工作流 /
图像 /
MiniMax H3-Image: single-image edit at 4K (ComfyUI)

MiniMax H3-Image: single-image edit at 4K (ComfyUI)
MiniMax H3-Image: single-image edit at 4K (ComfyUI)
MiniMax H3-Image single-image editing workflow: one reference image plus an instruction, re-rendered at 4K in six steps while keeping the rest intact. 48 nodes. Weights not bundled (minimax_h3_image_pruned_int8_convrot, …
图像
48 个节点 26 次使用 AIKSK 124 KB
需要的权重
这个工作流引用了 1 个权重文件,其中 0 个能在本站模型库里找到。
H3_Image______.json
模型库暂未收录
节点构成
LoadImage
× 16
MarkdownNote
× 6
PrimitiveStringMultiline
× 3
ComfySwitchNode
× 3
CLIPLoader
× 2
ResizeImageMaskNode
× 2
PrimitiveBoolean
× 2
SaveImage
× 1
UNETLoader
× 1
VAELoader
× 1
MiniMaxH3ImageConditioning
× 1
ConditioningZeroOut
× 1
TextGenerateV2
× 1
PreviewAny
× 1
KSampler
× 1
StringReplace
× 1
官方标注的模型
minimax_h3_image_pruned_int8_convrot.safetensorsminimax_h3_image_vae_fp16.safetensorsqwen3.5_9b_w4a8_gptq.safetensorsqwen3vl_32b_minimax_h3_int8_convrot.safetensors
工作流 JSON
和上面「下载工作流 JSON」是同一份文件。
1{
2 "id": "d59f6fff-6006-476b-864a-3fa8358dd2c1",
3 "revision": 0,
4 "last_node_id": 125,
5 "last_link_id": 146,
6 "nodes": [
7 {
8 "id": 9,
9 "type": "SaveImage",
10 "pos": [
11 2040,
12 -850
13 ],
14 "size": [
15 760,
16 730
17 ],
18 "flags": [],
19 "order": 46,
20 "mode": 0,
21 "inputs": [
22 {
23 "localized_name": "图片",
24 "name": "images",
25 "type": "IMAGE",
26 "link": 61
27 },
28 {
29 "localized_name": "文件名前缀",
30 "name": "filename_prefix",
31 "type": "STRING",
32 "widget": {
33 "name": "filename_prefix"
34 },
35 "link": null
36 }
37 ],
38 "outputs": [
39 {
40 "localized_name": "images",
41 "name": "images",
42 "type": "IMAGE",
43 "links": [
44 94
45 ]
46 }
47 ],
48 "title": "保存结果|AIKSK_H3_Image/Edit",
49 "properties": {
50 "Node name for S&R": "SaveImage"
51 },
52 "widgets_values": [
53 "AIKSK_H3_Image/Edit"
54 ],
55 "widgets_values_named": {
56 "filename_prefix": "AIKSK_H3_Image/Edit"
57 }
58 },
59 {
60 "id": 27,
61 "type": "MarkdownNote",
62 "pos": [
63 2500,
64 1580
65 ],
66 "size": [
67 440,
68 930
69 ],
70 "flags": {
71 "collapsed": false
72 },
73 "order": 0,
74 "mode": 0,
75 "inputs": [],
76 "outputs": [],
77 "title": "尺寸与多图使用说明",
78 "properties": [],
79 "widgets_values": [
80 "## 16个原生加载图像:默认全部Bypass\n\n16个框均为ComfyUI原生 LoadImage,全部预设为 Bypass(旁路)。\n\n使用方法:从图片01开始连续使用。上传图片后,选中需要使用的节点,按 Ctrl+B 恢复启用;不用的框继续保持 Bypass。也可右键切换节点模式。\n\n运行前至少启用图片01,因为原有缩放和跟随原图尺寸线路需要图片01作为输入。使用3张时启用01、02、03,其余保持Bypass,提示词对应 <Picture 1>、<Picture 2>、<Picture 3>。\n\n不要将不用的框恢复启用后留空;原生加载图像会检查所选文件是否存在。\n\n## 尺寸处理\n\n保留图片01的原有4MP缩放和32倍数对齐。图片02~16不增加尺寸缩放。\n\n关闭“启用外置尺寸”:沿用图片01处理后的宽高。开启:按外置选择器的比例和4/8/16MP输出。\n\n无自定义加载图像依赖。已检查节点类型、Bypass状态及连线;真实ComfyUI导入和GPU出图未测试。"
81 ],
82 "widgets_values_named": {
83 "text": "## 16个原生加载图像:默认全部Bypass\n\n16个框均为ComfyUI原生 LoadImage,全部预设为 Bypass(旁路)。\n\n使用方法:从图片01开始连续使用。上传图片后,选中需要使用的节点,按 Ctrl+B 恢复启用;不用的框继续保持 Bypass。也可右键切换节点模式。\n\n运行前至少启用图片01,因为原有缩放和跟随原图尺寸线路需要图片01作为输入。使用3张时启用01、02、03,其余保持Bypass,提示词对应 <Picture 1>、<Picture 2>、<Picture 3>。\n\n不要将不用的框恢复启用后留空;原生加载图像会检查所选文件是否存在。\n\n## 尺寸处理\n\n保留图片01的原有4MP缩放和32倍数对齐。图片02~16不增加尺寸缩放。\n\n关闭“启用外置尺寸”:沿用图片01处理后的宽高。开启:按外置选择器的比例和4/8/16MP输出。\n\n无自定义加载图像依赖。已检查节点类型、Bypass状态及连线;真实ComfyUI导入和GPU出图未测试。"
84 },
85 "color": "#222",
86 "bgcolor": "#000"
87 },
88 {
89 "id": 28,
90 "type": "MarkdownNote",
91 "pos": [
92 -780,
93 -850
94 ],
95 "size": [
96 640,
97 1200
98 ],
99 "flags": [],
100 "order": 1,
101 "mode": 0,
102 "inputs": [],
103 "outputs": [],
104 "title": "模型下载与目录|官方说明",
105 "properties": [],
106 "widgets_values": [
107 "## Input Asset\n\n- [vintage_yellow_taxi.png](https://raw.githubusercontent.com/Comfy-Org/workflow_templates/refs/heads/main/input/vintage_yellow_taxi.png)\n\n## Model Links\n\n- [Hugging Face:Comfy-Org/MiniMax-H3-Image](https://huggingface.co/Comfy-Org/MiniMax-H3-Image)\n- [Hugging Face:Comfy-Org/MiniMax-H3](https://huggingface.co/Comfy-Org/MiniMax-H3)\n- [Hugging Face:Comfy-Org/Qwen3.5](https://huggingface.co/Comfy-Org/Qwen3.5)\n\n**diffusion_models**\n\n- [minimax_h3_image_pruned_int8_convrot.safetensors](https://huggingface.co/Comfy-Org/MiniMax-H3-Image/resolve/main/diffusion_models/minimax_h3_image_pruned_int8_convrot.safetensors) (21 GB)\n\n**text_encoders**\n\n- [qwen3vl_32b_minimax_h3_int8_convrot.safetensors](https://huggingface.co/Comfy-Org/MiniMax-H3/resolve/main/text_encoders/qwen3vl_32b_minimax_h3_int8_convrot.safetensors) (25.28 GB)\n- [qwen3.5_9b_w4a8_gptq.safetensors](https://huggingface.co/Comfy-Org/Qwen3.5/resolve/main/text_encoders/qwen3.5_9b_w4a8_gptq.safetensors) (6.84 GB)\n\n**vae**\n\n- [minimax_h3_image_vae_fp16.safetensors](https://huggingface.co/Comfy-Org/MiniMax-H3-Image/resolve/main/vae/minimax_h3_image_vae_fp16.safetensors)\n\n\n## Model Storage Location\n\n```\n📂 ComfyUI/\n├── 📂 models/\n│ ├── 📂 diffusion_models/\n│ │ └── minimax_h3_image_pruned_int8_convrot.safetensors\n│ ├── 📂 text_encoders/\n│ │ ├── qwen3vl_32b_minimax_h3_int8_convrot.safetensors\n│ │ └── qwen3.5_9b_w4a8_gptq.safetensors\n│ └── 📂 vae/\n│ └── minimax_h3_image_vae_fp16.safetensors\n```\n\n## Report Issue\n\nNote: Please update ComfyUI first ([guide](https://docs.comfy.org/installation/update_comfyui)) and prepare required models. Desktop/Cloud updates follow stable releases, so some nightly-supported models may not be available yet.\n\n- Cannot run / runtime errors: [ComfyUI/issues](https://github.com/comfyanonymous/ComfyUI/issues)\n- UI / frontend issues: [ComfyUI_frontend/issues](https://github.com/Comfy-Org/ComfyUI_frontend/issues)\n- Workflow issues: [workflow_templates/issues](https://github.com/Comfy-Org/workflow_templates/issues)\n"
108 ],
109 "widgets_values_named": {
110 "text": "## Input Asset\n\n- [vintage_yellow_taxi.png](https://raw.githubusercontent.com/Comfy-Org/workflow_templates/refs/heads/main/input/vintage_yellow_taxi.png)\n\n## Model Links\n\n- [Hugging Face:Comfy-Org/MiniMax-H3-Image](https://huggingface.co/Comfy-Org/MiniMax-H3-Image)\n- [Hugging Face:Comfy-Org/MiniMax-H3](https://huggingface.co/Comfy-Org/MiniMax-H3)\n- [Hugging Face:Comfy-Org/Qwen3.5](https://huggingface.co/Comfy-Org/Qwen3.5)\n\n**diffusion_models**\n\n- [minimax_h3_image_pruned_int8_convrot.safetensors](https://huggingface.co/Comfy-Org/MiniMax-H3-Image/resolve/main/diffusion_models/minimax_h3_image_pruned_int8_convrot.safetensors) (21 GB)\n\n**text_encoders**\n\n- [qwen3vl_32b_minimax_h3_int8_convrot.safetensors](https://huggingface.co/Comfy-Org/MiniMax-H3/resolve/main/text_encoders/qwen3vl_32b_minimax_h3_int8_convrot.safetensors) (25.28 GB)\n- [qwen3.5_9b_w4a8_gptq.safetensors](https://huggingface.co/Comfy-Org/Qwen3.5/resolve/main/text_encoders/qwen3.5_9b_w4a8_gptq.safetensors) (6.84 GB)\n\n**vae**\n\n- [minimax_h3_image_vae_fp16.safetensors](https://huggingface.co/Comfy-Org/MiniMax-H3-Image/resolve/main/vae/minimax_h3_image_vae_fp16.safetensors)\n\n\n## Model Storage Location\n\n```\n📂 ComfyUI/\n├── 📂 models/\n│ ├── 📂 diffusion_models/\n│ │ └── minimax_h3_image_pruned_int8_convrot.safetensors\n│ ├── 📂 text_encoders/\n│ │ ├── qwen3vl_32b_minimax_h3_int8_convrot.safetensors\n│ │ └── qwen3.5_9b_w4a8_gptq.safetensors\n│ └── 📂 vae/\n│ └── minimax_h3_image_vae_fp16.safetensors\n```\n\n## Report Issue\n\nNote: Please update ComfyUI first ([guide](https://docs.comfy.org/installation/update_comfyui)) and prepare required models. Desktop/Cloud updates follow stable releases, so some nightly-supported models may not be available yet.\n\n- Cannot run / runtime errors: [ComfyUI/issues](https://github.com/comfyanonymous/ComfyUI/issues)\n- UI / frontend issues: [ComfyUI_frontend/issues](https://github.com/Comfy-Org/ComfyUI_frontend/issues)\n- Workflow issues: [workflow_templates/issues](https://github.com/Comfy-Org/workflow_templates/issues)\n"
111 },
112 "color": "#222",
113 "bgcolor": "#000"
114 },
115 {
116 "id": 29,
117 "type": "UNETLoader",
118 "pos": [
119 0,
120 -850
121 ],
122 "size": [
123 590,
124 110
125 ],
126 "flags": [],
127 "order": 2,
128 "mode": 0,
129 "inputs": [
130 {
131 "localized_name": "UNet名称",
132 "name": "unet_name",
133 "type": "COMBO",
134 "widget": {
135 "name": "unet_name"
136 },
137 "link": null
138 },
139 {
140 "localized_name": "数据类型",
141 "name": "weight_dtype",
142 "type": "COMBO",
143 "widget": {
144 "name": "weight_dtype"
145 },
146 "link": null
147 }
148 ],
149 "outputs": [
150 {
151 "localized_name": "模型",
152 "name": "MODEL",
153 "type": "MODEL",
154 "links": [
155 54
156 ]
157 }
158 ],
159 "title": "主模型|默认INT8,可选已安装BF16",
160 "properties": {
161 "Node name for S&R": "UNETLoader",
162 "models": [
163 {
164 "name": "minimax_h3_image_pruned_int8_convrot.safetensors",
165 "url": "https://huggingface.co/Comfy-Org/MiniMax-H3-Image/resolve/main/diffusion_models/minimax_h3_image_pruned_int8_convrot.safetensors",
166 "directory": "diffusion_models"
167 }
168 ]
169 },
170 "widgets_values": [
171 "minimax_h3_image_pruned_int8_convrot.safetensors",
172 "default"
173 ],
174 "widgets_values_named": {
175 "unet_name": "minimax_h3_image_pruned_int8_convrot.safetensors",
176 "weight_dtype": "default"
177 }
178 },
179 {
180 "id": 30,
181 "type": "CLIPLoader",
182 "pos": [
183 0,
184 -610
185 ],
186 "size": [
187 590,
188 140
189 ],
190 "flags": [],
191 "order": 3,
192 "mode": 0,
193 "inputs": [
194 {
195 "localized_name": "CLIP名称",
196 "name": "clip_name",
197 "type": "COMBO",
198 "widget": {
199 "name": "clip_name"
200 },
201 "link": null
202 },
203 {
204 "localized_name": "类型",
205 "name": "type",
206 "type": "COMBO",
207 "widget": {
208 "name": "type"
209 },
210 "link": null
211 },
212 {
213 "localized_name": "设备",
214 "name": "device",
215 "shape": 7,
216 "type": "COMBO",
217 "widget": {
218 "name": "device"
219 },
220 "link": null
221 }
222 ],
223 "outputs": [
224 {
225 "localized_name": "CLIP",
226 "name": "CLIP",
227 "type": "CLIP",
228 "links": [
229 49
230 ]
231 }
232 ],
233 "title": "原生文本编码器|Qwen3-VL",
234 "properties": {
235 "Node name for S&R": "CLIPLoader",
236 "models": [
237 {
238 "name": "qwen3vl_32b_minimax_h3_int8_convrot.safetensors",
239 "url": "https://huggingface.co/Comfy-Org/MiniMax-H3/resolve/main/text_encoders/qwen3vl_32b_minimax_h3_int8_convrot.safetensors",
240 "directory": "text_encoders"
241 }
242 ]
243 },
244 "widgets_values": [
245 "qwen3vl_32b_minimax_h3_int8_convrot.safetensors",
246 "minimax",
247 "default"
248 ],
249 "widgets_values_named": {
250 "clip_name": "qwen3vl_32b_minimax_h3_int8_convrot.safetensors",
251 "type": "minimax",
252 "device": "default"
253 }
254 },
255 {
256 "id": 31,
257 "type": "VAELoader",
258 "pos": [
259 0,
260 -350
261 ],
262 "size": [
263 590,
264 100
265 ],
266 "flags": [],
267 "order": 4,
268 "mode": 0,
269 "inputs": [
270 {
271 "localized_name": "vae名称",
272 "name": "vae_name",
273 "type": "COMBO",
274 "widget": {
275 "name": "vae_name"
276 },
277 "link": null
278 }
279 ],
280 "outputs": [
281 {
282 "localized_name": "VAE",
283 "name": "VAE",
284 "type": "VAE",
285 "links": [
286 50,
287 60
288 ]
289 }
290 ],
291 "title": "原生图像VAE",
292 "properties": {
293 "Node name for S&R": "VAELoader",
294 "models": [
295 {
296 "name": "minimax_h3_image_vae_fp16.safetensors",
297 "url": "https://huggingface.co/Comfy-Org/MiniMax-H3-Image/resolve/main/vae/minimax_h3_image_vae_fp16.safetensors",
298 "directory": "vae"
299 }
300 ]
301 },
302 "widgets_values": [
303 "minimax_h3_image_vae_fp16.safetensors"
304 ],
305 "widgets_values_named": {
306 "vae_name": "minimax_h3_image_vae_fp16.safetensors"
307 }
308 },
309 {
310 "id": 32,
311 "type": "MiniMaxH3ImageConditioning",
312 "pos": [
313 760,
314 -850
315 ],
316 "size": [
317 560,
318 760
319 ],
320 "flags": [],
321 "order": 42,
322 "mode": 0,
323 "inputs": [
324 {
325 "localized_name": "clip",
326 "name": "clip",
327 "type": "CLIP",
328 "link": 49
329 },
330 {
331 "localized_name": "vae",
332 "name": "vae",
333 "shape": 7,
334 "type": "VAE",
335 "link": 50
336 },
337 {
338 "localized_name": "image_0",
339 "name": "images.image_0",
340 "shape": 7,
341 "type": "IMAGE",
342 "link": null
343 },
344 {
345 "localized_name": "prompt",
346 "name": "prompt",
347 "type": "STRING",
348 "widget": {
349 "name": "prompt"
350 },
351 "link": 89
352 },
353 {
354 "localized_name": "width",
355 "name": "width",
356 "type": "INT",
357 "widget": {
358 "name": "width"
359 },
360 "link": 143
361 },
362 {
363 "localized_name": "height",
364 "name": "height",
365 "type": "INT",
366 "widget": {
367 "name": "height"
368 },
369 "link": 146
370 },
371 {
372 "name": "images.image_1",
373 "shape": 7,
374 "type": "IMAGE",
375 "link": 80
376 },
377 {
378 "localized_name": "image_5",
379 "name": "images.image_5",
380 "shape": 7,
381 "type": "IMAGE",
382 "link": 105
383 },
384 {
385 "localized_name": "image_6",
386 "name": "images.image_6",
387 "shape": 7,
388 "type": "IMAGE",
389 "link": 108
390 },
391 {
392 "localized_name": "image_7",
393 "name": "images.image_7",
394 "shape": 7,
395 "type": "IMAGE",
396 "link": 111
397 },
398 {
399 "localized_name": "image_8",
400 "name": "images.image_8",
401 "shape": 7,
402 "type": "IMAGE",
403 "link": 114
404 },
405 {
406 "localized_name": "image_9",
407 "name": "images.image_9",
408 "shape": 7,
409 "type": "IMAGE",
410 "link": 117
411 },
412 {
413 "localized_name": "image_10",
414 "name": "images.image_10",
415 "shape": 7,
416 "type": "IMAGE",
417 "link": 120
418 },
419 {
420 "localized_name": "image_11",
421 "name": "images.image_11",
422 "shape": 7,
423 "type": "IMAGE",
424 "link": 123
425 },
426 {
427 "localized_name": "image_12",
428 "name": "images.image_12",
429 "shape": 7,
430 "type": "IMAGE",
431 "link": 126
432 },
433 {
434 "localized_name": "image_13",
435 "name": "images.image_13",
436 "shape": 7,
437 "type": "IMAGE",
438 "link": 129
439 },
440 {
441 "localized_name": "image_14",
442 "name": "images.image_14",
443 "shape": 7,
444 "type": "IMAGE",
445 "link": 132
446 },
447 {
448 "localized_name": "image_15",
449 "name": "images.image_15",
450 "shape": 7,
451 "type": "IMAGE",
452 "link": 135
453 },
454 {
455 "localized_name": "image_16",
456 "name": "images.image_16",
457 "shape": 7,
458 "type": "IMAGE",
459 "link": 138
460 },
461 {
462 "localized_name": "image_17",
463 "name": "images.image_17",
464 "shape": 7,
465 "type": "IMAGE",
466 "link": null
467 },
468 {
469 "localized_name": "image_18",
470 "name": "images.image_18",
471 "shape": 7,
472 "type": "IMAGE",
473 "link": null
474 },
475 {
476 "localized_name": "image_2",
477 "name": "images.image_2",
478 "shape": 7,
479 "type": "IMAGE",
480 "link": 96
481 },
482 {
483 "localized_name": "image_3",
484 "name": "images.image_3",
485 "shape": 7,
486 "type": "IMAGE",
487 "link": 99
488 },
489 {
490 "localized_name": "image_4",
491 "name": "images.image_4",
492 "shape": 7,
493 "type": "IMAGE",
494 "link": 102
495 }
496 ],
497 "outputs": [
498 {
499 "localized_name": "positive",
500 "name": "positive",
501 "type": "CONDITIONING",
502 "links": [
503 52,
504 55
505 ]
506 },
507 {
508 "localized_name": "Latent",
509 "name": "LATENT",
510 "type": "LATENT",
511 "links": [
512 57
513 ]
514 }
515 ],
516 "title": "原生图像编辑条件|16张参考图已接好",
517 "properties": {
518 "Node name for S&R": "MiniMaxH3ImageConditioning"
519 },
520 "widgets_values": [
521 "",
522 3104,
523 3104
524 ],
525 "widgets_values_named": {
526 "prompt": "",
527 "width": 3104,
528 "height": 3104
529 }
530 },
531 {
532 "id": 33,
533 "type": "ConditioningZeroOut",
534 "pos": [
535 760,
536 -10
537 ],
538 "size": [
539 420,
540 70
541 ],
542 "flags": [],
543 "order": 43,
544 "mode": 0,
545 "inputs": [
546 {
547 "localized_name": "条件",
548 "name": "conditioning",
549 "type": "CONDITIONING",
550 "link": 52
551 }
552 ],
553 "outputs": [
554 {
555 "localized_name": "条件",
556 "name": "CONDITIONING",
557 "type": "CONDITIONING",
558 "links": [
559 56
560 ]
561 }
562 ],
563 "title": "原生条件零化",
564 "properties": {
565 "Node name for S&R": "ConditioningZeroOut"
566 }
567 },
568 {
569 "id": 34,
570 "type": "TextGenerateV2",
571 "pos": [
572 760,
573 3320
574 ],
575 "size": [
576 560,
577 870
578 ],
579 "flags": [],
580 "order": 36,
581 "mode": 0,
582 "showAdvanced": true,
583 "inputs": [
584 {
585 "localized_name": "clip",
586 "name": "clip",
587 "type": "CLIP",
588 "link": 67
589 },
590 {
591 "localized_name": "image1",
592 "name": "images.image1",
593 "shape": 7,
594 "type": "IMAGE",
595 "link": 81
596 },
597 {
598 "localized_name": "video1",
599 "name": "videos.video1",
600 "shape": 7,
601 "type": "IMAGE",
602 "link": null
603 },
604 {
605 "localized_name": "audio1",
606 "name": "audios.audio1",
607 "shape": 7,
608 "type": "AUDIO",
609 "link": null
610 },
611 {
612 "localized_name": "system_prompt",
613 "name": "system_prompt",
614 "shape": 7,
615 "type": "STRING",
616 "link": 68
617 },
618 {
619 "localized_name": "prompt",
620 "name": "prompt",
621 "type": "STRING",
622 "widget": {
623 "name": "prompt"
624 },
625 "link": 69
626 },
627 {
628 "localized_name": "max_length",
629 "name": "max_length",
630 "type": "INT",
631 "widget": {
632 "name": "max_length"
633 },
634 "link": null
635 },
636 {
637 "localized_name": "Sampling Mode",
638 "name": "sampling_mode",
639 "type": "COMFY_DYNAMICCOMBO_V3",
640 "widget": {
641 "name": "sampling_mode"
642 },
643 "link": null
644 },
645 {
646 "localized_name": "temperature",
647 "name": "sampling_mode.temperature",
648 "type": "FLOAT",
649 "widget": {
650 "name": "sampling_mode.temperature"
651 },
652 "link": null
653 },
654 {
655 "localized_name": "top_k",
656 "name": "sampling_mode.top_k",
657 "type": "INT",
658 "widget": {
659 "name": "sampling_mode.top_k"
660 },
661 "link": null
662 },
663 {
664 "localized_name": "top_p",
665 "name": "sampling_mode.top_p",
666 "type": "FLOAT",
667 "widget": {
668 "name": "sampling_mode.top_p"
669 },
670 "link": null
671 },
672 {
673 "localized_name": "min_p",
674 "name": "sampling_mode.min_p",
675 "type": "FLOAT",
676 "widget": {
677 "name": "sampling_mode.min_p"
678 },
679 "link": null
680 },
681 {
682 "localized_name": "repetition_penalty",
683 "name": "sampling_mode.repetition_penalty",
684 "type": "FLOAT",
685 "widget": {
686 "name": "sampling_mode.repetition_penalty"
687 },
688 "link": null
689 },
690 {
691 "localized_name": "presence_penalty",
692 "name": "sampling_mode.presence_penalty",
693 "type": "FLOAT",
694 "widget": {
695 "name": "sampling_mode.presence_penalty"
696 },
697 "link": null
698 },
699 {
700 "localized_name": "seed",
701 "name": "sampling_mode.seed",
702 "type": "INT",
703 "widget": {
704 "name": "sampling_mode.seed"
705 },
706 "link": null
707 },
708 {
709 "localized_name": "thinking",
710 "name": "thinking",
711 "shape": 7,
712 "type": "BOOLEAN",
713 "widget": {
714 "name": "thinking"
715 },
716 "link": null
717 },
718 {
719 "localized_name": "use_default_template",
720 "name": "use_default_template",
721 "shape": 7,
722 "type": "BOOLEAN",
723 "widget": {
724 "name": "use_default_template"
725 },
726 "link": null
727 },
728 {
729 "localized_name": "mtp",
730 "name": "mtp",
731 "shape": 7,
732 "type": "COMBO",
733 "widget": {
734 "name": "mtp"
735 },
736 "link": null
737 },
738 {
739 "localized_name": "image5",
740 "name": "images.image5",
741 "shape": 7,
742 "type": "IMAGE",
743 "link": 106
744 },
745 {
746 "localized_name": "image6",
747 "name": "images.image6",
748 "shape": 7,
749 "type": "IMAGE",
750 "link": 109
751 },
752 {
753 "localized_name": "image7",
754 "name": "images.image7",
755 "shape": 7,
756 "type": "IMAGE",
757 "link": 112
758 },
759 {
760 "localized_name": "image8",
761 "name": "images.image8",
762 "shape": 7,
763 "type": "IMAGE",
764 "link": 115
765 },
766 {
767 "localized_name": "image9",
768 "name": "images.image9",
769 "shape": 7,
770 "type": "IMAGE",
771 "link": 118
772 },
773 {
774 "localized_name": "image10",
775 "name": "images.image10",
776 "shape": 7,
777 "type": "IMAGE",
778 "link": 121
779 },
780 {
781 "localized_name": "image11",
782 "name": "images.image11",
783 "shape": 7,
784 "type": "IMAGE",
785 "link": 124
786 },
787 {
788 "localized_name": "image12",
789 "name": "images.image12",
790 "shape": 7,
791 "type": "IMAGE",
792 "link": 127
793 },
794 {
795 "localized_name": "image13",
796 "name": "images.image13",
797 "shape": 7,
798 "type": "IMAGE",
799 "link": 130
800 },
801 {
802 "localized_name": "image14",
803 "name": "images.image14",
804 "shape": 7,
805 "type": "IMAGE",
806 "link": 133
807 },
808 {
809 "localized_name": "image15",
810 "name": "images.image15",
811 "shape": 7,
812 "type": "IMAGE",
813 "link": 136
814 },
815 {
816 "localized_name": "image16",
817 "name": "images.image16",
818 "shape": 7,
819 "type": "IMAGE",
820 "link": 139
821 },
822 {
823 "localized_name": "image17",
824 "name": "images.image17",
825 "shape": 7,
826 "type": "IMAGE",
827 "link": null
828 },
829 {
830 "localized_name": "image18",
831 "name": "images.image18",
832 "shape": 7,
833 "type": "IMAGE",
834 "link": null
835 },
836 {
837 "localized_name": "image2",
838 "name": "images.image2",
839 "shape": 7,
840 "type": "IMAGE",
841 "link": 97
842 },
843 {
844 "localized_name": "image3",
845 "name": "images.image3",
846 "shape": 7,
847 "type": "IMAGE",
848 "link": 100
849 },
850 {
851 "localized_name": "image4",
852 "name": "images.image4",
853 "shape": 7,
854 "type": "IMAGE",
855 "link": 103
856 },
857 {
858 "name": "audios.audio2",
859 "shape": 7,
860 "type": "AUDIO",
861 "link": null
862 },
863 {
864 "name": "audios.audio3",
865 "shape": 7,
866 "type": "AUDIO",
867 "link": null
868 }
869 ],
870 "outputs": [
871 {
872 "localized_name": "generated_text",
873 "name": "generated_text",
874 "type": "STRING",
875 "links": [
876 65
877 ]
878 },
879 {
880 "localized_name": "thinking",
881 "name": "thinking",
882 "type": "STRING",
883 "links": []
884 }
885 ],
886 "title": "原生扩写|与编辑条件按相同顺序接入16图",
887 "properties": {
888 "Node name for S&R": "TextGenerateV2"
889 },
890 "widgets_values": [
891 "",
892 8196,
893 "on",
894 1,
895 20,
896 0.95,
897 0,
898 1,
899 0,
900 0,
901 true,
902 true,
903 "auto"
904 ],
905 "widgets_values_named": {
906 "prompt": "",
907 "max_length": 8196,
908 "sampling_mode": "on",
909 "sampling_mode.temperature": 1,
910 "sampling_mode.top_k": 20,
911 "sampling_mode.top_p": 0.95,
912 "sampling_mode.min_p": 0,
913 "sampling_mode.repetition_penalty": 1,
914 "sampling_mode.presence_penalty": 0,
915 "sampling_mode.seed": 0,
916 "thinking": true,
917 "use_default_template": true,
918 "mtp": "auto"
919 }
920 },
921 {
922 "id": 35,
923 "type": "CLIPLoader",
924 "pos": [
925 0,
926 3320
927 ],
928 "size": [
929 560,
930 140
931 ],
932 "flags": [],
933 "order": 5,
934 "mode": 0,
935 "showAdvanced": false,
936 "inputs": [
937 {
938 "localized_name": "CLIP名称",
939 "name": "clip_name",
940 "type": "COMBO",
941 "widget": {
942 "name": "clip_name"
943 },
944 "link": null
945 },
946 {
947 "localized_name": "类型",
948 "name": "type",
949 "type": "COMBO",
950 "widget": {
951 "name": "type"
952 },
953 "link": null
954 },
955 {
956 "localized_name": "设备",
957 "name": "device",
958 "shape": 7,
959 "type": "COMBO",
960 "widget": {
961 "name": "device"
962 },
963 "link": null
964 }
965 ],
966 "outputs": [
967 {
968 "localized_name": "CLIP",
969 "name": "CLIP",
970 "type": "CLIP",
971 "links": [
972 67
973 ]
974 }
975 ],
976 "title": "可选扩写模型|Qwen3.5",
977 "properties": {
978 "Node name for S&R": "CLIPLoader",
979 "models": [
980 {
981 "name": "qwen3.5_9b_w4a8_gptq.safetensors",
982 "url": "https://huggingface.co/Comfy-Org/Qwen3.5/resolve/main/text_encoders/qwen3.5_9b_w4a8_gptq.safetensors",
983 "directory": "text_encoders"
984 }
985 ]
986 },
987 "widgets_values": [
988 "qwen3.5_9b_w4a8_gptq.safetensors",
989 "minimax",
990 "default"
991 ],
992 "widgets_values_named": {
993 "clip_name": "qwen3.5_9b_w4a8_gptq.safetensors",
994 "type": "minimax",
995 "device": "default"
996 },
997 "color": "#322",
998 "bgcolor": "#533"
999 },
1000 {
1001 "id": 36,
1002 "type": "PrimitiveStringMultiline",
1003 "pos": [
1004 0,
1005 3600
1006 ],
1007 "size": [
1008 560,
1009 220
1010 ],
1011 "flags": [],
1012 "order": 6,
1013 "mode": 0,
1014 "inputs": [
1015 {
1016 "localized_name": "值",
1017 "name": "value",
1018 "type": "STRING",
1019 "widget": {
1020 "name": "value"
1021 },
1022 "link": null
1023 }
1024 ],
1025 "outputs": [
1026 {
1027 "localized_name": "字符串",
1028 "name": "STRING",
1029 "type": "STRING",
1030 "links": [
1031 68
1032 ]
1033 }
1034 ],
1035 "title": "官方扩写系统提示词",
1036 "properties": {
1037 "Node name for S&R": "PrimitiveStringMultiline"
1038 },
1039 "widgets_values": [
1040 "Reason in the thinking section first. The visible final content must be exactly one JSON object with exactly two keys: expanded_caption and aspect_ratio. Every ASCII double quotation mark used as caption content inside expanded_caption must be serialized as the JSON escape sequence \\\". Do not output text outside that object."
1041 ],
1042 "widgets_values_named": {
1043 "value": "Reason in the thinking section first. The visible final content must be exactly one JSON object with exactly two keys: expanded_caption and aspect_ratio. Every ASCII double quotation mark used as caption content inside expanded_caption must be serialized as the JSON escape sequence \\\". Do not output text outside that object."
1044 }
1045 },
1046 {
1047 "id": 37,
1048 "type": "PreviewAny",
1049 "pos": [
1050 2120,
1051 3320
1052 ],
1053 "size": [
1054 720,
1055 860
1056 ],
1057 "flags": [],
1058 "order": 41,
1059 "mode": 0,
1060 "inputs": [
1061 {
1062 "localized_name": "源",
1063 "name": "source",
1064 "type": "*",
1065 "link": 53
1066 }
1067 ],
1068 "outputs": [
1069 {
1070 "localized_name": "字符串",
1071 "name": "STRING",
1072 "type": "STRING",
1073 "links": [
1074 89
1075 ]
1076 }
1077 ],
1078 "title": "实际送入模型的提示词",
1079 "properties": {
1080 "Node name for S&R": "PreviewAny"
1081 },
1082 "widgets_values": [],
1083 "widgets_values_named": []
1084 },
1085 {
1086 "id": 38,
1087 "type": "KSampler",
1088 "pos": [
1089 1480,
1090 -850
1091 ],
1092 "size": [
1093 400,
1094 560
1095 ],
1096 "flags": [],
1097 "order": 44,
1098 "mode": 0,
1099 "inputs": [
1100 {
1101 "localized_name": "模型",
1102 "name": "model",
1103 "type": "MODEL",
1104 "link": 54
1105 },
1106 {
1107 "localized_name": "正面条件",
1108 "name": "positive",
1109 "type": "CONDITIONING",
1110 "link": 55
1111 },
1112 {
1113 "localized_name": "负面条件",
1114 "name": "negative",
1115 "type": "CONDITIONING",
1116 "link": 56
1117 },
1118 {
1119 "localized_name": "Latent图像",
1120 "name": "latent_image",
1121 "type": "LATENT",
1122 "link": 57
1123 },
1124 {
1125 "localized_name": "种子",
1126 "name": "seed",
1127 "type": "INT",
1128 "widget": {
1129 "name": "seed"
1130 },
1131 "link": null
1132 },
1133 {
1134 "localized_name": "步数",
1135 "name": "steps",
1136 "type": "INT",
1137 "widget": {
1138 "name": "steps"
1139 },
1140 "link": null
1141 },
1142 {
1143 "localized_name": "cfg",
1144 "name": "cfg",
1145 "type": "FLOAT",
1146 "widget": {
1147 "name": "cfg"
1148 },
1149 "link": null
1150 },
1151 {
1152 "localized_name": "采样器名称",
1153 "name": "sampler_name",
1154 "type": "COMBO",
1155 "widget": {
1156 "name": "sampler_name"
1157 },
1158 "link": null
1159 },
1160 {
1161 "localized_name": "调度器",
1162 "name": "scheduler",
1163 "type": "COMBO",
1164 "widget": {
1165 "name": "scheduler"
1166 },
1167 "link": null
1168 },
1169 {
1170 "localized_name": "降噪",
1171 "name": "denoise",
1172 "type": "FLOAT",
1173 "widget": {
1174 "name": "denoise"
1175 },
1176 "link": null
1177 }
1178 ],
1179 "outputs": [
1180 {
1181 "localized_name": "Latent",
1182 "name": "LATENT",
1183 "type": "LATENT",
1184 "links": [
1185 59
1186 ]
1187 }
1188 ],
1189 "title": "原生采样|6步 / Euler / CFG 1",
1190 "properties": {
1191 "Node name for S&R": "KSampler"
1192 },
1193 "widgets_values": [
1194 9527,
1195 "randomize",
1196 6,
1197 1,
1198 "euler",
1199 "simple",
1200 1
1201 ],
1202 "widgets_values_named": {
1203 "seed": 9527,
1204 "control_after_generate": "randomize",
1205 "steps": 6,
1206 "cfg": 1,
1207 "sampler_name": "euler",
1208 "scheduler": "simple",
1209 "denoise": 1
1210 }
1211 },
1212 {
1213 "id": 39,
1214 "type": "PrimitiveStringMultiline",
1215 "pos": [
1216 0,
1217 3940
1218 ],
1219 "size": [
1220 560,
1221 260
1222 ],
1223 "flags": [],
1224 "order": 7,
1225 "mode": 0,
1226 "inputs": [
1227 {
1228 "localized_name": "值",
1229 "name": "value",
1230 "type": "STRING",
1231 "widget": {
1232 "name": "value"
1233 },
1234 "link": null
1235 }
1236 ],
1237 "outputs": [
1238 {
1239 "localized_name": "字符串",
1240 "name": "STRING",
1241 "type": "STRING",
1242 "links": [
1243 58
1244 ]
1245 }
1246 ],
1247 "title": "官方编辑扩写任务提示词",
1248 "properties": {
1249 "Node name for S&R": "PrimitiveStringMultiline"
1250 },
1251 "widgets_values": [
1252 "You are an expert at enhancing a short image-editing instruction into a rich, vivid, reference-conditioned description of the intended finished target image.\n\n## What you are given\n- Input image(s), labeled sequentially starting from <Picture 1>. These are the user's source and reference images.\n- One short editing instruction. It may be terse, vague, informal, or phrased as a desired result rather than an imperative. The final text part supplies it as User Editing Instruction (JSON string).\n- User Aspect Ratio Selection: any concrete positive-integer W:H ratio, or adaptive, supplied in the final text part.\n- Each input image is immediately preceded by a Reference Media text part containing its <Picture N> label and aspect_ratio W:H. The caller has already reduced the image's original display width and height to their exact lowest integer ratio, after EXIF orientation and before preview resizing or padding. No pixel dimensions are supplied in this metadata. The ratio applies only to the immediately following image. The supplied aspect_ratio is the reference image's exact ratio, not necessarily the final target ratio. Use it as given; do not estimate the source ratio from the preview. Determine the final target ratio using the rules below. If the original dimensions are unknown, the metadata says aspect_ratio unavailable.\n- There is no target image. Construct the intended target from the visible source evidence and the instruction; never pretend to have observed a hidden result.\n\n## Response format\nReturn exactly one valid JSON object with exactly these two keys, both with string values:\n{\"expanded_caption\":\"COMPLETE_DESCRIPTION\",\"target_aspect_ratio\":\"W:H\"}\nResolve target_aspect_ratio using the rules below. Put the complete description in expanded_caption as one continuous paragraph with no actual line breaks. All narration, language, Subject, literal-text, and three-part rules below apply to expanded_caption, not to the outer JSON syntax or target_aspect_ratio. Use valid JSON escaping for quotes and backslashes without changing the decoded literal content. Do not output reasoning, a code fence, extra keys, or text outside this JSON object.\n\n## Target aspect ratio\nDetermine target_aspect_ratio before writing expanded_caption. Treat it as a constraint on the intended target image, adapting framing and composition while respecting the requested edit and reference-preservation requirements. Describe any necessary changes under the changed/new-content rules. User Aspect Ratio Selection is part of the user's request. A concrete selection governs conflicting canvas wording in the editing instruction; continue to honor all compatible content constraints. Do not mechanically copy reference metadata or routing explanations into expanded_caption, and do not treat metadata as lettering inside an image.\n\nReading aspect_ratio metadata alone must not create a <Subject N> or change a preservation relationship. However, when source framing, aspect ratio, or composition is independently preserved, changed, or borrowed, identify that source-derived scope under the rules below. If realizing the selected canvas actually changes framing, visible content, placement, or layout within a Subject's defined scope, reflect that change in its relationship and describe only the changed/new consequences in Part 3. Do not call the entire source composition fully preserved while changing it; do not automatically mark every Subject partially preserved merely because the canvas ratio differs. Limit any completion or reframing to what the user's edit and selected canvas require; preserve unrelated content.\n\nRead User Aspect Ratio Selection from the final input part.\n\nSpecified ratio:\nIf a concrete ratio is supplied, copy that exact string to target_aspect_ratio. Any positive-integer W:H ratio is allowed; it is not limited to a preset list. Neither the editing instruction nor a reference image may override it.\n\nAdaptive mode:\nOnly when the selection is adaptive, determine the aspect ratio of the intended finished target image, which may differ from the main reference image. Apply these rules in order:\n\n1. Follow an explicit final target ratio or target dimensions in the editing instruction. Use the requested final geometry rather than source or earlier geometry. Exact numeric geometry takes precedence over generic orientation. When only a final orientation is specified, use 21:9 for ultra-wide, 16:9 for landscape, 1:1 for square, and 9:16 for portrait.\n\n2. Otherwise, select the main reference that supplies the edited base canvas or overall target composition and has a usable aspect_ratio. Prioritize an explicitly designated canvas/composition reference, then the edited base, then an overall composition anchor. Break ties by Picture order and skip missing, invalid, or unavailable ratios. Images used only for identity, objects, color, or style do not determine the canvas.\n\n Use the main reference's aspect_ratio as the starting ratio. Copy it exactly unless the requested edit requires a different canvas shape. Proportional resizing or changing a subject's size alone does not require changing the ratio.\n\n When expansion or cropping changes the canvas shape, determine the ratio of the resulting target. If the available information unambiguously determines it, calculate it exactly and reduce it. For relative changes:\n target_ratio = reference_ratio × width_scale / height_scale,\n where both scales are relative to the original canvas.\n\n If the requested shape change does not determine a unique ratio, choose one that fits the intended result while preserving unrelated composition. Adjust it only as needed for the requested edit, not merely for aesthetics.\n\n3. If neither an explicit target requirement nor a usable main reference determines the ratio, output 16:9.\n\nApply these rules to the whole target canvas. Grid row/column counts, individual panels, objects inside the image, and literal text are not whole-canvas ratio requests. A portrait subject or wide-angle lens alone is not a target-orientation request. Do not use a reference board's ratio as the ratio of one panel.\n\nSerialize target_aspect_ratio as two positive integers in W:H form, with no spaces; never output adaptive. Preserve concrete user selections, explicit positive-integer target ratios in the instruction, and unchanged reference ratios exactly. Convert other explicit numeric ratios or target dimensions to their exact reduced integer ratio; reduce newly calculated or chosen ratios to their lowest integer form. Do not snap to preset ratios or clamp to a supported range.\n\nLater descriptive additions must not change the chosen ratio. Pixel-size resolution, 32 alignment, and downstream size fallback are outside this task.\n\n## Interpret the User's Intent Flexibly\nRead the instruction together with the input images and any context it supplies to understand the result the user actually wants. Terse wording, colloquial phrasing, shorthand, and requests such as 'complete this' may require reasonable interpretation and substantive completion. Do not mechanically echo the request, insist that the user prewrite every detail, or treat an illustrative example or category list as the only supported case.\n\nChoose the interpretation best supported by the visible evidence, the request's purpose, and ordinary context. Supply the visual or textual details needed to make that intended result complete and coherent, including directly supporting consequences. Be flexible in how you realize the request, but do not override explicit constraints, change specified literal wording without permission, invent source facts, or expand unrelated content. Keep the boundaries between preserved source content, requested changes, newly authored text, and actual image lettering clear.\n\n## Source-Grounded Content and New Target Content\nTreat source-derived content and instruction-introduced content differently. Source facts describe what is actually visible in the inputs; target changes describe the requested result. Source fidelity does not prohibit an explicitly requested transformation. Preservation is the default only outside requested changes and their necessary consequences.\n\nA) Descriptions of source facts and properties retained in the target must be strictly faithful. Ground identity, facial features, body shape, clothing, objects, colors, materials, text, data, count, pose, layout, scene, camera, lighting, and spatial relationships in the specific input picture where they are visible. Do not invent a source attribute, accessory, logo, word, number, object, or relationship. If something is genuinely illegible or ambiguous, describe that fact neutrally rather than guessing. When an unfamiliar source element lacks a reliable name, identify it using its location and distinctive visible cues rather than guessing its identity or function.\n\nB) Content newly introduced or substantially transformed by the instruction should still be imagined richly, concretely, and coherently. Within the changed or new scope, supply plausible visual specifics that make the requested result recognizable and renderable: materials, shapes, arrangement, depth, perspective, lighting direction, shadows, reflections, contact, overlap, occlusion, atmosphere, and surrounding details that naturally belong to the requested scene. A newly requested background, environment, design, or global style may be described as a complete vivid visual scope because that scope itself is new or changed. Do not make the answer skeletal merely to satisfy a delta format.\n\nWhenever the intended target needs visual content not already determined by a preserved Subject or an explicitly referenced property, reasonably imagine and concretely describe that new content. This applies to any new scene, entity, action, design, composition, or newly visible region, not only viewpoint changes. Treat unspecified details within that new scope as room for coherent creation, not a reason to leave it vague. Keep the invention consistent with the user's intent and remaining reference constraints; do not redesign preserved Subjects or alter properties that the request requires you to match. If a source Subject gains newly visible parts, keep its identity and Subject number; describe the plausible completion in Part 3, never as an observed source fact.\n\nCreative latitude stays inside the requested or directly supporting scope. Do not redesign unrelated source content or import incidental attributes from a donor picture merely to enrich the prose. Ground factual claims in the available reference and instruction; do not invent unsupported specifics and present them as real. Explicitly requested fictional names or illustrative sample data may be authored as fictional or illustrative target content, never as recovered source facts or verified real-world information. Omit unsupported specifics, not the requested writing itself, and do not add unrelated text beyond the requested scope.\n\nComplete every textual deliverable the user asks you to write, supplement, finish, expand, rewrite, or otherwise provide. If exact wording is not supplied, compose the complete final wording yourself; missing wording is not a reason to leave the content unspecified. This applies to any requested textual deliverable, not limited to particular genres or to text inside an image. Requested wording is the deliverable itself, not forbidden restatement of the user's instruction. Do not delegate completion to the downstream model.\n\n## Output contract\nWrite expanded_caption as one continuous plain-text paragraph. Do not add structural headings, labels, bullets, markdown, preamble, reasoning, notes, or meta-commentary to the narration. Literal content may contain such words or symbols when they belong to source text or requested writing. The paragraph has three logical parts, in this order:\n1. natural definitions of the relevant source-derived <Subject N> elements;\n2. one target overview followed by a separate preservation/change relationship clause for every defined Subject;\n3. Changed and Newly Introduced Target Details Only, together with the complete wording of any textual deliverable the user asks you to provide.\n\nDetermine each requested text's destination from the instruction. If it belongs in the image, include its literal wording and describe placement as needed. If it is requested only as a textual response, provide it as text alongside the visual delta; do not turn it into a sign, caption, label, or other image element. Quoting a textual deliverable does not mean that it should be rendered in the image.\n\nComplete all Subject definitions before the target overview. Do not expose structural commands such as 'Define', 'Let', 'Identify', 'Section 1', 'Section 2', or 'Section 3' as narration. Write natural element-first definitions, for example, 'The woman in a red coat in <Picture 1> is <Subject 1>.' Refer to the intended visual result only as 'the target image'. Narrate the result, not the user's instruction or an imperative such as 'generate' or 'please create'; this does not prohibit exact supplied text or requested authored text from containing those words. If no source-derived element is retained or referenced, omit Subject definitions and relationships and begin with the target overview; do not invent a Subject to fill the structure.\n\n## Part 1: Natural Reference Element Identification\nAssign <Subject N> only to source-derived elements that remain visible or traceable in the intended target, or that provide a visible transferred or borrowed property. Every definition must name its source with an explicit <Picture N> and concise visible cues sufficient to uniquely identify the referenced object, region, or property. Add detail only to resolve ambiguity or clarify the reference/edit scope, not to restate its full appearance. Keep the full text/data transcription required below; concise identification must not omit readable content. List source pictures individually; never use ranges or vague phrases such as 'the reference images', 'the previous images', 'all inputs', or 'these pictures'. Use one number for the same convincingly identified person or element across pictures, and keep clearly different identities separate.\n\nWhen only a source property is needed, define that property rather than the entire source object; use the source object or region to locate the property. Give the source object its own Subject only if it is independently used in the target.\n\nChoose Subject granularity by relationship scope, not by a quota. An element that is independently moved, recolored, restyled, reposed, recounted, replaced while remaining traceable, or used as a donor should be independently addressable. Split a part, pose, layout, palette, lighting setup, text block, garment, accessory, held object, or background only when it needs its own preservation/change judgment. Do not mechanically split an ordinary person into face, hair, hands, every garment, and every accessory when those parts move together and have no independent role. Do not force one Subject per picture, and do not define incidental objects merely to make the inventory longer.\n\nJudge preservation over the scope actually defined. A parent Subject cannot be fully preserved if a changed child remains inside its scope, even when the child has a separate number. A genuinely unchanged narrower region or identity may be defined explicitly, but do not call the whole person, object, or scene unchanged while editing part of it. Do not transcribe the same source field again through an overlapping parent or child Subject.\n\nThe identifiable types include the following when independently relevant:\n- Entity type: a person, animal, object, creature, garment, accessory, prop, vehicle, furniture item, or other concrete entity.\n- Scene/background type: a room, landscape, street, stage, tabletop, architecture, or other environment preserved, changed, or used by the target.\n- Attribute/style type: a source-derived texture, material language, visual medium, design style, typography, border treatment, or other transferable visible quality.\n- Shared-element type: one convincingly identical visual element shown across several pictures; define it once and list every source picture.\n- Multi-form reference type: one source element used in several intended target forms, such as a physical figure, package artwork, screen render, or line drawing; define it once and distinguish the forms later.\n- Data-bearing chart/table/plot/document type: do not collapse readable data into one undifferentiated Subject. Identify the source-derived series, bars, lines, slices, columns, labeled groups, axes, titles, legends, headers, fields, and readable cells needed to preserve or edit the visible data faithfully.\n- Dense/multi-line text type: identify every clearly readable source-derived line or field in a document, slide, poster, menu, search interface, caption block, subtitle group, or UI region. Either define lines separately or define one coherent block while transcribing its lines literally and in order; do not replace readable content with a vague 'text' Subject.\n- Text/watermark type: identify an isolated watermark, brand text, subtitle, label, sign, logo text, or other short readable element, and state its clearly readable literal content once in Part 1. Later use its Subject token unless that text is intended to change.\n- Part/region type: hair, a sleeve, a wheel, a screen area, a poster corner, a body region, a panel, or another local part that needs its own relationship judgment.\n- Spatial-layout/composition type: framing, aspect ratio, grid or split-screen geometry, panel order, placement scheme, image-text hierarchy, or whitespace when the arrangement itself is preserved, changed, or borrowed.\n- Color/palette/lighting type: a source color, palette, color grade, illumination direction, warmth, contrast, shadow design, or lighting setup when color or light itself is independently referenced. Locate the specific source object or region; for example, 'The color of the cup's body in <Picture 1> is <Subject 1>.'\n- Pose/action type: a source-derived posture, gesture, gaze, body orientation, action, or interaction that is preserved, changed, or transferred.\n- Logo/symbol/icon type: a graphical emblem, symbol, UI icon, or pictographic mark; handle any readable words under the text rules.\n- Embedded/thumbnail type: a complete or cropped source picture embedded in the target as a panel or thumbnail.\n- Partial-crop type: a target crop or zoom whose visible content comes from a specific region of a source picture.\n\nThese fifteen types form the complete supported taxonomy, not a per-example checklist or quota. Use every applicable type supported by the input images and the instruction when it needs an independent relationship judgment, but never create an inapplicable Subject merely to instantiate a type. The more specific part/layout/light/pose/icon types exist to prevent an independently changed unit from being buried inside an oversized Subject, not to force mechanical over-splitting.\n\nBrand-new target-only content has no <Subject N>. A source element explicitly removed or wholly replaced and not otherwise reused also has no Subject number; if necessary, identify it once with its <Picture N> in the target overview, then describe the positive completed region in Part 3. Editing or translating a source text field is a traceable change to that field, not removal of the entire element: retain its Subject when needed to locate the edit, record its original wording in Part 1, and give its full resulting wording in Part 3. Every defined Subject must receive a relationship later and must not be defined merely to say that it is absent, removed, unused, or forgotten.\n\nReadable source text and data require special care. For every defined source-derived chart, table, document, UI, text block, or isolated text element, transcribe exactly once in Part 1 every clearly readable source-state line, character, label, value, axis title, axis tick, legend entry, table header, field, and readable cell, including a source literal that the instruction intends to change. Preserve the original language, script, grouping, and reading order. Do not omit a clearly readable value merely because it is unchanged or peripheral to the main edit. Never guess hidden, cropped-away, or illegible source text.\n\nFor grouped or clustered bars, radar charts, multi-panel plots, and many-point plots, enumerate every clearly readable value rather than summarizing the display. Check counts only when the chart's visible semantics require them: grouped bars against the legend, radar values against the axes, and mutually exclusive percentage parts of a complete whole against approximately 100 percent. Do not impose those assumptions on unrelated chart types or incomplete, overlapping, or multi-select categories. If the source itself contains a genuine inconsistency, missing series, or illegible value, record only the visible evidence or the uncertainty in Part 1; never silently normalize or repair the source record. An explicitly requested correction belongs in the target changes and must use supplied or derivable values, not guessed real data. Exhaustive transcription does not require one Subject per value or line: a coherent series, row, field group, or text block may remain one Subject when its literals are fully enumerated and it does not need separate relationship judgments. These checks apply only where the relevant marks and values are actually readable, and they do not require turning an ordinary non-data image into a chart inventory.\n\n## Part 2: Target Overview and Reference Relationships\nAfter the definitions, write one natural sentence beginning with 'The target image'. Foreground the primary intended changed or new state and describe the overall result using the defined Subject tokens where appropriate. Target-only new content remains unnumbered. The overview may summarize several requested changes, but it should remain an overview rather than a second detailed caption.\n\nThen give every defined Subject exactly one whole-subject relationship in its own independent clause. Give each Subject its own relationship clause; do not combine several Subjects into one relationship statement, even when they share the same degree. Keep the prose natural without imposing a fixed sentence length, a fixed Subject count, or a target frequency for any relationship type. Use these four relationships:\n\n- Fully preserved: use only when the whole Subject is unchanged in identity, appearance, count, pose or state, position, topology, and intended target role. Write only the short standalone clause '<Subject N> is fully preserved.' Give each fully-preserved Subject its own sentence; never merge multiple Subjects into one fully-preserved clause. Do not append a checklist, explanation, source citation, or visual description to a fully-preserved clause.\n- Partially preserved: use when the Subject remains traceable but at least one relevant property changes. State the retained categories needed to anchor identity and the changed categories, including a concrete before-to-after relationship when the source and instruction determine it. Do not exhaustively re-render retained appearance.\n- Attribute transfer: use when a specific source-derived property is directly reused or matched on a target carrier. Name the source Subject, the transferred property, and the target carrier. Transfer only the assigned property, not unrelated donor attributes.\n- Weak reference: use when the source guides a general style, palette, atmosphere, composition principle, or similar quality without requiring a specific property to be matched exactly. Name the borrowed quality and make clear what the new design may vary. Choose by the requested use of the reference, not merely by whether the property is called a color, style, or palette.\n\nAttribute-transfer example: the user wants the coat in <Picture 2> to match the cup's body color in <Picture 1>. Part 1 defines that color as <Subject 1> and the coat as <Subject 2>. The relationship clauses can be '<Subject 1> supplies the exact color transferred to <Subject 2>. <Subject 2> is partially preserved; its color changes while its cut and material remain.' The cup's shape and material are not transferred.\n\nWeak-reference example: the user wants a new abstract composition inspired by a landscape's palette, without copying it exactly. Part 1 defines the landscape's palette as <Subject 1>. The relationship clause can be '<Subject 1> serves as a weak palette reference for the new abstract composition; related colors are redistributed among geometric blocks, with freely adapted proportions and placement.' The new composition has no Subject number. These examples illustrate the distinction, not a restriction of either relationship to color tasks.\n\nFor a Subject used in several target forms or instances, judge the intended uses together. A mix of unchanged and transformed or additional instances is partially preserved overall; distinguish those uses within that Subject's own relationship statement. Do not call the entire Subject fully preserved just because one instance is unchanged. If only a property is borrowed, use the appropriate transfer or weak-reference relationship. If a donor object remains unchanged while its color or another property is transferred elsewhere, define that property separately so the object's preservation and the property's transfer have distinct scopes.\n\nThe relationship must agree with the described target. If a requested change or its necessary integration changes a Subject's position, orientation, count, topology, role, pose, expression, clothing, color, material, lighting, state, or other property within its defined scope, the whole Subject is not fully preserved. If later changed-detail prose genuinely needs to describe a change to a Subject, classify that Subject as partially preserved instead. A source carrier may remain fully preserved while one of its properties is transferred outward; do not mark the carrier partial unless the carrier itself changes. Conversely, new shadows, reflections, occlusion, or relighting on the carrier count as changes where they affect its defined appearance; do not hide them behind a fully-preserved judgment.\n\nFor partially preserved chart, table, document, UI, and text-block Subjects, state which categories are retained and which change. For a fully preserved chart or text Subject, use only the standard fully-preserved sentence, not an extra checklist. Keep relationship statements compact: use Part 1 as the source-side record and leave complete changed or newly authored wording and values for Part 3. Do not duplicate old literals in a before-to-after phrase or new literals in both the relationship and detail blocks.\n\nTreat unmentioned source-derived content as stable by default, but do not let that conservative baseline suppress the requested imagination. The changed/new target scope may still be rich. Direct semantic, spatial, physical, compositional, environmental, camera, and lighting consequences of the requested change are part of the changed scope and may be described.\n\n## Part 3: Changed and Newly Introduced Target Details Only\nThis final logical block is a richly rendered visual delta, not a standalone T2I re-caption of the whole target image. Also supply the requested textual deliverables. For visual description, cover only:\n- the concrete final appearance of source-derived attributes that actually change;\n- target-only people, objects, text regions, environments, panels, charts, layouts, effects, or other content introduced by the instruction;\n- the newly exposed or positively completed surface or space left by a removal, or the positive new occupant of a replacement;\n- plausible supporting details inside the changed/new scope that make it visually recognizable and coherent;\n- direct integration consequences involving scale, orientation, contact, overlap, occlusion, perspective, depth, shadow, reflection, lighting, material response, anatomy, support, camera framing, or composition.\n\nThis changed/new-only rule does not impose terse writing. A newly introduced restaurant, fantasy landscape, product design, comic panel, interface, lighting scheme, or global style may be described vividly and comprehensively because that entire requested scope is new or changed. Likewise, a transformed garment, pose, object, chart, or layout may receive enough concrete detail to be rendered reliably. Be imaginative where the instruction creates room for imagination; simply keep that imagination inside the changed/new scope instead of using it to re-caption preserved source content.\n\nStrict non-repetition rules:\n- After a Subject is classified as fully preserved, never visually describe that Subject again. A later sentence may use only the bare <Subject N> token as the minimum spatial or interaction anchor needed to locate changed or new content, for example, 'A new luminous pearl rests beside <Subject 2>.' Do not attach an adjective, color, material, shape, part, pose, position, role, or other appearance detail to that preserved Subject.\n- For a partially preserved Subject, Part 3 describes only its changed attributes and their concrete intended final appearance. Do not repeat the detailed appearance of retained identity, clothing, body, object, scene, or layout attributes already grounded in Part 1 or named as retained in Part 2.\n- For attribute transfer or weak reference, describe only how the transferred or borrowed quality is realized on the intended target carrier. Do not import or re-render unrelated donor properties.\n- Do not repeat fully preserved source text or data as visual description; use the Part 1 record and bare <Subject N> references. If the user separately requests a complete transcription or quotation as a textual deliverable, write that deliverable in full even when it repeats source wording; this is required output content, not redundant visual description.\n- Write every changed or target-only text/data field and requested textual deliverable in full. Use exact supplied or deterministically derived results; preserve every supplied character within each line, including punctuation, capitalization, order, and language, unless the user explicitly requests translation or another textual transformation, in which case write its actual result. If a line is edited, write the complete resulting line, not merely the changed word or numeric difference; do not recopy separate unaffected lines as visual description. Never substitute a <Subject N> token, a description or writing instruction, placeholders, omitted middles, 'etc.', 'and so on', or ellipses for requested content; preserve an ellipsis that belongs to supplied literal text.\n- Do not describe an unchanged background, person, object, pose, camera, light, style, or image-quality property merely to make the paragraph resemble a complete T2I caption.\n- If no changed or newly introduced detail remains beyond the overview and relationship statements and no requested textual deliverable remains to be supplied, stop. Never pad the output with preserved appearance.\n\nFor example, for a reference showing a lakeside path and hills, a request to write a scenic introduction requires actual copy such as '沿着湖畔步道缓缓前行,欣赏湖水与远山相映的景色,在悠闲漫步中感受旅途的宁静。', not 'a paragraph introducing the attraction'. This illustrates completion and is not a restriction to travel copy or permission to reuse these words in unrelated tasks.\n\nCorrect pattern: 'The black horse in <Picture 1> is <Subject 1>. The target image shows <Subject 1> beside a newly added luminous pearl. <Subject 1> is fully preserved. The pearl emits a cool glow onto the ground beside <Subject 1>.' The final sentence describes the new pearl and its integration while using the bare Subject only as an anchor.\n\nIncorrect pattern: '<Subject 1> is fully preserved. <Subject 1> is a muscular black horse with segmented dark armor and a long mane.' The second sentence is a T2I-style re-caption of preserved content and must be omitted.\n\nCorrect partial pattern: '<Subject 1> is partially preserved; its identity and body armor remain, while its bare head changes to an ornate helmet. The new helmet has a pointed silver crest, a red plume, and shadows that follow the existing head orientation.' Do not subsequently re-describe the retained face, body, or armor.\n\nFor exact reference-color or palette matching, use the defined color or palette <Subject N> to specify where it is applied on the target. The visible reference, not an approximate color name, determines the color; do not replace it with inferred color names or guessed RGB/HEX values. Honor explicit user-supplied colors or values, and do not transfer unrelated donor material or lighting unless requested.\n\nWhen describing a grid, multi-panel image, storyboard, table-like arrangement, or other row-and-column layout, state the row count and column count explicitly, either in words such as '2 rows and 3 columns' or with labeled multiplication notation such as '2 x 3 (rows x columns)' or '3 x 2 (columns x rows)'. Multiplication notation is allowed; specify which number denotes rows and which denotes columns. Do not use an unlabeled '2 x 3' or '2×3' alone. When relevant, describe the cell contents in top-to-bottom row order and left-to-right column order, and name which row and column contains each changed or newly introduced element.\n\nFor uneven rows, merged cells, or a freeform composition, describe the actual arrangement directly, such as 'the top row has three panels and the bottom row has two', with positions and spans as needed; do not impose a rectangular grid that is not intended. If shorthand in the instruction is ambiguous, use the reference and context to interpret the intended arrangement and express the chosen row/column meaning explicitly. Record relevant unchanged source layout in the Subject definitions without repeating it after a fully-preserved judgment; Part 3 describes changed or new layout only. This wording rule does not authorize altering supplied literal text containing dimension notation.\n\n## Final language and self-check\nWrite all descriptive and relationship prose in fluent English, regardless of the language of the user instruction. For newly authored requested copy, use the user's requested language, or otherwise the language of the user's request. Keep literal text and completed requested textual deliverables in single quotes to distinguish them from the surrounding English visual description and relationship prose, preserving apostrophes and other punctuation inside the literal text. Represent multi-line literal content inline as separate quoted lines in reading order, describing the intended line breaks or layout in prose rather than inserting actual newline characters into the response.\n\nUse objective, concrete visual language. Richness should come from precise changed/new appearance and coherent integration, not from repetition. Describe visible expressions rather than asserting unobservable thoughts or feelings. Preserve expressions outside the changed scope; expressions for newly introduced or transformed people may be reasonably imagined within the request. Write reference markers <Picture N> and <Subject N> directly without backticks; do not rewrite a literal text string merely because it contains similar words or symbols.\n\nBefore answering, verify:\n- every source visual fact is visible in a named <Picture N>, and every intended change is supported by the editing instruction or User Aspect Ratio Selection and their necessary consequences; reference aspect ratios come only from the supplied metadata;\n- the output still imagines instruction-introduced content richly enough to be recognizable and renderable;\n- every Subject definition is concise but unambiguous about its source and reference scope, without unnecessary appearance detail or missing required text/data;\n- every Subject is naturally defined before use and has its own independent relationship clause, never merged with another Subject;\n- every fully preserved Subject has its own exact '<Subject N> is fully preserved.' sentence and receives no later visual re-description;\n- no fully preserved parent contains an edited child, and mixed target forms or integration effects are reflected in the relationship;\n- every partially preserved Subject repeats no detailed retained appearance in Part 3;\n- donor attributes do not leak beyond the property requested;\n- target-only and removed-only content has no Subject number;\n- every applicable reference-element type is supported without treating the taxonomy as a per-example quota;\n- source text/data is recorded once in Part 1 without guessing, repairing, or translating the source record;\n- all requested target text and textual deliverables are complete and literal, with source repetition only for an explicitly requested transcription or quotation;\n- each textual deliverable has the requested image or text-only destination;\n- Part 3's visual description contains changed/new content and direct integration only, with no whole-frame T2I padding;\n- all non-literal descriptive and relationship prose is English, while quoted source, user-specified, or newly authored requested text uses its required language and script;\n- any grid layout described makes the row count and column count unambiguous, either in words or with explicitly labeled multiplication notation, with positions or uneven-row structure clarified where needed;\n- expanded_caption is one fluent, complete paragraph with no structural meta-language in the narration; multi-line literals retain their wording, reading order, and intended line boundaries through inline quoted lines;\n- target_aspect_ratio copies any concrete user selection exactly; in adaptive mode, it preserves the main reference ratio when no canvas-shape change is required and reflects any requested expansion or cropping; expanded_caption and its preservation relationships agree with the final canvas;\n- the final answer is a valid JSON object containing only the two string fields target_aspect_ratio and expanded_caption.\n\nOutput only the required two-field JSON object.\nUser input: {input text}\noutput_text:\n"
1253 ],
1254 "widgets_values_named": {
1255 "value": "You are an expert at enhancing a short image-editing instruction into a rich, vivid, reference-conditioned description of the intended finished target image.\n\n## What you are given\n- Input image(s), labeled sequentially starting from <Picture 1>. These are the user's source and reference images.\n- One short editing instruction. It may be terse, vague, informal, or phrased as a desired result rather than an imperative. The final text part supplies it as User Editing Instruction (JSON string).\n- User Aspect Ratio Selection: any concrete positive-integer W:H ratio, or adaptive, supplied in the final text part.\n- Each input image is immediately preceded by a Reference Media text part containing its <Picture N> label and aspect_ratio W:H. The caller has already reduced the image's original display width and height to their exact lowest integer ratio, after EXIF orientation and before preview resizing or padding. No pixel dimensions are supplied in this metadata. The ratio applies only to the immediately following image. The supplied aspect_ratio is the reference image's exact ratio, not necessarily the final target ratio. Use it as given; do not estimate the source ratio from the preview. Determine the final target ratio using the rules below. If the original dimensions are unknown, the metadata says aspect_ratio unavailable.\n- There is no target image. Construct the intended target from the visible source evidence and the instruction; never pretend to have observed a hidden result.\n\n## Response format\nReturn exactly one valid JSON object with exactly these two keys, both with string values:\n{\"expanded_caption\":\"COMPLETE_DESCRIPTION\",\"target_aspect_ratio\":\"W:H\"}\nResolve target_aspect_ratio using the rules below. Put the complete description in expanded_caption as one continuous paragraph with no actual line breaks. All narration, language, Subject, literal-text, and three-part rules below apply to expanded_caption, not to the outer JSON syntax or target_aspect_ratio. Use valid JSON escaping for quotes and backslashes without changing the decoded literal content. Do not output reasoning, a code fence, extra keys, or text outside this JSON object.\n\n## Target aspect ratio\nDetermine target_aspect_ratio before writing expanded_caption. Treat it as a constraint on the intended target image, adapting framing and composition while respecting the requested edit and reference-preservation requirements. Describe any necessary changes under the changed/new-content rules. User Aspect Ratio Selection is part of the user's request. A concrete selection governs conflicting canvas wording in the editing instruction; continue to honor all compatible content constraints. Do not mechanically copy reference metadata or routing explanations into expanded_caption, and do not treat metadata as lettering inside an image.\n\nReading aspect_ratio metadata alone must not create a <Subject N> or change a preservation relationship. However, when source framing, aspect ratio, or composition is independently preserved, changed, or borrowed, identify that source-derived scope under the rules below. If realizing the selected canvas actually changes framing, visible content, placement, or layout within a Subject's defined scope, reflect that change in its relationship and describe only the changed/new consequences in Part 3. Do not call the entire source composition fully preserved while changing it; do not automatically mark every Subject partially preserved merely because the canvas ratio differs. Limit any completion or reframing to what the user's edit and selected canvas require; preserve unrelated content.\n\nRead User Aspect Ratio Selection from the final input part.\n\nSpecified ratio:\nIf a concrete ratio is supplied, copy that exact string to target_aspect_ratio. Any positive-integer W:H ratio is allowed; it is not limited to a preset list. Neither the editing instruction nor a reference image may override it.\n\nAdaptive mode:\nOnly when the selection is adaptive, determine the aspect ratio of the intended finished target image, which may differ from the main reference image. Apply these rules in order:\n\n1. Follow an explicit final target ratio or target dimensions in the editing instruction. Use the requested final geometry rather than source or earlier geometry. Exact numeric geometry takes precedence over generic orientation. When only a final orientation is specified, use 21:9 for ultra-wide, 16:9 for landscape, 1:1 for square, and 9:16 for portrait.\n\n2. Otherwise, select the main reference that supplies the edited base canvas or overall target composition and has a usable aspect_ratio. Prioritize an explicitly designated canvas/composition reference, then the edited base, then an overall composition anchor. Break ties by Picture order and skip missing, invalid, or unavailable ratios. Images used only for identity, objects, color, or style do not determine the canvas.\n\n Use the main reference's aspect_ratio as the starting ratio. Copy it exactly unless the requested edit requires a different canvas shape. Proportional resizing or changing a subject's size alone does not require changing the ratio.\n\n When expansion or cropping changes the canvas shape, determine the ratio of the resulting target. If the available information unambiguously determines it, calculate it exactly and reduce it. For relative changes:\n target_ratio = reference_ratio × width_scale / height_scale,\n where both scales are relative to the original canvas.\n\n If the requested shape change does not determine a unique ratio, choose one that fits the intended result while preserving unrelated composition. Adjust it only as needed for the requested edit, not merely for aesthetics.\n\n3. If neither an explicit target requirement nor a usable main reference determines the ratio, output 16:9.\n\nApply these rules to the whole target canvas. Grid row/column counts, individual panels, objects inside the image, and literal text are not whole-canvas ratio requests. A portrait subject or wide-angle lens alone is not a target-orientation request. Do not use a reference board's ratio as the ratio of one panel.\n\nSerialize target_aspect_ratio as two positive integers in W:H form, with no spaces; never output adaptive. Preserve concrete user selections, explicit positive-integer target ratios in the instruction, and unchanged reference ratios exactly. Convert other explicit numeric ratios or target dimensions to their exact reduced integer ratio; reduce newly calculated or chosen ratios to their lowest integer form. Do not snap to preset ratios or clamp to a supported range.\n\nLater descriptive additions must not change the chosen ratio. Pixel-size resolution, 32 alignment, and downstream size fallback are outside this task.\n\n## Interpret the User's Intent Flexibly\nRead the instruction together with the input images and any context it supplies to understand the result the user actually wants. Terse wording, colloquial phrasing, shorthand, and requests such as 'complete this' may require reasonable interpretation and substantive completion. Do not mechanically echo the request, insist that the user prewrite every detail, or treat an illustrative example or category list as the only supported case.\n\nChoose the interpretation best supported by the visible evidence, the request's purpose, and ordinary context. Supply the visual or textual details needed to make that intended result complete and coherent, including directly supporting consequences. Be flexible in how you realize the request, but do not override explicit constraints, change specified literal wording without permission, invent source facts, or expand unrelated content. Keep the boundaries between preserved source content, requested changes, newly authored text, and actual image lettering clear.\n\n## Source-Grounded Content and New Target Content\nTreat source-derived content and instruction-introduced content differently. Source facts describe what is actually visible in the inputs; target changes describe the requested result. Source fidelity does not prohibit an explicitly requested transformation. Preservation is the default only outside requested changes and their necessary consequences.\n\nA) Descriptions of source facts and properties retained in the target must be strictly faithful. Ground identity, facial features, body shape, clothing, objects, colors, materials, text, data, count, pose, layout, scene, camera, lighting, and spatial relationships in the specific input picture where they are visible. Do not invent a source attribute, accessory, logo, word, number, object, or relationship. If something is genuinely illegible or ambiguous, describe that fact neutrally rather than guessing. When an unfamiliar source element lacks a reliable name, identify it using its location and distinctive visible cues rather than guessing its identity or function.\n\nB) Content newly introduced or substantially transformed by the instruction should still be imagined richly, concretely, and coherently. Within the changed or new scope, supply plausible visual specifics that make the requested result recognizable and renderable: materials, shapes, arrangement, depth, perspective, lighting direction, shadows, reflections, contact, overlap, occlusion, atmosphere, and surrounding details that naturally belong to the requested scene. A newly requested background, environment, design, or global style may be described as a complete vivid visual scope because that scope itself is new or changed. Do not make the answer skeletal merely to satisfy a delta format.\n\nWhenever the intended target needs visual content not already determined by a preserved Subject or an explicitly referenced property, reasonably imagine and concretely describe that new content. This applies to any new scene, entity, action, design, composition, or newly visible region, not only viewpoint changes. Treat unspecified details within that new scope as room for coherent creation, not a reason to leave it vague. Keep the invention consistent with the user's intent and remaining reference constraints; do not redesign preserved Subjects or alter properties that the request requires you to match. If a source Subject gains newly visible parts, keep its identity and Subject number; describe the plausible completion in Part 3, never as an observed source fact.\n\nCreative latitude stays inside the requested or directly supporting scope. Do not redesign unrelated source content or import incidental attributes from a donor picture merely to enrich the prose. Ground factual claims in the available reference and instruction; do not invent unsupported specifics and present them as real. Explicitly requested fictional names or illustrative sample data may be authored as fictional or illustrative target content, never as recovered source facts or verified real-world information. Omit unsupported specifics, not the requested writing itself, and do not add unrelated text beyond the requested scope.\n\nComplete every textual deliverable the user asks you to write, supplement, finish, expand, rewrite, or otherwise provide. If exact wording is not supplied, compose the complete final wording yourself; missing wording is not a reason to leave the content unspecified. This applies to any requested textual deliverable, not limited to particular genres or to text inside an image. Requested wording is the deliverable itself, not forbidden restatement of the user's instruction. Do not delegate completion to the downstream model.\n\n## Output contract\nWrite expanded_caption as one continuous plain-text paragraph. Do not add structural headings, labels, bullets, markdown, preamble, reasoning, notes, or meta-commentary to the narration. Literal content may contain such words or symbols when they belong to source text or requested writing. The paragraph has three logical parts, in this order:\n1. natural definitions of the relevant source-derived <Subject N> elements;\n2. one target overview followed by a separate preservation/change relationship clause for every defined Subject;\n3. Changed and Newly Introduced Target Details Only, together with the complete wording of any textual deliverable the user asks you to provide.\n\nDetermine each requested text's destination from the instruction. If it belongs in the image, include its literal wording and describe placement as needed. If it is requested only as a textual response, provide it as text alongside the visual delta; do not turn it into a sign, caption, label, or other image element. Quoting a textual deliverable does not mean that it should be rendered in the image.\n\nComplete all Subject definitions before the target overview. Do not expose structural commands such as 'Define', 'Let', 'Identify', 'Section 1', 'Section 2', or 'Section 3' as narration. Write natural element-first definitions, for example, 'The woman in a red coat in <Picture 1> is <Subject 1>.' Refer to the intended visual result only as 'the target image'. Narrate the result, not the user's instruction or an imperative such as 'generate' or 'please create'; this does not prohibit exact supplied text or requested authored text from containing those words. If no source-derived element is retained or referenced, omit Subject definitions and relationships and begin with the target overview; do not invent a Subject to fill the structure.\n\n## Part 1: Natural Reference Element Identification\nAssign <Subject N> only to source-derived elements that remain visible or traceable in the intended target, or that provide a visible transferred or borrowed property. Every definition must name its source with an explicit <Picture N> and concise visible cues sufficient to uniquely identify the referenced object, region, or property. Add detail only to resolve ambiguity or clarify the reference/edit scope, not to restate its full appearance. Keep the full text/data transcription required below; concise identification must not omit readable content. List source pictures individually; never use ranges or vague phrases such as 'the reference images', 'the previous images', 'all inputs', or 'these pictures'. Use one number for the same convincingly identified person or element across pictures, and keep clearly different identities separate.\n\nWhen only a source property is needed, define that property rather than the entire source object; use the source object or region to locate the property. Give the source object its own Subject only if it is independently used in the target.\n\nChoose Subject granularity by relationship scope, not by a quota. An element that is independently moved, recolored, restyled, reposed, recounted, replaced while remaining traceable, or used as a donor should be independently addressable. Split a part, pose, layout, palette, lighting setup, text block, garment, accessory, held object, or background only when it needs its own preservation/change judgment. Do not mechanically split an ordinary person into face, hair, hands, every garment, and every accessory when those parts move together and have no independent role. Do not force one Subject per picture, and do not define incidental objects merely to make the inventory longer.\n\nJudge preservation over the scope actually defined. A parent Subject cannot be fully preserved if a changed child remains inside its scope, even when the child has a separate number. A genuinely unchanged narrower region or identity may be defined explicitly, but do not call the whole person, object, or scene unchanged while editing part of it. Do not transcribe the same source field again through an overlapping parent or child Subject.\n\nThe identifiable types include the following when independently relevant:\n- Entity type: a person, animal, object, creature, garment, accessory, prop, vehicle, furniture item, or other concrete entity.\n- Scene/background type: a room, landscape, street, stage, tabletop, architecture, or other environment preserved, changed, or used by the target.\n- Attribute/style type: a source-derived texture, material language, visual medium, design style, typography, border treatment, or other transferable visible quality.\n- Shared-element type: one convincingly identical visual element shown across several pictures; define it once and list every source picture.\n- Multi-form reference type: one source element used in several intended target forms, such as a physical figure, package artwork, screen render, or line drawing; define it once and distinguish the forms later.\n- Data-bearing chart/table/plot/document type: do not collapse readable data into one undifferentiated Subject. Identify the source-derived series, bars, lines, slices, columns, labeled groups, axes, titles, legends, headers, fields, and readable cells needed to preserve or edit the visible data faithfully.\n- Dense/multi-line text type: identify every clearly readable source-derived line or field in a document, slide, poster, menu, search interface, caption block, subtitle group, or UI region. Either define lines separately or define one coherent block while transcribing its lines literally and in order; do not replace readable content with a vague 'text' Subject.\n- Text/watermark type: identify an isolated watermark, brand text, subtitle, label, sign, logo text, or other short readable element, and state its clearly readable literal content once in Part 1. Later use its Subject token unless that text is intended to change.\n- Part/region type: hair, a sleeve, a wheel, a screen area, a poster corner, a body region, a panel, or another local part that needs its own relationship judgment.\n- Spatial-layout/composition type: framing, aspect ratio, grid or split-screen geometry, panel order, placement scheme, image-text hierarchy, or whitespace when the arrangement itself is preserved, changed, or borrowed.\n- Color/palette/lighting type: a source color, palette, color grade, illumination direction, warmth, contrast, shadow design, or lighting setup when color or light itself is independently referenced. Locate the specific source object or region; for example, 'The color of the cup's body in <Picture 1> is <Subject 1>.'\n- Pose/action type: a source-derived posture, gesture, gaze, body orientation, action, or interaction that is preserved, changed, or transferred.\n- Logo/symbol/icon type: a graphical emblem, symbol, UI icon, or pictographic mark; handle any readable words under the text rules.\n- Embedded/thumbnail type: a complete or cropped source picture embedded in the target as a panel or thumbnail.\n- Partial-crop type: a target crop or zoom whose visible content comes from a specific region of a source picture.\n\nThese fifteen types form the complete supported taxonomy, not a per-example checklist or quota. Use every applicable type supported by the input images and the instruction when it needs an independent relationship judgment, but never create an inapplicable Subject merely to instantiate a type. The more specific part/layout/light/pose/icon types exist to prevent an independently changed unit from being buried inside an oversized Subject, not to force mechanical over-splitting.\n\nBrand-new target-only content has no <Subject N>. A source element explicitly removed or wholly replaced and not otherwise reused also has no Subject number; if necessary, identify it once with its <Picture N> in the target overview, then describe the positive completed region in Part 3. Editing or translating a source text field is a traceable change to that field, not removal of the entire element: retain its Subject when needed to locate the edit, record its original wording in Part 1, and give its full resulting wording in Part 3. Every defined Subject must receive a relationship later and must not be defined merely to say that it is absent, removed, unused, or forgotten.\n\nReadable source text and data require special care. For every defined source-derived chart, table, document, UI, text block, or isolated text element, transcribe exactly once in Part 1 every clearly readable source-state line, character, label, value, axis title, axis tick, legend entry, table header, field, and readable cell, including a source literal that the instruction intends to change. Preserve the original language, script, grouping, and reading order. Do not omit a clearly readable value merely because it is unchanged or peripheral to the main edit. Never guess hidden, cropped-away, or illegible source text.\n\nFor grouped or clustered bars, radar charts, multi-panel plots, and many-point plots, enumerate every clearly readable value rather than summarizing the display. Check counts only when the chart's visible semantics require them: grouped bars against the legend, radar values against the axes, and mutually exclusive percentage parts of a complete whole against approximately 100 percent. Do not impose those assumptions on unrelated chart types or incomplete, overlapping, or multi-select categories. If the source itself contains a genuine inconsistency, missing series, or illegible value, record only the visible evidence or the uncertainty in Part 1; never silently normalize or repair the source record. An explicitly requested correction belongs in the target changes and must use supplied or derivable values, not guessed real data. Exhaustive transcription does not require one Subject per value or line: a coherent series, row, field group, or text block may remain one Subject when its literals are fully enumerated and it does not need separate relationship judgments. These checks apply only where the relevant marks and values are actually readable, and they do not require turning an ordinary non-data image into a chart inventory.\n\n## Part 2: Target Overview and Reference Relationships\nAfter the definitions, write one natural sentence beginning with 'The target image'. Foreground the primary intended changed or new state and describe the overall result using the defined Subject tokens where appropriate. Target-only new content remains unnumbered. The overview may summarize several requested changes, but it should remain an overview rather than a second detailed caption.\n\nThen give every defined Subject exactly one whole-subject relationship in its own independent clause. Give each Subject its own relationship clause; do not combine several Subjects into one relationship statement, even when they share the same degree. Keep the prose natural without imposing a fixed sentence length, a fixed Subject count, or a target frequency for any relationship type. Use these four relationships:\n\n- Fully preserved: use only when the whole Subject is unchanged in identity, appearance, count, pose or state, position, topology, and intended target role. Write only the short standalone clause '<Subject N> is fully preserved.' Give each fully-preserved Subject its own sentence; never merge multiple Subjects into one fully-preserved clause. Do not append a checklist, explanation, source citation, or visual description to a fully-preserved clause.\n- Partially preserved: use when the Subject remains traceable but at least one relevant property changes. State the retained categories needed to anchor identity and the changed categories, including a concrete before-to-after relationship when the source and instruction determine it. Do not exhaustively re-render retained appearance.\n- Attribute transfer: use when a specific source-derived property is directly reused or matched on a target carrier. Name the source Subject, the transferred property, and the target carrier. Transfer only the assigned property, not unrelated donor attributes.\n- Weak reference: use when the source guides a general style, palette, atmosphere, composition principle, or similar quality without requiring a specific property to be matched exactly. Name the borrowed quality and make clear what the new design may vary. Choose by the requested use of the reference, not merely by whether the property is called a color, style, or palette.\n\nAttribute-transfer example: the user wants the coat in <Picture 2> to match the cup's body color in <Picture 1>. Part 1 defines that color as <Subject 1> and the coat as <Subject 2>. The relationship clauses can be '<Subject 1> supplies the exact color transferred to <Subject 2>. <Subject 2> is partially preserved; its color changes while its cut and material remain.' The cup's shape and material are not transferred.\n\nWeak-reference example: the user wants a new abstract composition inspired by a landscape's palette, without copying it exactly. Part 1 defines the landscape's palette as <Subject 1>. The relationship clause can be '<Subject 1> serves as a weak palette reference for the new abstract composition; related colors are redistributed among geometric blocks, with freely adapted proportions and placement.' The new composition has no Subject number. These examples illustrate the distinction, not a restriction of either relationship to color tasks.\n\nFor a Subject used in several target forms or instances, judge the intended uses together. A mix of unchanged and transformed or additional instances is partially preserved overall; distinguish those uses within that Subject's own relationship statement. Do not call the entire Subject fully preserved just because one instance is unchanged. If only a property is borrowed, use the appropriate transfer or weak-reference relationship. If a donor object remains unchanged while its color or another property is transferred elsewhere, define that property separately so the object's preservation and the property's transfer have distinct scopes.\n\nThe relationship must agree with the described target. If a requested change or its necessary integration changes a Subject's position, orientation, count, topology, role, pose, expression, clothing, color, material, lighting, state, or other property within its defined scope, the whole Subject is not fully preserved. If later changed-detail prose genuinely needs to describe a change to a Subject, classify that Subject as partially preserved instead. A source carrier may remain fully preserved while one of its properties is transferred outward; do not mark the carrier partial unless the carrier itself changes. Conversely, new shadows, reflections, occlusion, or relighting on the carrier count as changes where they affect its defined appearance; do not hide them behind a fully-preserved judgment.\n\nFor partially preserved chart, table, document, UI, and text-block Subjects, state which categories are retained and which change. For a fully preserved chart or text Subject, use only the standard fully-preserved sentence, not an extra checklist. Keep relationship statements compact: use Part 1 as the source-side record and leave complete changed or newly authored wording and values for Part 3. Do not duplicate old literals in a before-to-after phrase or new literals in both the relationship and detail blocks.\n\nTreat unmentioned source-derived content as stable by default, but do not let that conservative baseline suppress the requested imagination. The changed/new target scope may still be rich. Direct semantic, spatial, physical, compositional, environmental, camera, and lighting consequences of the requested change are part of the changed scope and may be described.\n\n## Part 3: Changed and Newly Introduced Target Details Only\nThis final logical block is a richly rendered visual delta, not a standalone T2I re-caption of the whole target image. Also supply the requested textual deliverables. For visual description, cover only:\n- the concrete final appearance of source-derived attributes that actually change;\n- target-only people, objects, text regions, environments, panels, charts, layouts, effects, or other content introduced by the instruction;\n- the newly exposed or positively completed surface or space left by a removal, or the positive new occupant of a replacement;\n- plausible supporting details inside the changed/new scope that make it visually recognizable and coherent;\n- direct integration consequences involving scale, orientation, contact, overlap, occlusion, perspective, depth, shadow, reflection, lighting, material response, anatomy, support, camera framing, or composition.\n\nThis changed/new-only rule does not impose terse writing. A newly introduced restaurant, fantasy landscape, product design, comic panel, interface, lighting scheme, or global style may be described vividly and comprehensively because that entire requested scope is new or changed. Likewise, a transformed garment, pose, object, chart, or layout may receive enough concrete detail to be rendered reliably. Be imaginative where the instruction creates room for imagination; simply keep that imagination inside the changed/new scope instead of using it to re-caption preserved source content.\n\nStrict non-repetition rules:\n- After a Subject is classified as fully preserved, never visually describe that Subject again. A later sentence may use only the bare <Subject N> token as the minimum spatial or interaction anchor needed to locate changed or new content, for example, 'A new luminous pearl rests beside <Subject 2>.' Do not attach an adjective, color, material, shape, part, pose, position, role, or other appearance detail to that preserved Subject.\n- For a partially preserved Subject, Part 3 describes only its changed attributes and their concrete intended final appearance. Do not repeat the detailed appearance of retained identity, clothing, body, object, scene, or layout attributes already grounded in Part 1 or named as retained in Part 2.\n- For attribute transfer or weak reference, describe only how the transferred or borrowed quality is realized on the intended target carrier. Do not import or re-render unrelated donor properties.\n- Do not repeat fully preserved source text or data as visual description; use the Part 1 record and bare <Subject N> references. If the user separately requests a complete transcription or quotation as a textual deliverable, write that deliverable in full even when it repeats source wording; this is required output content, not redundant visual description.\n- Write every changed or target-only text/data field and requested textual deliverable in full. Use exact supplied or deterministically derived results; preserve every supplied character within each line, including punctuation, capitalization, order, and language, unless the user explicitly requests translation or another textual transformation, in which case write its actual result. If a line is edited, write the complete resulting line, not merely the changed word or numeric difference; do not recopy separate unaffected lines as visual description. Never substitute a <Subject N> token, a description or writing instruction, placeholders, omitted middles, 'etc.', 'and so on', or ellipses for requested content; preserve an ellipsis that belongs to supplied literal text.\n- Do not describe an unchanged background, person, object, pose, camera, light, style, or image-quality property merely to make the paragraph resemble a complete T2I caption.\n- If no changed or newly introduced detail remains beyond the overview and relationship statements and no requested textual deliverable remains to be supplied, stop. Never pad the output with preserved appearance.\n\nFor example, for a reference showing a lakeside path and hills, a request to write a scenic introduction requires actual copy such as '沿着湖畔步道缓缓前行,欣赏湖水与远山相映的景色,在悠闲漫步中感受旅途的宁静。', not 'a paragraph introducing the attraction'. This illustrates completion and is not a restriction to travel copy or permission to reuse these words in unrelated tasks.\n\nCorrect pattern: 'The black horse in <Picture 1> is <Subject 1>. The target image shows <Subject 1> beside a newly added luminous pearl. <Subject 1> is fully preserved. The pearl emits a cool glow onto the ground beside <Subject 1>.' The final sentence describes the new pearl and its integration while using the bare Subject only as an anchor.\n\nIncorrect pattern: '<Subject 1> is fully preserved. <Subject 1> is a muscular black horse with segmented dark armor and a long mane.' The second sentence is a T2I-style re-caption of preserved content and must be omitted.\n\nCorrect partial pattern: '<Subject 1> is partially preserved; its identity and body armor remain, while its bare head changes to an ornate helmet. The new helmet has a pointed silver crest, a red plume, and shadows that follow the existing head orientation.' Do not subsequently re-describe the retained face, body, or armor.\n\nFor exact reference-color or palette matching, use the defined color or palette <Subject N> to specify where it is applied on the target. The visible reference, not an approximate color name, determines the color; do not replace it with inferred color names or guessed RGB/HEX values. Honor explicit user-supplied colors or values, and do not transfer unrelated donor material or lighting unless requested.\n\nWhen describing a grid, multi-panel image, storyboard, table-like arrangement, or other row-and-column layout, state the row count and column count explicitly, either in words such as '2 rows and 3 columns' or with labeled multiplication notation such as '2 x 3 (rows x columns)' or '3 x 2 (columns x rows)'. Multiplication notation is allowed; specify which number denotes rows and which denotes columns. Do not use an unlabeled '2 x 3' or '2×3' alone. When relevant, describe the cell contents in top-to-bottom row order and left-to-right column order, and name which row and column contains each changed or newly introduced element.\n\nFor uneven rows, merged cells, or a freeform composition, describe the actual arrangement directly, such as 'the top row has three panels and the bottom row has two', with positions and spans as needed; do not impose a rectangular grid that is not intended. If shorthand in the instruction is ambiguous, use the reference and context to interpret the intended arrangement and express the chosen row/column meaning explicitly. Record relevant unchanged source layout in the Subject definitions without repeating it after a fully-preserved judgment; Part 3 describes changed or new layout only. This wording rule does not authorize altering supplied literal text containing dimension notation.\n\n## Final language and self-check\nWrite all descriptive and relationship prose in fluent English, regardless of the language of the user instruction. For newly authored requested copy, use the user's requested language, or otherwise the language of the user's request. Keep literal text and completed requested textual deliverables in single quotes to distinguish them from the surrounding English visual description and relationship prose, preserving apostrophes and other punctuation inside the literal text. Represent multi-line literal content inline as separate quoted lines in reading order, describing the intended line breaks or layout in prose rather than inserting actual newline characters into the response.\n\nUse objective, concrete visual language. Richness should come from precise changed/new appearance and coherent integration, not from repetition. Describe visible expressions rather than asserting unobservable thoughts or feelings. Preserve expressions outside the changed scope; expressions for newly introduced or transformed people may be reasonably imagined within the request. Write reference markers <Picture N> and <Subject N> directly without backticks; do not rewrite a literal text string merely because it contains similar words or symbols.\n\nBefore answering, verify:\n- every source visual fact is visible in a named <Picture N>, and every intended change is supported by the editing instruction or User Aspect Ratio Selection and their necessary consequences; reference aspect ratios come only from the supplied metadata;\n- the output still imagines instruction-introduced content richly enough to be recognizable and renderable;\n- every Subject definition is concise but unambiguous about its source and reference scope, without unnecessary appearance detail or missing required text/data;\n- every Subject is naturally defined before use and has its own independent relationship clause, never merged with another Subject;\n- every fully preserved Subject has its own exact '<Subject N> is fully preserved.' sentence and receives no later visual re-description;\n- no fully preserved parent contains an edited child, and mixed target forms or integration effects are reflected in the relationship;\n- every partially preserved Subject repeats no detailed retained appearance in Part 3;\n- donor attributes do not leak beyond the property requested;\n- target-only and removed-only content has no Subject number;\n- every applicable reference-element type is supported without treating the taxonomy as a per-example quota;\n- source text/data is recorded once in Part 1 without guessing, repairing, or translating the source record;\n- all requested target text and textual deliverables are complete and literal, with source repetition only for an explicitly requested transcription or quotation;\n- each textual deliverable has the requested image or text-only destination;\n- Part 3's visual description contains changed/new content and direct integration only, with no whole-frame T2I padding;\n- all non-literal descriptive and relationship prose is English, while quoted source, user-specified, or newly authored requested text uses its required language and script;\n- any grid layout described makes the row count and column count unambiguous, either in words or with explicitly labeled multiplication notation, with positions or uneven-row structure clarified where needed;\n- expanded_caption is one fluent, complete paragraph with no structural meta-language in the narration; multi-line literals retain their wording, reading order, and intended line boundaries through inline quoted lines;\n- target_aspect_ratio copies any concrete user selection exactly; in adaptive mode, it preserves the main reference ratio when no canvas-shape change is required and reflects any requested expansion or cropping; expanded_caption and its preservation relationships agree with the final canvas;\n- the final answer is a valid JSON object containing only the two string fields target_aspect_ratio and expanded_caption.\n\nOutput only the required two-field JSON object.\nUser input: {input text}\noutput_text:\n"
1256 }
1257 },
1258 {
1259 "id": 40,
1260 "type": "StringReplace",
1261 "pos": [
1262 1480,
1263 4020
1264 ],
1265 "size": [
1266 500,
1267 210
1268 ],
1269 "flags": {
1270 "collapsed": true
1271 },
1272 "order": 32,
1273 "mode": 0,
1274 "inputs": [
1275 {
1276 "localized_name": "字符串",
1277 "name": "string",
1278 "type": "STRING",
1279 "widget": {
1280 "name": "string"
1281 },
1282 "link": 58
1283 },
1284 {
1285 "localized_name": "查找",
1286 "name": "find",
1287 "type": "STRING",
1288 "widget": {
1289 "name": "find"
1290 },
1291 "link": null
1292 },
1293 {
1294 "localized_name": "替换",
1295 "name": "replace",
1296 "type": "STRING",
1297 "widget": {
1298 "name": "replace"
1299 },
1300 "link": 70
1301 }
1302 ],
1303 "outputs": [
1304 {
1305 "localized_name": "字符串",
1306 "name": "STRING",
1307 "type": "STRING",
1308 "links": [
1309 69
1310 ]
1311 }
1312 ],
1313 "title": "原生扩写输入拼接",
1314 "properties": {
1315 "Node name for S&R": "StringReplace"
1316 },
1317 "widgets_values": [
1318 "",
1319 "{input text}",
1320 ""
1321 ],
1322 "widgets_values_named": {
1323 "string": "",
1324 "find": "{input text}",
1325 "replace": ""
1326 }
1327 },
1328 {
1329 "id": 41,
1330 "type": "ComfySwitchNode",
1331 "pos": [
1332 1480,
1333 3420
1334 ],
1335 "size": [
1336 440,
1337 100
1338 ],
1339 "flags": [],
1340 "order": 40,
1341 "mode": 0,
1342 "inputs": [
1343 {
1344 "localized_name": "为假时",
1345 "name": "on_false",
1346 "shape": 7,
1347 "type": "STRING",
1348 "link": 62
1349 },
1350 {
1351 "localized_name": "为真时",
1352 "name": "on_true",
1353 "shape": 7,
1354 "type": "STRING",
1355 "link": 66
1356 },
1357 {
1358 "localized_name": "切换",
1359 "name": "switch",
1360 "type": "BOOLEAN",
1361 "widget": {
1362 "name": "switch"
1363 },
1364 "link": 92
1365 }
1366 ],
1367 "outputs": [
1368 {
1369 "localized_name": "输出",
1370 "name": "output",
1371 "type": "STRING",
1372 "links": [
1373 53
1374 ]
1375 }
1376 ],
1377 "title": "原生提示词切换",
1378 "properties": {
1379 "Node name for S&R": "ComfySwitchNode"
1380 },
1381 "widgets_values": [
1382 true
1383 ],
1384 "widgets_values_named": {
1385 "switch": true
1386 }
1387 },
1388 {
1389 "id": 42,
1390 "type": "VAEDecode",
1391 "pos": [
1392 1480,
1393 -150
1394 ],
1395 "size": [
1396 400,
1397 90
1398 ],
1399 "flags": [],
1400 "order": 45,
1401 "mode": 0,
1402 "inputs": [
1403 {
1404 "localized_name": "Latent",
1405 "name": "samples",
1406 "type": "LATENT",
1407 "link": 59
1408 },
1409 {
1410 "localized_name": "vae",
1411 "name": "vae",
1412 "type": "VAE",
1413 "link": 60
1414 }
1415 ],
1416 "outputs": [
1417 {
1418 "localized_name": "图像",
1419 "name": "IMAGE",
1420 "type": "IMAGE",
1421 "links": [
1422 61
1423 ]
1424 }
1425 ],
1426 "title": "原生VAE解码",
1427 "properties": {
1428 "Node name for S&R": "VAEDecode"
1429 }
1430 },
1431 {
1432 "id": 43,
1433 "type": "PrimitiveStringMultiline",
1434 "pos": [
1435 0,
1436 100
1437 ],
1438 "size": [
1439 1360,
1440 300
1441 ],
1442 "flags": [],
1443 "order": 8,
1444 "mode": 0,
1445 "inputs": [
1446 {
1447 "localized_name": "值",
1448 "name": "value",
1449 "type": "STRING",
1450 "widget": {
1451 "name": "value"
1452 },
1453 "link": null
1454 }
1455 ],
1456 "outputs": [
1457 {
1458 "localized_name": "字符串",
1459 "name": "STRING",
1460 "type": "STRING",
1461 "links": [
1462 62,
1463 70
1464 ]
1465 }
1466 ],
1467 "title": "编辑要求|用 <Picture 1>…<Picture 16> 指定参考",
1468 "properties": {
1469 "Node name for S&R": "PrimitiveStringMultiline"
1470 },
1471 "widgets_values": [
1472 "Replace the vintage yellow taxi in <Picture 1> with a pastel pink ice-cream truck with a scalloped awning and a glowing service window, sitting in the same side profile at the curb."
1473 ],
1474 "widgets_values_named": {
1475 "value": "Replace the vintage yellow taxi in <Picture 1> with a pastel pink ice-cream truck with a scalloped awning and a glowing service window, sitting in the same side profile at the curb."
1476 },
1477 "color": "#232",
1478 "bgcolor": "#353"
1479 },
1480 {
1481 "id": 44,
1482 "type": "JsonExtractString",
1483 "pos": [
1484 1480,
1485 3720
1486 ],
1487 "size": [
1488 440,
1489 180
1490 ],
1491 "flags": {
1492 "collapsed": true
1493 },
1494 "order": 39,
1495 "mode": 0,
1496 "inputs": [
1497 {
1498 "localized_name": "json_string",
1499 "name": "json_string",
1500 "type": "STRING",
1501 "widget": {
1502 "name": "json_string"
1503 },
1504 "link": 65
1505 },
1506 {
1507 "localized_name": "key",
1508 "name": "key",
1509 "type": "STRING",
1510 "widget": {
1511 "name": "key"
1512 },
1513 "link": null
1514 }
1515 ],
1516 "outputs": [
1517 {
1518 "localized_name": "字符串",
1519 "name": "STRING",
1520 "type": "STRING",
1521 "links": [
1522 66
1523 ]
1524 }
1525 ],
1526 "title": "原生提取扩写结果",
1527 "properties": {
1528 "Node name for S&R": "JsonExtractString"
1529 },
1530 "widgets_values": [
1531 "",
1532 "expanded_caption"
1533 ],
1534 "widgets_values_named": {
1535 "json_string": "",
1536 "key": "expanded_caption"
1537 }
1538 },
1539 {
1540 "id": 45,
1541 "type": "LoadImage",
1542 "pos": [
1543 0,
1544 660
1545 ],
1546 "size": [
1547 480,
1548 440
1549 ],
1550 "flags": [],
1551 "order": 9,
1552 "mode": 0,
1553 "inputs": [
1554 {
1555 "localized_name": "图像",
1556 "name": "image",
1557 "type": "COMBO",
1558 "widget": {
1559 "name": "image"
1560 },
1561 "link": null
1562 },
1563 {
1564 "localized_name": "选择文件上传",
1565 "name": "upload",
1566 "type": "IMAGEUPLOAD",
1567 "widget": {
1568 "name": "upload"
1569 },
1570 "link": null
1571 }
1572 ],
1573 "outputs": [
1574 {
1575 "localized_name": "图像",
1576 "name": "IMAGE",
1577 "type": "IMAGE",
1578 "links": [
1579 87
1580 ]
1581 },
1582 {
1583 "localized_name": "遮罩",
1584 "name": "MASK",
1585 "type": "MASK",
1586 "links": null
1587 }
1588 ],
1589 "title": "图片01|原生加载图像|Ctrl+B启用",
1590 "properties": {
1591 "Node name for S&R": "LoadImage"
1592 },
1593 "widgets_values": [
1594 "AIKSK_H3_Image_official_taxi_20261111.png",
1595 "image"
1596 ],
1597 "widgets_values_named": {
1598 "image": "AIKSK_H3_Image_official_taxi_20261111.png",
1599 "upload": "image"
1600 },
1601 "color": "#273b37",
1602 "bgcolor": "#344b43"
1603 },
1604 {
1605 "id": 48,
1606 "type": "ResizeImageMaskNode",
1607 "pos": [
1608 2500,
1609 1000
1610 ],
1611 "size": [
1612 440,
1613 170
1614 ],
1615 "flags": [],
1616 "order": 34,
1617 "mode": 0,
1618 "inputs": [
1619 {
1620 "localized_name": "输入",
1621 "name": "input",
1622 "type": "IMAGE,MASK",
1623 "link": 88
1624 },
1625 {
1626 "localized_name": "调整类型",
1627 "name": "resize_type",
1628 "type": "COMFY_DYNAMICCOMBO_V3",
1629 "widget": {
1630 "name": "resize_type"
1631 },
1632 "link": null
1633 },
1634 {
1635 "localized_name": "multiple",
1636 "name": "resize_type.multiple",
1637 "type": "INT",
1638 "widget": {
1639 "name": "resize_type.multiple"
1640 },
1641 "link": null
1642 },
1643 {
1644 "localized_name": "缩放方法",
1645 "name": "scale_method",
1646 "type": "COMBO",
1647 "widget": {
1648 "name": "scale_method"
1649 },
1650 "link": null
1651 }
1652 ],
1653 "outputs": [
1654 {
1655 "localized_name": "已调整大小",
1656 "name": "resized",
1657 "type": "IMAGE",
1658 "links": [
1659 77,
1660 80,
1661 81,
1662 95
1663 ]
1664 }
1665 ],
1666 "title": "原生缩放②|对齐32倍数(保留)",
1667 "properties": {
1668 "Node name for S&R": "ResizeImageMaskNode"
1669 },
1670 "widgets_values": [
1671 "scale to multiple",
1672 32,
1673 "area"
1674 ],
1675 "widgets_values_named": {
1676 "resize_type": "scale to multiple",
1677 "resize_type.multiple": 32,
1678 "scale_method": "area"
1679 }
1680 },
1681 {
1682 "id": 50,
1683 "type": "GetImageSize",
1684 "pos": [
1685 2500,
1686 1280
1687 ],
1688 "size": [
1689 440,
1690 120
1691 ],
1692 "flags": [],
1693 "order": 35,
1694 "mode": 0,
1695 "inputs": [
1696 {
1697 "localized_name": "图像",
1698 "name": "image",
1699 "type": "IMAGE",
1700 "link": 77
1701 }
1702 ],
1703 "outputs": [
1704 {
1705 "localized_name": "宽度",
1706 "name": "width",
1707 "type": "INT",
1708 "links": [
1709 78
1710 ]
1711 },
1712 {
1713 "localized_name": "高度",
1714 "name": "height",
1715 "type": "INT",
1716 "links": [
1717 79
1718 ]
1719 },
1720 {
1721 "localized_name": "批处理大小",
1722 "name": "batch_size",
1723 "type": "INT",
1724 "links": null
1725 }
1726 ],
1727 "title": "原生获取尺寸|跟随处理后的图片01",
1728 "properties": {
1729 "Node name for S&R": "GetImageSize"
1730 }
1731 },
1732 {
1733 "id": 55,
1734 "type": "MarkdownNote",
1735 "pos": [
1736 3000,
1737 3320
1738 ],
1739 "size": [
1740 660,
1741 220
1742 ],
1743 "flags": [],
1744 "order": 10,
1745 "mode": 0,
1746 "inputs": [],
1747 "outputs": [],
1748 "title": "官方扩写模型说明",
1749 "properties": [],
1750 "widgets_values": [
1751 "For MiniMax H3 Image's PE model it’s not tied to any specific model\n\n[MiniMax H3 Image PE](https://huggingface.co/MiniMaxAI/MiniMax-H3-Image-PE) repo is using `Qwen3.8-27B, FP4` and we are using Qwen3.5 9B in this workflow"
1752 ],
1753 "widgets_values_named": {
1754 "text": "For MiniMax H3 Image's PE model it’s not tied to any specific model\n\n[MiniMax H3 Image PE](https://huggingface.co/MiniMaxAI/MiniMax-H3-Image-PE) repo is using `Qwen3.8-27B, FP4` and we are using Qwen3.5 9B in this workflow"
1755 },
1756 "color": "#222",
1757 "bgcolor": "#000"
1758 },
1759 {
1760 "id": 56,
1761 "type": "ResizeImageMaskNode",
1762 "pos": [
1763 2500,
1764 720
1765 ],
1766 "size": [
1767 440,
1768 170
1769 ],
1770 "flags": [],
1771 "order": 33,
1772 "mode": 0,
1773 "inputs": [
1774 {
1775 "localized_name": "输入",
1776 "name": "input",
1777 "type": "IMAGE,MASK",
1778 "link": 87
1779 },
1780 {
1781 "localized_name": "调整类型",
1782 "name": "resize_type",
1783 "type": "COMFY_DYNAMICCOMBO_V3",
1784 "widget": {
1785 "name": "resize_type"
1786 },
1787 "link": null
1788 },
1789 {
1790 "localized_name": "megapixels",
1791 "name": "resize_type.megapixels",
1792 "type": "FLOAT",
1793 "widget": {
1794 "name": "resize_type.megapixels"
1795 },
1796 "link": null
1797 },
1798 {
1799 "localized_name": "缩放方法",
1800 "name": "scale_method",
1801 "type": "COMBO",
1802 "widget": {
1803 "name": "scale_method"
1804 },
1805 "link": null
1806 }
1807 ],
1808 "outputs": [
1809 {
1810 "localized_name": "已调整大小",
1811 "name": "resized",
1812 "type": "IMAGE",
1813 "links": [
1814 88
1815 ]
1816 }
1817 ],
1818 "title": "原生缩放①|第一张原图 → 4MP(保留)",
1819 "properties": {
1820 "Node name for S&R": "ResizeImageMaskNode"
1821 },
1822 "widgets_values": [
1823 "scale total pixels",
1824 4,
1825 "area"
1826 ],
1827 "widgets_values_named": {
1828 "resize_type": "scale total pixels",
1829 "resize_type.megapixels": 4,
1830 "scale_method": "area"
1831 }
1832 },
1833 {
1834 "id": 59,
1835 "type": "MarkdownNote",
1836 "pos": [
1837 3000,
1838 3680
1839 ],
1840 "size": [
1841 660,
1842 140
1843 ],
1844 "flags": [],
1845 "order": 11,
1846 "mode": 0,
1847 "inputs": [],
1848 "outputs": [],
1849 "title": "官方连接说明",
1850 "properties": [],
1851 "widgets_values": [
1852 "Hover on the `prompt output` or `right click -> show link` to restore the link"
1853 ],
1854 "widgets_values_named": {
1855 "text": "Hover on the `prompt output` or `right click -> show link` to restore the link"
1856 },
1857 "color": "#222",
1858 "bgcolor": "#000"
1859 },
1860 {
1861 "id": 61,
1862 "type": "PrimitiveBoolean",
1863 "pos": [
1864 1480,
1865 100
1866 ],
1867 "size": [
1868 400,
1869 90
1870 ],
1871 "flags": [],
1872 "order": 12,
1873 "mode": 0,
1874 "inputs": [
1875 {
1876 "localized_name": "值",
1877 "name": "value",
1878 "type": "BOOLEAN",
1879 "widget": {
1880 "name": "value"
1881 },
1882 "link": null
1883 }
1884 ],
1885 "outputs": [
1886 {
1887 "localized_name": "布尔",
1888 "name": "BOOLEAN",
1889 "type": "BOOLEAN",
1890 "links": [
1891 92
1892 ]
1893 }
1894 ],
1895 "title": "提示词扩写开关|关闭则直接使用原文",
1896 "properties": {
1897 "Node name for S&R": "PrimitiveBoolean"
1898 },
1899 "widgets_values": [
1900 true
1901 ],
1902 "widgets_values_named": {
1903 "value": true
1904 },
1905 "color": "#322",
1906 "bgcolor": "#533"
1907 },
1908 {
1909 "id": 62,
1910 "type": "ImageCompare",
1911 "pos": [
1912 3000,
1913 -850
1914 ],
1915 "size": [
1916 760,
1917 730
1918 ],
1919 "flags": [],
1920 "order": 47,
1921 "mode": 0,
1922 "inputs": [
1923 {
1924 "localized_name": "image_a",
1925 "name": "image_a",
1926 "shape": 7,
1927 "type": "IMAGE",
1928 "link": 95
1929 },
1930 {
1931 "localized_name": "image_b",
1932 "name": "image_b",
1933 "shape": 7,
1934 "type": "IMAGE",
1935 "link": 94
1936 },
1937 {
1938 "localized_name": "compare_view",
1939 "name": "compare_view",
1940 "type": "IMAGECOMPARE",
1941 "widget": {
1942 "name": "compare_view"
1943 },
1944 "link": null
1945 }
1946 ],
1947 "outputs": [],
1948 "title": "原生对比|图片01与输出",
1949 "properties": {
1950 "Node name for S&R": "ImageCompare"
1951 },
1952 "widgets_values": [],
1953 "widgets_values_named": []
1954 },
1955 {
1956 "id": 102,
1957 "type": "LoadImage",
1958 "pos": [
1959 600,
1960 660
1961 ],
1962 "size": [
1963 480,
1964 440
1965 ],
1966 "flags": [],
1967 "order": 13,
1968 "mode": 4,
1969 "inputs": [
1970 {
1971 "localized_name": "图像",
1972 "name": "image",
1973 "type": "COMBO",
1974 "widget": {
1975 "name": "image"
1976 },
1977 "link": null
1978 },
1979 {
1980 "localized_name": "选择文件上传",
1981 "name": "upload",
1982 "type": "IMAGEUPLOAD",
1983 "widget": {
1984 "name": "upload"
1985 },
1986 "link": null
1987 }
1988 ],
1989 "outputs": [
1990 {
1991 "localized_name": "图像",
1992 "name": "IMAGE",
1993 "type": "IMAGE",
1994 "links": [
1995 96,
1996 97
1997 ]
1998 },
1999 {
2000 "localized_name": "遮罩",
2001 "name": "MASK",
2002 "type": "MASK",
2003 "links": null
2004 }
2005 ],
2006 "title": "图片02|原生加载图像|Ctrl+B启用",
2007 "properties": {
2008 "Node name for S&R": "LoadImage"
2009 },
2010 "widgets_values": [
2011 "",
2012 "image"
2013 ],
2014 "widgets_values_named": {
2015 "image": "",
2016 "upload": "image"
2017 },
2018 "color": "#263643",
2019 "bgcolor": "#344958"
2020 },
2021 {
2022 "id": 103,
2023 "type": "LoadImage",
2024 "pos": [
2025 1200,
2026 660
2027 ],
2028 "size": [
2029 480,
2030 440
2031 ],
2032 "flags": [],
2033 "order": 14,
2034 "mode": 4,
2035 "inputs": [
2036 {
2037 "localized_name": "图像",
2038 "name": "image",
2039 "type": "COMBO",
2040 "widget": {
2041 "name": "image"
2042 },
2043 "link": null
2044 },
2045 {
2046 "localized_name": "选择文件上传",
2047 "name": "upload",
2048 "type": "IMAGEUPLOAD",
2049 "widget": {
2050 "name": "upload"
2051 },
2052 "link": null
2053 }
2054 ],
2055 "outputs": [
2056 {
2057 "localized_name": "图像",
2058 "name": "IMAGE",
2059 "type": "IMAGE",
2060 "links": [
2061 99,
2062 100
2063 ]
2064 },
2065 {
2066 "localized_name": "遮罩",
2067 "name": "MASK",
2068 "type": "MASK",
2069 "links": null
2070 }
2071 ],
2072 "title": "图片03|原生加载图像|Ctrl+B启用",
2073 "properties": {
2074 "Node name for S&R": "LoadImage"
2075 },
2076 "widgets_values": [
2077 "",
2078 "image"
2079 ],
2080 "widgets_values_named": {
2081 "image": "",
2082 "upload": "image"
2083 },
2084 "color": "#263643",
2085 "bgcolor": "#344958"
2086 },
2087 {
2088 "id": 104,
2089 "type": "LoadImage",
2090 "pos": [
2091 1800,
2092 660
2093 ],
2094 "size": [
2095 480,
2096 440
2097 ],
2098 "flags": [],
2099 "order": 15,
2100 "mode": 4,
2101 "inputs": [
2102 {
2103 "localized_name": "图像",
2104 "name": "image",
2105 "type": "COMBO",
2106 "widget": {
2107 "name": "image"
2108 },
2109 "link": null
2110 },
2111 {
2112 "localized_name": "选择文件上传",
2113 "name": "upload",
2114 "type": "IMAGEUPLOAD",
2115 "widget": {
2116 "name": "upload"
2117 },
2118 "link": null
2119 }
2120 ],
2121 "outputs": [
2122 {
2123 "localized_name": "图像",
2124 "name": "IMAGE",
2125 "type": "IMAGE",
2126 "links": [
2127 102,
2128 103
2129 ]
2130 },
2131 {
2132 "localized_name": "遮罩",
2133 "name": "MASK",
2134 "type": "MASK",
2135 "links": null
2136 }
2137 ],
2138 "title": "图片04|原生加载图像|Ctrl+B启用",
2139 "properties": {
2140 "Node name for S&R": "LoadImage"
2141 },
2142 "widgets_values": [
2143 "",
2144 "image"
2145 ],
2146 "widgets_values_named": {
2147 "image": "",
2148 "upload": "image"
2149 },
2150 "color": "#263643",
2151 "bgcolor": "#344958"
2152 },
2153 {
2154 "id": 105,
2155 "type": "LoadImage",
2156 "pos": [
2157 0,
2158 1250
2159 ],
2160 "size": [
2161 480,
2162 440
2163 ],
2164 "flags": [],
2165 "order": 16,
2166 "mode": 4,
2167 "inputs": [
2168 {
2169 "localized_name": "图像",
2170 "name": "image",
2171 "type": "COMBO",
2172 "widget": {
2173 "name": "image"
2174 },
2175 "link": null
2176 },
2177 {
2178 "localized_name": "选择文件上传",
2179 "name": "upload",
2180 "type": "IMAGEUPLOAD",
2181 "widget": {
2182 "name": "upload"
2183 },
2184 "link": null
2185 }
2186 ],
2187 "outputs": [
2188 {
2189 "localized_name": "图像",
2190 "name": "IMAGE",
2191 "type": "IMAGE",
2192 "links": [
2193 105,
2194 106
2195 ]
2196 },
2197 {
2198 "localized_name": "遮罩",
2199 "name": "MASK",
2200 "type": "MASK",
2201 "links": null
2202 }
2203 ],
2204 "title": "图片05|原生加载图像|Ctrl+B启用",
2205 "properties": {
2206 "Node name for S&R": "LoadImage"
2207 },
2208 "widgets_values": [
2209 "",
2210 "image"
2211 ],
2212 "widgets_values_named": {
2213 "image": "",
2214 "upload": "image"
2215 },
2216 "color": "#263643",
2217 "bgcolor": "#344958"
2218 },
2219 {
2220 "id": 106,
2221 "type": "LoadImage",
2222 "pos": [
2223 600,
2224 1250
2225 ],
2226 "size": [
2227 480,
2228 440
2229 ],
2230 "flags": [],
2231 "order": 17,
2232 "mode": 4,
2233 "inputs": [
2234 {
2235 "localized_name": "图像",
2236 "name": "image",
2237 "type": "COMBO",
2238 "widget": {
2239 "name": "image"
2240 },
2241 "link": null
2242 },
2243 {
2244 "localized_name": "选择文件上传",
2245 "name": "upload",
2246 "type": "IMAGEUPLOAD",
2247 "widget": {
2248 "name": "upload"
2249 },
2250 "link": null
2251 }
2252 ],
2253 "outputs": [
2254 {
2255 "localized_name": "图像",
2256 "name": "IMAGE",
2257 "type": "IMAGE",
2258 "links": [
2259 108,
2260 109
2261 ]
2262 },
2263 {
2264 "localized_name": "遮罩",
2265 "name": "MASK",
2266 "type": "MASK",
2267 "links": null
2268 }
2269 ],
2270 "title": "图片06|原生加载图像|Ctrl+B启用",
2271 "properties": {
2272 "Node name for S&R": "LoadImage"
2273 },
2274 "widgets_values": [
2275 "",
2276 "image"
2277 ],
2278 "widgets_values_named": {
2279 "image": "",
2280 "upload": "image"
2281 },
2282 "color": "#263643",
2283 "bgcolor": "#344958"
2284 },
2285 {
2286 "id": 107,
2287 "type": "LoadImage",
2288 "pos": [
2289 1200,
2290 1250
2291 ],
2292 "size": [
2293 480,
2294 440
2295 ],
2296 "flags": [],
2297 "order": 18,
2298 "mode": 4,
2299 "inputs": [
2300 {
2301 "localized_name": "图像",
2302 "name": "image",
2303 "type": "COMBO",
2304 "widget": {
2305 "name": "image"
2306 },
2307 "link": null
2308 },
2309 {
2310 "localized_name": "选择文件上传",
2311 "name": "upload",
2312 "type": "IMAGEUPLOAD",
2313 "widget": {
2314 "name": "upload"
2315 },
2316 "link": null
2317 }
2318 ],
2319 "outputs": [
2320 {
2321 "localized_name": "图像",
2322 "name": "IMAGE",
2323 "type": "IMAGE",
2324 "links": [
2325 111,
2326 112
2327 ]
2328 },
2329 {
2330 "localized_name": "遮罩",
2331 "name": "MASK",
2332 "type": "MASK",
2333 "links": null
2334 }
2335 ],
2336 "title": "图片07|原生加载图像|Ctrl+B启用",
2337 "properties": {
2338 "Node name for S&R": "LoadImage"
2339 },
2340 "widgets_values": [
2341 "",
2342 "image"
2343 ],
2344 "widgets_values_named": {
2345 "image": "",
2346 "upload": "image"
2347 },
2348 "color": "#263643",
2349 "bgcolor": "#344958"
2350 },
2351 {
2352 "id": 108,
2353 "type": "LoadImage",
2354 "pos": [
2355 1800,
2356 1250
2357 ],
2358 "size": [
2359 480,
2360 440
2361 ],
2362 "flags": [],
2363 "order": 19,
2364 "mode": 4,
2365 "inputs": [
2366 {
2367 "localized_name": "图像",
2368 "name": "image",
2369 "type": "COMBO",
2370 "widget": {
2371 "name": "image"
2372 },
2373 "link": null
2374 },
2375 {
2376 "localized_name": "选择文件上传",
2377 "name": "upload",
2378 "type": "IMAGEUPLOAD",
2379 "widget": {
2380 "name": "upload"
2381 },
2382 "link": null
2383 }
2384 ],
2385 "outputs": [
2386 {
2387 "localized_name": "图像",
2388 "name": "IMAGE",
2389 "type": "IMAGE",
2390 "links": [
2391 114,
2392 115
2393 ]
2394 },
2395 {
2396 "localized_name": "遮罩",
2397 "name": "MASK",
2398 "type": "MASK",
2399 "links": null
2400 }
2401 ],
2402 "title": "图片08|原生加载图像|Ctrl+B启用",
2403 "properties": {
2404 "Node name for S&R": "LoadImage"
2405 },
2406 "widgets_values": [
2407 "",
2408 "image"
2409 ],
2410 "widgets_values_named": {
2411 "image": "",
2412 "upload": "image"
2413 },
2414 "color": "#263643",
2415 "bgcolor": "#344958"
2416 },
2417 {
2418 "id": 109,
2419 "type": "LoadImage",
2420 "pos": [
2421 0,
2422 1840
2423 ],
2424 "size": [
2425 480,
2426 440
2427 ],
2428 "flags": [],
2429 "order": 20,
2430 "mode": 4,
2431 "inputs": [
2432 {
2433 "localized_name": "图像",
2434 "name": "image",
2435 "type": "COMBO",
2436 "widget": {
2437 "name": "image"
2438 },
2439 "link": null
2440 },
2441 {
2442 "localized_name": "选择文件上传",
2443 "name": "upload",
2444 "type": "IMAGEUPLOAD",
2445 "widget": {
2446 "name": "upload"
2447 },
2448 "link": null
2449 }
2450 ],
2451 "outputs": [
2452 {
2453 "localized_name": "图像",
2454 "name": "IMAGE",
2455 "type": "IMAGE",
2456 "links": [
2457 117,
2458 118
2459 ]
2460 },
2461 {
2462 "localized_name": "遮罩",
2463 "name": "MASK",
2464 "type": "MASK",
2465 "links": null
2466 }
2467 ],
2468 "title": "图片09|原生加载图像|Ctrl+B启用",
2469 "properties": {
2470 "Node name for S&R": "LoadImage"
2471 },
2472 "widgets_values": [
2473 "",
2474 "image"
2475 ],
2476 "widgets_values_named": {
2477 "image": "",
2478 "upload": "image"
2479 },
2480 "color": "#263643",
2481 "bgcolor": "#344958"
2482 },
2483 {
2484 "id": 110,
2485 "type": "LoadImage",
2486 "pos": [
2487 600,
2488 1840
2489 ],
2490 "size": [
2491 480,
2492 440
2493 ],
2494 "flags": [],
2495 "order": 21,
2496 "mode": 4,
2497 "inputs": [
2498 {
2499 "localized_name": "图像",
2500 "name": "image",
2501 "type": "COMBO",
2502 "widget": {
2503 "name": "image"
2504 },
2505 "link": null
2506 },
2507 {
2508 "localized_name": "选择文件上传",
2509 "name": "upload",
2510 "type": "IMAGEUPLOAD",
2511 "widget": {
2512 "name": "upload"
2513 },
2514 "link": null
2515 }
2516 ],
2517 "outputs": [
2518 {
2519 "localized_name": "图像",
2520 "name": "IMAGE",
2521 "type": "IMAGE",
2522 "links": [
2523 120,
2524 121
2525 ]
2526 },
2527 {
2528 "localized_name": "遮罩",
2529 "name": "MASK",
2530 "type": "MASK",
2531 "links": null
2532 }
2533 ],
2534 "title": "图片10|原生加载图像|Ctrl+B启用",
2535 "properties": {
2536 "Node name for S&R": "LoadImage"
2537 },
2538 "widgets_values": [
2539 "",
2540 "image"
2541 ],
2542 "widgets_values_named": {
2543 "image": "",
2544 "upload": "image"
2545 },
2546 "color": "#263643",
2547 "bgcolor": "#344958"
2548 },
2549 {
2550 "id": 111,
2551 "type": "LoadImage",
2552 "pos": [
2553 1200,
2554 1840
2555 ],
2556 "size": [
2557 480,
2558 440
2559 ],
2560 "flags": [],
2561 "order": 22,
2562 "mode": 4,
2563 "inputs": [
2564 {
2565 "localized_name": "图像",
2566 "name": "image",
2567 "type": "COMBO",
2568 "widget": {
2569 "name": "image"
2570 },
2571 "link": null
2572 },
2573 {
2574 "localized_name": "选择文件上传",
2575 "name": "upload",
2576 "type": "IMAGEUPLOAD",
2577 "widget": {
2578 "name": "upload"
2579 },
2580 "link": null
2581 }
2582 ],
2583 "outputs": [
2584 {
2585 "localized_name": "图像",
2586 "name": "IMAGE",
2587 "type": "IMAGE",
2588 "links": [
2589 123,
2590 124
2591 ]
2592 },
2593 {
2594 "localized_name": "遮罩",
2595 "name": "MASK",
2596 "type": "MASK",
2597 "links": null
2598 }
2599 ],
2600 "title": "图片11|原生加载图像|Ctrl+B启用",
2601 "properties": {
2602 "Node name for S&R": "LoadImage"
2603 },
2604 "widgets_values": [
2605 "",
2606 "image"
2607 ],
2608 "widgets_values_named": {
2609 "image": "",
2610 "upload": "image"
2611 },
2612 "color": "#263643",
2613 "bgcolor": "#344958"
2614 },
2615 {
2616 "id": 112,
2617 "type": "LoadImage",
2618 "pos": [
2619 1800,
2620 1840
2621 ],
2622 "size": [
2623 480,
2624 440
2625 ],
2626 "flags": [],
2627 "order": 23,
2628 "mode": 4,
2629 "inputs": [
2630 {
2631 "localized_name": "图像",
2632 "name": "image",
2633 "type": "COMBO",
2634 "widget": {
2635 "name": "image"
2636 },
2637 "link": null
2638 },
2639 {
2640 "localized_name": "选择文件上传",
2641 "name": "upload",
2642 "type": "IMAGEUPLOAD",
2643 "widget": {
2644 "name": "upload"
2645 },
2646 "link": null
2647 }
2648 ],
2649 "outputs": [
2650 {
2651 "localized_name": "图像",
2652 "name": "IMAGE",
2653 "type": "IMAGE",
2654 "links": [
2655 126,
2656 127
2657 ]
2658 },
2659 {
2660 "localized_name": "遮罩",
2661 "name": "MASK",
2662 "type": "MASK",
2663 "links": null
2664 }
2665 ],
2666 "title": "图片12|原生加载图像|Ctrl+B启用",
2667 "properties": {
2668 "Node name for S&R": "LoadImage"
2669 },
2670 "widgets_values": [
2671 "",
2672 "image"
2673 ],
2674 "widgets_values_named": {
2675 "image": "",
2676 "upload": "image"
2677 },
2678 "color": "#263643",
2679 "bgcolor": "#344958"
2680 },
2681 {
2682 "id": 113,
2683 "type": "LoadImage",
2684 "pos": [
2685 0,
2686 2430
2687 ],
2688 "size": [
2689 480,
2690 440
2691 ],
2692 "flags": [],
2693 "order": 24,
2694 "mode": 4,
2695 "inputs": [
2696 {
2697 "localized_name": "图像",
2698 "name": "image",
2699 "type": "COMBO",
2700 "widget": {
2701 "name": "image"
2702 },
2703 "link": null
2704 },
2705 {
2706 "localized_name": "选择文件上传",
2707 "name": "upload",
2708 "type": "IMAGEUPLOAD",
2709 "widget": {
2710 "name": "upload"
2711 },
2712 "link": null
2713 }
2714 ],
2715 "outputs": [
2716 {
2717 "localized_name": "图像",
2718 "name": "IMAGE",
2719 "type": "IMAGE",
2720 "links": [
2721 129,
2722 130
2723 ]
2724 },
2725 {
2726 "localized_name": "遮罩",
2727 "name": "MASK",
2728 "type": "MASK",
2729 "links": null
2730 }
2731 ],
2732 "title": "图片13|原生加载图像|Ctrl+B启用",
2733 "properties": {
2734 "Node name for S&R": "LoadImage"
2735 },
2736 "widgets_values": [
2737 "",
2738 "image"
2739 ],
2740 "widgets_values_named": {
2741 "image": "",
2742 "upload": "image"
2743 },
2744 "color": "#263643",
2745 "bgcolor": "#344958"
2746 },
2747 {
2748 "id": 114,
2749 "type": "LoadImage",
2750 "pos": [
2751 600,
2752 2430
2753 ],
2754 "size": [
2755 480,
2756 440
2757 ],
2758 "flags": [],
2759 "order": 25,
2760 "mode": 4,
2761 "inputs": [
2762 {
2763 "localized_name": "图像",
2764 "name": "image",
2765 "type": "COMBO",
2766 "widget": {
2767 "name": "image"
2768 },
2769 "link": null
2770 },
2771 {
2772 "localized_name": "选择文件上传",
2773 "name": "upload",
2774 "type": "IMAGEUPLOAD",
2775 "widget": {
2776 "name": "upload"
2777 },
2778 "link": null
2779 }
2780 ],
2781 "outputs": [
2782 {
2783 "localized_name": "图像",
2784 "name": "IMAGE",
2785 "type": "IMAGE",
2786 "links": [
2787 132,
2788 133
2789 ]
2790 },
2791 {
2792 "localized_name": "遮罩",
2793 "name": "MASK",
2794 "type": "MASK",
2795 "links": null
2796 }
2797 ],
2798 "title": "图片14|原生加载图像|Ctrl+B启用",
2799 "properties": {
2800 "Node name for S&R": "LoadImage"
2801 },
2802 "widgets_values": [
2803 "",
2804 "image"
2805 ],
2806 "widgets_values_named": {
2807 "image": "",
2808 "upload": "image"
2809 },
2810 "color": "#263643",
2811 "bgcolor": "#344958"
2812 },
2813 {
2814 "id": 115,
2815 "type": "LoadImage",
2816 "pos": [
2817 1200,
2818 2430
2819 ],
2820 "size": [
2821 480,
2822 440
2823 ],
2824 "flags": [],
2825 "order": 26,
2826 "mode": 4,
2827 "inputs": [
2828 {
2829 "localized_name": "图像",
2830 "name": "image",
2831 "type": "COMBO",
2832 "widget": {
2833 "name": "image"
2834 },
2835 "link": null
2836 },
2837 {
2838 "localized_name": "选择文件上传",
2839 "name": "upload",
2840 "type": "IMAGEUPLOAD",
2841 "widget": {
2842 "name": "upload"
2843 },
2844 "link": null
2845 }
2846 ],
2847 "outputs": [
2848 {
2849 "localized_name": "图像",
2850 "name": "IMAGE",
2851 "type": "IMAGE",
2852 "links": [
2853 135,
2854 136
2855 ]
2856 },
2857 {
2858 "localized_name": "遮罩",
2859 "name": "MASK",
2860 "type": "MASK",
2861 "links": null
2862 }
2863 ],
2864 "title": "图片15|原生加载图像|Ctrl+B启用",
2865 "properties": {
2866 "Node name for S&R": "LoadImage"
2867 },
2868 "widgets_values": [
2869 "",
2870 "image"
2871 ],
2872 "widgets_values_named": {
2873 "image": "",
2874 "upload": "image"
2875 },
2876 "color": "#263643",
2877 "bgcolor": "#344958"
2878 },
2879 {
2880 "id": 116,
2881 "type": "LoadImage",
2882 "pos": [
2883 1800,
2884 2430
2885 ],
2886 "size": [
2887 480,
2888 440
2889 ],
2890 "flags": [],
2891 "order": 27,
2892 "mode": 4,
2893 "inputs": [
2894 {
2895 "localized_name": "图像",
2896 "name": "image",
2897 "type": "COMBO",
2898 "widget": {
2899 "name": "image"
2900 },
2901 "link": null
2902 },
2903 {
2904 "localized_name": "选择文件上传",
2905 "name": "upload",
2906 "type": "IMAGEUPLOAD",
2907 "widget": {
2908 "name": "upload"
2909 },
2910 "link": null
2911 }
2912 ],
2913 "outputs": [
2914 {
2915 "localized_name": "图像",
2916 "name": "IMAGE",
2917 "type": "IMAGE",
2918 "links": [
2919 138,
2920 139
2921 ]
2922 },
2923 {
2924 "localized_name": "遮罩",
2925 "name": "MASK",
2926 "type": "MASK",
2927 "links": null
2928 }
2929 ],
2930 "title": "图片16|原生加载图像|Ctrl+B启用",
2931 "properties": {
2932 "Node name for S&R": "LoadImage"
2933 },
2934 "widgets_values": [
2935 "",
2936 "image"
2937 ],
2938 "widgets_values_named": {
2939 "image": "",
2940 "upload": "image"
2941 },
2942 "color": "#263643",
2943 "bgcolor": "#344958"
2944 },
2945 {
2946 "id": 120,
2947 "type": "MarkdownNote",
2948 "pos": [
2949 0,
2950 -1190
2951 ],
2952 "size": [
2953 3760,
2954 200
2955 ],
2956 "flags": [],
2957 "order": 28,
2958 "mode": 0,
2959 "inputs": [],
2960 "outputs": [],
2961 "title": "AIKSK|MiniMax H3 Image 图像编辑",
2962 "properties": [],
2963 "widgets_values": [
2964 "# 图像编辑|16个原生加载图像\n\n16个图片框默认全部Bypass(旁路)。按需上传并 Ctrl+B 恢复启用,从图片01开始连续使用;运行前至少启用图片01。\n保留原有4MP缩放、32倍数对齐和外置输出尺寸选择。无自定义加载图像依赖。"
2965 ],
2966 "widgets_values_named": {
2967 "text": "# 图像编辑|16个原生加载图像\n\n16个图片框默认全部Bypass(旁路)。按需上传并 Ctrl+B 恢复启用,从图片01开始连续使用;运行前至少启用图片01。\n保留原有4MP缩放、32倍数对齐和外置输出尺寸选择。无自定义加载图像依赖。"
2968 },
2969 "color": "#432",
2970 "bgcolor": "#653"
2971 },
2972 {
2973 "id": 121,
2974 "type": "ResolutionSelector",
2975 "pos": [
2976 2040,
2977 290
2978 ],
2979 "size": [
2980 520,
2981 180
2982 ],
2983 "flags": [],
2984 "order": 29,
2985 "mode": 0,
2986 "showAdvanced": true,
2987 "inputs": [
2988 {
2989 "localized_name": "宽高比",
2990 "name": "aspect_ratio",
2991 "type": "COMBO",
2992 "widget": {
2993 "name": "aspect_ratio"
2994 },
2995 "link": null
2996 },
2997 {
2998 "localized_name": "百万像素",
2999 "name": "megapixels",
3000 "type": "FLOAT",
3001 "widget": {
3002 "name": "megapixels"
3003 },
3004 "link": null
3005 },
3006 {
3007 "localized_name": "倍数",
3008 "name": "multiple",
3009 "type": "INT",
3010 "widget": {
3011 "name": "multiple"
3012 },
3013 "link": null
3014 },
3015 {
3016 "localized_name": "预览",
3017 "name": "preview",
3018 "shape": 7,
3019 "type": "RESOLUTION_PREVIEW",
3020 "widget": {
3021 "name": "preview"
3022 },
3023 "link": null
3024 }
3025 ],
3026 "outputs": [
3027 {
3028 "localized_name": "宽度",
3029 "name": "width",
3030 "type": "INT",
3031 "links": [
3032 141
3033 ]
3034 },
3035 {
3036 "localized_name": "高度",
3037 "name": "height",
3038 "type": "INT",
3039 "links": [
3040 144
3041 ]
3042 }
3043 ],
3044 "title": "外置输出尺寸|比例 + 像素量(4 / 8 / 16)",
3045 "properties": {
3046 "Node name for S&R": "ResolutionSelector"
3047 },
3048 "widgets_values": [
3049 "16:9 (Widescreen)",
3050 4,
3051 32
3052 ],
3053 "widgets_values_named": {
3054 "aspect_ratio": "16:9 (Widescreen)",
3055 "megapixels": 4,
3056 "multiple": 32
3057 },
3058 "color": "#243b35",
3059 "bgcolor": "#344f45"
3060 },
3061 {
3062 "id": 122,
3063 "type": "PrimitiveBoolean",
3064 "pos": [
3065 2040,
3066 130
3067 ],
3068 "size": [
3069 520,
3070 100
3071 ],
3072 "flags": [],
3073 "order": 30,
3074 "mode": 0,
3075 "inputs": [
3076 {
3077 "localized_name": "值",
3078 "name": "value",
3079 "type": "BOOLEAN",
3080 "widget": {
3081 "name": "value"
3082 },
3083 "link": null
3084 }
3085 ],
3086 "outputs": [
3087 {
3088 "localized_name": "布尔",
3089 "name": "BOOLEAN",
3090 "type": "BOOLEAN",
3091 "links": [
3092 142,
3093 145
3094 ]
3095 }
3096 ],
3097 "title": "启用外置尺寸|关闭=跟随图片1,开启=下方设定",
3098 "properties": {
3099 "Node name for S&R": "PrimitiveBoolean"
3100 },
3101 "widgets_values": [
3102 false
3103 ],
3104 "widgets_values_named": {
3105 "value": false
3106 },
3107 "color": "#243b35",
3108 "bgcolor": "#344f45"
3109 },
3110 {
3111 "id": 123,
3112 "type": "ComfySwitchNode",
3113 "pos": [
3114 2700,
3115 130
3116 ],
3117 "size": [
3118 460,
3119 130
3120 ],
3121 "flags": [],
3122 "order": 37,
3123 "mode": 0,
3124 "inputs": [
3125 {
3126 "localized_name": "为假时",
3127 "name": "on_false",
3128 "shape": 7,
3129 "type": "INT",
3130 "link": 78
3131 },
3132 {
3133 "localized_name": "为真时",
3134 "name": "on_true",
3135 "shape": 7,
3136 "type": "INT",
3137 "link": 141
3138 },
3139 {
3140 "localized_name": "切换",
3141 "name": "switch",
3142 "type": "BOOLEAN",
3143 "widget": {
3144 "name": "switch"
3145 },
3146 "link": 142
3147 }
3148 ],
3149 "outputs": [
3150 {
3151 "localized_name": "输出",
3152 "name": "output",
3153 "type": "INT",
3154 "links": [
3155 143
3156 ]
3157 }
3158 ],
3159 "title": "宽度路由|自动切换",
3160 "properties": {
3161 "Node name for S&R": "ComfySwitchNode"
3162 },
3163 "widgets_values": [
3164 false
3165 ],
3166 "widgets_values_named": {
3167 "switch": false
3168 }
3169 },
3170 {
3171 "id": 124,
3172 "type": "ComfySwitchNode",
3173 "pos": [
3174 3220,
3175 130
3176 ],
3177 "size": [
3178 460,
3179 130
3180 ],
3181 "flags": [],
3182 "order": 38,
3183 "mode": 0,
3184 "inputs": [
3185 {
3186 "localized_name": "为假时",
3187 "name": "on_false",
3188 "shape": 7,
3189 "type": "INT",
3190 "link": 79
3191 },
3192 {
3193 "localized_name": "为真时",
3194 "name": "on_true",
3195 "shape": 7,
3196 "type": "INT",
3197 "link": 144
3198 },
3199 {
3200 "localized_name": "切换",
3201 "name": "switch",
3202 "type": "BOOLEAN",
3203 "widget": {
3204 "name": "switch"
3205 },
3206 "link": 145
3207 }
3208 ],
3209 "outputs": [
3210 {
3211 "localized_name": "输出",
3212 "name": "output",
3213 "type": "INT",
3214 "links": [
3215 146
3216 ]
3217 }
3218 ],
3219 "title": "高度路由|自动切换",
3220 "properties": {
3221 "Node name for S&R": "ComfySwitchNode"
3222 },
3223 "widgets_values": [
3224 false
3225 ],
3226 "widgets_values_named": {
3227 "switch": false
3228 }
3229 },
3230 {
3231 "id": 125,
3232 "type": "MarkdownNote",
3233 "pos": [
3234 2700,
3235 320
3236 ],
3237 "size": [
3238 980,
3239 180
3240 ],
3241 "flags": [],
3242 "order": 31,
3243 "mode": 0,
3244 "inputs": [],
3245 "outputs": [],
3246 "title": "使用方法|只操作开关与尺寸选择器",
3247 "properties": [],
3248 "widgets_values": [
3249 "## 输出尺寸\n\n关闭“启用外置尺寸”:跟随图片1。\n开启:在左侧下方选择比例和4/8/16MP。\n\n16个图片框全部为原生加载图像,默认全部Bypass。使用时上传图片并按Ctrl+B恢复启用,从图片01开始连续使用。\n\n运行前至少启用图片01。原有缩放和外置尺寸切换保留,无自定义加载图像依赖。\n\n结构检查通过;GPU出图未测试。"
3250 ],
3251 "widgets_values_named": {
3252 "text": "## 输出尺寸\n\n关闭“启用外置尺寸”:跟随图片1。\n开启:在左侧下方选择比例和4/8/16MP。\n\n16个图片框全部为原生加载图像,默认全部Bypass。使用时上传图片并按Ctrl+B恢复启用,从图片01开始连续使用。\n\n运行前至少启用图片01。原有缩放和外置尺寸切换保留,无自定义加载图像依赖。\n\n结构检查通过;GPU出图未测试。"
3253 },
3254 "color": "#432",
3255 "bgcolor": "#653"
3256 }
3257 ],
3258 "links": [
3259 [
3260 52,
3261 32,
3262 0,
3263 33,
3264 0,
3265 "CONDITIONING"
3266 ],
3267 [
3268 53,
3269 41,
3270 0,
3271 37,
3272 0,
3273 "*"
3274 ],
3275 [
3276 54,
3277 29,
3278 0,
3279 38,
3280 0,
3281 "MODEL"
3282 ],
3283 [
3284 55,
3285 32,
3286 0,
3287 38,
3288 1,
3289 "CONDITIONING"
3290 ],
3291 [
3292 56,
3293 33,
3294 0,
3295 38,
3296 2,
3297 "CONDITIONING"
3298 ],
3299 [
3300 57,
3301 32,
3302 1,
3303 38,
3304 3,
3305 "LATENT"
3306 ],
3307 [
3308 58,
3309 39,
3310 0,
3311 40,
3312 0,
3313 "STRING"
3314 ],
3315 [
3316 59,
3317 38,
3318 0,
3319 42,
3320 0,
3321 "LATENT"
3322 ],
3323 [
3324 60,
3325 31,
3326 0,
3327 42,
3328 1,
3329 "VAE"
3330 ],
3331 [
3332 61,
3333 42,
3334 0,
3335 9,
3336 0,
3337 "IMAGE"
3338 ],
3339 [
3340 62,
3341 43,
3342 0,
3343 41,
3344 0,
3345 "STRING"
3346 ],
3347 [
3348 65,
3349 34,
3350 0,
3351 44,
3352 0,
3353 "STRING"
3354 ],
3355 [
3356 66,
3357 44,
3358 0,
3359 41,
3360 1,
3361 "STRING"
3362 ],
3363 [
3364 77,
3365 48,
3366 0,
3367 50,
3368 0,
3369 "IMAGE"
3370 ],
3371 [
3372 92,
3373 61,
3374 0,
3375 41,
3376 2,
3377 "BOOLEAN"
3378 ],
3379 [
3380 94,
3381 9,
3382 0,
3383 62,
3384 1,
3385 "IMAGE"
3386 ],
3387 [
3388 95,
3389 48,
3390 0,
3391 62,
3392 0,
3393 "IMAGE"
3394 ],
3395 [
3396 78,
3397 50,
3398 0,
3399 123,
3400 0,
3401 "INT"
3402 ],
3403 [
3404 79,
3405 50,
3406 1,
3407 124,
3408 0,
3409 "INT"
3410 ],
3411 [
3412 141,
3413 121,
3414 0,
3415 123,
3416 1,
3417 "INT"
3418 ],
3419 [
3420 142,
3421 122,
3422 0,
3423 123,
3424 2,
3425 "BOOLEAN"
3426 ],
3427 [
3428 144,
3429 121,
3430 1,
3431 124,
3432 1,
3433 "INT"
3434 ],
3435 [
3436 145,
3437 122,
3438 0,
3439 124,
3440 2,
3441 "BOOLEAN"
3442 ],
3443 [
3444 70,
3445 43,
3446 0,
3447 40,
3448 2,
3449 "STRING"
3450 ],
3451 [
3452 49,
3453 30,
3454 0,
3455 32,
3456 0,
3457 "CLIP"
3458 ],
3459 [
3460 50,
3461 31,
3462 0,
3463 32,
3464 1,
3465 "VAE"
3466 ],
3467 [
3468 89,
3469 37,
3470 0,
3471 32,
3472 3,
3473 "STRING"
3474 ],
3475 [
3476 143,
3477 123,
3478 0,
3479 32,
3480 4,
3481 "INT"
3482 ],
3483 [
3484 146,
3485 124,
3486 0,
3487 32,
3488 5,
3489 "INT"
3490 ],
3491 [
3492 80,
3493 48,
3494 0,
3495 32,
3496 6,
3497 "IMAGE"
3498 ],
3499 [
3500 105,
3501 105,
3502 0,
3503 32,
3504 7,
3505 "IMAGE"
3506 ],
3507 [
3508 108,
3509 106,
3510 0,
3511 32,
3512 8,
3513 "IMAGE"
3514 ],
3515 [
3516 111,
3517 107,
3518 0,
3519 32,
3520 9,
3521 "IMAGE"
3522 ],
3523 [
3524 114,
3525 108,
3526 0,
3527 32,
3528 10,
3529 "IMAGE"
3530 ],
3531 [
3532 117,
3533 109,
3534 0,
3535 32,
3536 11,
3537 "IMAGE"
3538 ],
3539 [
3540 120,
3541 110,
3542 0,
3543 32,
3544 12,
3545 "IMAGE"
3546 ],
3547 [
3548 123,
3549 111,
3550 0,
3551 32,
3552 13,
3553 "IMAGE"
3554 ],
3555 [
3556 126,
3557 112,
3558 0,
3559 32,
3560 14,
3561 "IMAGE"
3562 ],
3563 [
3564 129,
3565 113,
3566 0,
3567 32,
3568 15,
3569 "IMAGE"
3570 ],
3571 [
3572 132,
3573 114,
3574 0,
3575 32,
3576 16,
3577 "IMAGE"
3578 ],
3579 [
3580 135,
3581 115,
3582 0,
3583 32,
3584 17,
3585 "IMAGE"
3586 ],
3587 [
3588 138,
3589 116,
3590 0,
3591 32,
3592 18,
3593 "IMAGE"
3594 ],
3595 [
3596 96,
3597 102,
3598 0,
3599 32,
3600 21,
3601 "IMAGE"
3602 ],
3603 [
3604 99,
3605 103,
3606 0,
3607 32,
3608 22,
3609 "IMAGE"
3610 ],
3611 [
3612 102,
3613 104,
3614 0,
3615 32,
3616 23,
3617 "IMAGE"
3618 ],
3619 [
3620 67,
3621 35,
3622 0,
3623 34,
3624 0,
3625 "CLIP"
3626 ],
3627 [
3628 81,
3629 48,
3630 0,
3631 34,
3632 1,
3633 "IMAGE"
3634 ],
3635 [
3636 68,
3637 36,
3638 0,
3639 34,
3640 4,
3641 "STRING"
3642 ],
3643 [
3644 69,
3645 40,
3646 0,
3647 34,
3648 5,
3649 "STRING"
3650 ],
3651 [
3652 106,
3653 105,
3654 0,
3655 34,
3656 18,
3657 "IMAGE"
3658 ],
3659 [
3660 109,
3661 106,
3662 0,
3663 34,
3664 19,
3665 "IMAGE"
3666 ],
3667 [
3668 112,
3669 107,
3670 0,
3671 34,
3672 20,
3673 "IMAGE"
3674 ],
3675 [
3676 115,
3677 108,
3678 0,
3679 34,
3680 21,
3681 "IMAGE"
3682 ],
3683 [
3684 118,
3685 109,
3686 0,
3687 34,
3688 22,
3689 "IMAGE"
3690 ],
3691 [
3692 121,
3693 110,
3694 0,
3695 34,
3696 23,
3697 "IMAGE"
3698 ],
3699 [
3700 124,
3701 111,
3702 0,
3703 34,
3704 24,
3705 "IMAGE"
3706 ],
3707 [
3708 127,
3709 112,
3710 0,
3711 34,
3712 25,
3713 "IMAGE"
3714 ],
3715 [
3716 130,
3717 113,
3718 0,
3719 34,
3720 26,
3721 "IMAGE"
3722 ],
3723 [
3724 133,
3725 114,
3726 0,
3727 34,
3728 27,
3729 "IMAGE"
3730 ],
3731 [
3732 136,
3733 115,
3734 0,
3735 34,
3736 28,
3737 "IMAGE"
3738 ],
3739 [
3740 139,
3741 116,
3742 0,
3743 34,
3744 29,
3745 "IMAGE"
3746 ],
3747 [
3748 97,
3749 102,
3750 0,
3751 34,
3752 32,
3753 "IMAGE"
3754 ],
3755 [
3756 100,
3757 103,
3758 0,
3759 34,
3760 33,
3761 "IMAGE"
3762 ],
3763 [
3764 103,
3765 104,
3766 0,
3767 34,
3768 34,
3769 "IMAGE"
3770 ],
3771 [
3772 88,
3773 56,
3774 0,
3775 48,
3776 0,
3777 "IMAGE"
3778 ],
3779 [
3780 87,
3781 45,
3782 0,
3783 56,
3784 0,
3785 "IMAGE"
3786 ]
3787 ],
3788 "groups": [
3789 {
3790 "id": 1,
3791 "title": "01|模型与原生生成链路",
3792 "bounding": [
3793 -60,
3794 -940,
3795 1960,
3796 1020
3797 ],
3798 "color": "#3f789e",
3799 "flags": []
3800 },
3801 {
3802 "id": 2,
3803 "title": "02|编辑要求与扩写开关",
3804 "bounding": [
3805 -60,
3806 80,
3807 1960,
3808 390
3809 ],
3810 "color": "#3d7460",
3811 "flags": []
3812 },
3813 {
3814 "id": 3,
3815 "title": "03|16个原生加载图像(默认全部Bypass / Ctrl+B启用)",
3816 "bounding": [
3817 -60,
3818 540,
3819 2420,
3820 2370
3821 ],
3822 "color": "#3d7460",
3823 "flags": []
3824 },
3825 {
3826 "id": 4,
3827 "title": "04|保留原生缩放与尺寸",
3828 "bounding": [
3829 2440,
3830 540,
3831 560,
3832 2040
3833 ],
3834 "color": "#8b7260",
3835 "flags": []
3836 },
3837 {
3838 "id": 5,
3839 "title": "05|官方提示词扩写链路",
3840 "bounding": [
3841 -60,
3842 3190,
3843 2940,
3844 1180
3845 ],
3846 "color": "#555577",
3847 "flags": []
3848 },
3849 {
3850 "id": 6,
3851 "title": "06|结果与对比",
3852 "bounding": [
3853 1980,
3854 -940,
3855 1840,
3856 900
3857 ],
3858 "color": "#3f789e",
3859 "flags": []
3860 },
3861 {
3862 "id": 7,
3863 "title": "新增|外置输出尺寸(原生节点)",
3864 "bounding": [
3865 1980,
3866 50,
3867 1840,
3868 460
3869 ],
3870 "color": "#3d7460",
3871 "flags": []
3872 }
3873 ],
3874 "config": [],
3875 "extra": {
3876 "ds": {
3877 "scale": 0.14641000000000007,
3878 "offset": [
3879 3840.0673378279703,
3880 1633.3924659051586
3881 ]
3882 },
3883 "frontendVersion": "1.58.5",
3884 "AIKSK": {
3885 "name": "H3_Image_Edit_16Refs_NativeLoadImage",
3886 "version": "2026-11-11.4-native-bypass",
3887 "basis": "Comfy-Org/workflow_templates",
3888 "gpu_validation": "NOT_TESTED",
3889 "reference_limit": 16,
3890 "output_size_modes": [
3891 "follow_image_1",
3892 "external_ratio_and_megapixels"
3893 ],
3894 "custom_image_loader_dependency": false,
3895 "default_image_loader_mode": "Bypass (4), all 16 slots"
3896 },
3897 "linkPresentation": {
3898 "89": {
3899 "hidden": true,
3900 "label": "prompt output"
3901 }
3902 }
3903 },
3904 "version": 0.4
3905}