工作流 /
图像 /
GPT Image 2.6 Sunburst：图像编辑
GPT Image 2.6 Sunburst：图像编辑
GPT Image 2.6 Sunburst: Image Edit
只改一处，其余保持原样：机位、构图、主体位置和光线都不动，只替换你点名的那一块。GPT Image 2.6 Sunburst 在细纹理和小字号上更稳，海报和高密度场景多改几轮也不容易崩。
图像
4 个节点 1 个输入 283 次使用 ComfyUI ≥ 0.44.1 ComfyUI 9 KB
节点构成
MarkdownNote
× 1
LoadImage
× 1
SaveImageAdvanced
× 1
OpenAIGPTImageNodeV2
× 1
官方标注的模型
GPT Image 2.6
工作流 JSON
和上面「下载工作流 JSON」是同一份文件。
1{
2 "id": "7a06a6de-067d-4d41-b7bf-7d6add20ec8c",
3 "revision": 0,
4 "last_node_id": 394,
5 "last_link_id": 575,
6 "nodes": [
7 {
8 "id": 375,
9 "type": "MarkdownNote",
10 "pos": [
11 -1720,
12 7010
13 ],
14 "size": [
15 360,
16 360
17 ],
18 "flags": [],
19 "order": 0,
20 "mode": 0,
21 "inputs": [],
22 "outputs": [],
23 "properties": [],
24 "widgets_values": [
25 "## Input Asset\n\n- [api_openai_gpt_image_26_sunburst_image_edit_input.png](https://raw.githubusercontent.com/Comfy-Org/workflow_templates/refs/heads/main/input/api_openai_gpt_image_26_sunburst_image_edit_input.png)\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"
26 ],
27 "widgets_values_named": {
28 "text": "## Input Asset\n\n- [api_openai_gpt_image_26_sunburst_image_edit_input.png](https://raw.githubusercontent.com/Comfy-Org/workflow_templates/refs/heads/main/input/api_openai_gpt_image_26_sunburst_image_edit_input.png)\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"
29 },
30 "color": "#222",
31 "bgcolor": "#000"
32 },
33 {
34 "id": 376,
35 "type": "LoadImage",
36 "pos": [
37 -1320,
38 7010
39 ],
40 "size": [
41 390,
42 450
43 ],
44 "flags": [],
45 "order": 1,
46 "mode": 0,
47 "inputs": [],
48 "outputs": [
49 {
50 "name": "IMAGE",
51 "type": "IMAGE",
52 "links": [
53 573
54 ]
55 },
56 {
57 "name": "MASK",
58 "type": "MASK",
59 "links": null
60 }
61 ],
62 "properties": {
63 "Node name for S&R": "LoadImage"
64 },
65 "widgets_values": [
66 "api_openai_gpt_image_26_sunburst_image_edit_input.png",
67 "image"
68 ],
69 "widgets_values_named": {
70 "image": "api_openai_gpt_image_26_sunburst_image_edit_input.png",
71 "upload": "image"
72 }
73 },
74 {
75 "id": 391,
76 "type": "SaveImageAdvanced",
77 "pos": [
78 -430,
79 7010
80 ],
81 "size": [
82 550,
83 434
84 ],
85 "flags": [],
86 "order": 3,
87 "mode": 0,
88 "showAdvanced": true,
89 "inputs": [
90 {
91 "name": "images",
92 "type": "IMAGE",
93 "link": 572
94 }
95 ],
96 "outputs": [
97 {
98 "name": "images",
99 "type": "IMAGE",
100 "links": null
101 }
102 ],
103 "properties": [],
104 "widgets_values": [
105 "api_openai_gpt_image_26_sunburst_image_edit",
106 "png",
107 "8-bit",
108 "sRGB"
109 ],
110 "widgets_values_named": {
111 "filename_prefix": "api_openai_gpt_image_26_sunburst_image_edit",
112 "format": "png",
113 "format.bit_depth": "8-bit",
114 "format.input_color_space": "sRGB"
115 }
116 },
117 {
118 "id": 392,
119 "type": "OpenAIGPTImageNodeV2",
120 "pos": [
121 -880,
122 7010
123 ],
124 "size": [
125 400,
126 368
127 ],
128 "flags": [],
129 "order": 2,
130 "mode": 0,
131 "inputs": [
132 {
133 "name": "model.images.image_1",
134 "shape": 7,
135 "type": "IMAGE",
136 "link": 573
137 },
138 {
139 "name": "model.images.image_2",
140 "shape": 7,
141 "type": "IMAGE",
142 "link": null
143 },
144 {
145 "name": "model.mask",
146 "shape": 7,
147 "type": "MASK",
148 "link": null
149 }
150 ],
151 "outputs": [
152 {
153 "name": "IMAGE",
154 "type": "IMAGE",
155 "links": [
156 572
157 ]
158 }
159 ],
160 "properties": {
161 "Node name for S&R": "OpenAIGPTImageNodeV2"
162 },
163 "widgets_values": [
164 "Edit this color photo into a photoreal Magritte-like surreal photograph. CRITICAL continuity locks: keep the Volkswagen Type 2 bus in the EXACT same position, scale, orientation, three-quarter angle, hover height, and circular-hole piercing relationship as the reference — do NOT move, rotate, enlarge, shrink, or restage the vehicle. Keep the same camera height and framing. Photoreal fine-art photography only — NO illustration, NO comic, NO CGI cartoon look.\n\nReplace the concrete walls with a monumental vertical WALL MADE OF REAL CLOUDS: dense sculpted cumulus that still reads as a flat architectural plane meeting at a right angle, with a PERFECT circular aperture cut through the cloud-mass — the bus must protrude through that cloud-hole exactly as it did through concrete (front hanging in open air, rear still inside the cloud wall). The courtyard floor becomes a soft continuous cloud deck / mist floor beneath the hovering bus. Beyond the cloud walls: authentic open sky and distant real atmospheric clouds under soft overcast daylight.\n\nRelight the cream VW for open-sky conditions: cool skylight fill, soft diffuse shadows, faint sky-color bounce on body panels, natural specular on chrome and glass matching the cloudy sky — no studio look.\n\nIntegrate LARGE environmental typography formed in the cloud wall itself or as soft shadow-lettering across the cloud face (partially occluded by the bus), readable exact text: \"GPT IMAGE 2.6\" and \"SUNBURST\". No other slogans, no people.\n\nThe vertical plane must literally be clouds the bus is stuck in — not a cliff above clouds.",
165 "gpt-image-2.6-sunburst",
166 "auto",
167 1024,
168 1024,
169 "auto",
170 "low",
171 "high",
172 1,
173 553435342,
174 "randomize"
175 ],
176 "widgets_values_named": {
177 "prompt": "Edit this color photo into a photoreal Magritte-like surreal photograph. CRITICAL continuity locks: keep the Volkswagen Type 2 bus in the EXACT same position, scale, orientation, three-quarter angle, hover height, and circular-hole piercing relationship as the reference — do NOT move, rotate, enlarge, shrink, or restage the vehicle. Keep the same camera height and framing. Photoreal fine-art photography only — NO illustration, NO comic, NO CGI cartoon look.\n\nReplace the concrete walls with a monumental vertical WALL MADE OF REAL CLOUDS: dense sculpted cumulus that still reads as a flat architectural plane meeting at a right angle, with a PERFECT circular aperture cut through the cloud-mass — the bus must protrude through that cloud-hole exactly as it did through concrete (front hanging in open air, rear still inside the cloud wall). The courtyard floor becomes a soft continuous cloud deck / mist floor beneath the hovering bus. Beyond the cloud walls: authentic open sky and distant real atmospheric clouds under soft overcast daylight.\n\nRelight the cream VW for open-sky conditions: cool skylight fill, soft diffuse shadows, faint sky-color bounce on body panels, natural specular on chrome and glass matching the cloudy sky — no studio look.\n\nIntegrate LARGE environmental typography formed in the cloud wall itself or as soft shadow-lettering across the cloud face (partially occluded by the bus), readable exact text: \"GPT IMAGE 2.6\" and \"SUNBURST\". No other slogans, no people.\n\nThe vertical plane must literally be clouds the bus is stuck in — not a cliff above clouds.",
178 "model": "gpt-image-2.6-sunburst",
179 "model.size": "auto",
180 "model.custom_width": 1024,
181 "model.custom_height": 1024,
182 "model.background": "auto",
183 "model.quality": "low",
184 "model.complexity": "high",
185 "n": 1,
186 "seed": 553435342,
187 "control_after_generate": "randomize"
188 },
189 "color": "#432",
190 "bgcolor": "#653"
191 }
192 ],
193 "links": [
194 [
195 572,
196 392,
197 0,
198 391,
199 0,
200 "IMAGE"
201 ],
202 [
203 573,
204 376,
205 0,
206 392,
207 0,
208 "IMAGE"
209 ]
210 ],
211 "groups": [],
212 "config": [],
213 "extra": {
214 "ds": {
215 "scale": 0.3561419643870965,
216 "offset": [
217 3934.981964397097,
218 -5615.695748001764
219 ]
220 },
221 "frontendVersion": "1.58.5",
222 "VHS_latentpreview": false,
223 "VHS_latentpreviewrate": 0,
224 "VHS_MetadataImage": true,
225 "VHS_KeepIntermediate": true
226 },
227 "version": 0.4
228}