工作流 /
图像工具 /
图像扩图: Bria
图像扩图: Bria
Image Outpainting: Bria
使用Bria的扩图API将图像扩展到原始边界之外，生成与原始光照、透视和纹理相匹配的新周围内容，同时保持源图像不变。从预设宽高比、自定义尺寸或手动画布放置中选择，可选文本提示词引导扩展方向，留空时自动生成提示词。非常适合扩展构图以用于打印布局、创建现有照片的全景版本，以及将艺术作品适配到新格式或社交媒体尺寸。
图像工具
4 个节点 1 个输入 42 次使用 ComfyUI ≥ 0.33.0 ComfyUI 6 KB
节点构成
LoadImage
× 1
SaveImageAdvanced
× 1
MarkdownNote
× 1
BriaExpandImage
× 1
官方标注的模型
BRIA
工作流 JSON
和上面「下载工作流 JSON」是同一份文件。
1{
2 "id": "fb5b92b3-fd02-4753-8d96-4f8848898ee8",
3 "revision": 0,
4 "last_node_id": 16,
5 "last_link_id": 32,
6 "nodes": [
7 {
8 "id": 2,
9 "type": "LoadImage",
10 "pos": [
11 -2900,
12 4860
13 ],
14 "size": [
15 360,
16 470
17 ],
18 "flags": [],
19 "order": 0,
20 "mode": 0,
21 "inputs": [],
22 "outputs": [
23 {
24 "name": "IMAGE",
25 "type": "IMAGE",
26 "links": [
27 31
28 ]
29 },
30 {
31 "name": "MASK",
32 "type": "MASK",
33 "links": []
34 }
35 ],
36 "properties": {
37 "Node name for S&R": "LoadImage",
38 "image": "clipspace-painted-masked-1787233871783.png [input]"
39 },
40 "widgets_values": [
41 "red_sports_car_night.png",
42 "image"
43 ]
44 },
45 {
46 "id": 4,
47 "type": "SaveImageAdvanced",
48 "pos": [
49 -2040,
50 4860
51 ],
52 "size": [
53 550,
54 377.875
55 ],
56 "flags": [],
57 "order": 3,
58 "mode": 0,
59 "inputs": [
60 {
61 "name": "images",
62 "type": "IMAGE",
63 "link": 32
64 }
65 ],
66 "outputs": [
67 {
68 "name": "images",
69 "type": "IMAGE",
70 "links": []
71 }
72 ],
73 "properties": [],
74 "widgets_values": [
75 "Bria_Expand_Image",
76 "png",
77 "8-bit",
78 "sRGB"
79 ]
80 },
81 {
82 "id": 13,
83 "type": "MarkdownNote",
84 "pos": [
85 -3380,
86 4860
87 ],
88 "size": [
89 440,
90 899.640625
91 ],
92 "flags": [],
93 "order": 1,
94 "mode": 0,
95 "inputs": [],
96 "outputs": [],
97 "title": "Note: Bria Expand Image",
98 "properties": [],
99 "widgets_values": [
100 "# Bria Expand Image Template\n\nUse this template to **expand an image beyond its original borders** with Bria's Expand API (outpainting). The original image stays untouched; the surrounding area is generated to match its lighting, perspective and texture.\n\n## Input Assets\n\n- [red_sports_car_night.png](https://raw.githubusercontent.com/Comfy-Org/workflow_templates/refs/heads/main/input/red_sports_car_night.png)\n\n## How to use\n\n1. **Load an image**: Upload your image in `LoadImage` and connect it to `BriaExpandImage`.\n2. **Choose an expansion mode** in the `expand_mode` dropdown:\n - **Preset aspect ratio**: Pick one of `1:1`, `2:3`, `3:2`, `3:4`, `4:3`, `4:5`, `5:4`, `9:16`, `16:9`. The image is centered and the edges are expanded to fit that ratio.\n - **Custom ratio**: Pick `custom_ratio` and set `ratio_width` / `ratio_height` (e.g. 21 and 9 for 21:9). Bria accepts width-to-height ratios between 0.5 and 3.0, so anything taller than 1:2 needs the manual mode.\n - **Manual**: Pick `manual` and place the original image on a canvas with `canvas_width` / `canvas_height` (max 5000 × 5000), `image_width` / `image_height`, and `image_x` / `image_y` (the image's top-left corner; can be negative to crop).\n3. **Prompt (optional)**: Two behaviors:\n - **Leave it empty**: Bria automatically generates a prompt from the input image and expands to match its scene and style. This works well for simple expansions.\n - **Write your own**: Guide the expansion direction explicitly, e.g. \"the street continuing to the right\", \"add a streetlamp\", \"no other cars\". English only; special characters are not supported.\n4. **Optional**: `negative_prompt` to avoid unwanted content, `seed` to reproduce a result, `moderation` to enable content moderation.\n5. **Save**: connect the image output to `SaveImage`.\n\n## Notes\n\n- The `prompt` output on the node shows the prompt actually used: your input, or the one auto-generated by Bria when the field is empty.\n- If the target aspect ratio equals the image's current ratio (e.g. a 1:1 image set to `1:1`), no changes are made and the image is returned as-is.\n- Canvas area is limited to 5,000 × 5,000 pixels. Keep the main subject above **15% of the total canvas area** for best results.\n- `preserve_alpha` keeps transparency (the API inherits alpha from the nearest original pixels). Expansion is rejected (422) if any image edge has fully transparent pixels."
101 ],
102 "color": "#222",
103 "bgcolor": "#000"
104 },
105 {
106 "id": 16,
107 "type": "BriaExpandImage",
108 "pos": [
109 -2490,
110 4860
111 ],
112 "size": [
113 400,
114 361.890625
115 ],
116 "flags": [],
117 "order": 2,
118 "mode": 0,
119 "inputs": [
120 {
121 "name": "image",
122 "type": "IMAGE",
123 "link": 31
124 }
125 ],
126 "outputs": [
127 {
128 "name": "IMAGE",
129 "type": "IMAGE",
130 "links": [
131 32
132 ]
133 },
134 {
135 "name": "prompt",
136 "type": "STRING",
137 "links": null
138 }
139 ],
140 "properties": {
141 "Node name for S&R": "BriaExpandImage"
142 },
143 "widgets_values": [
144 "16:9",
145 "The night city street continuing seamlessly beyond the original frame, matching the existing scene: blurred colorful city lights in blue, orange and yellow, wet asphalt road with reflections of the neon lights, modern building silhouettes in the background, the same motion-blurred aesthetic and cool blue-grey color grading, empty street with no other cars, realistic night photography style",
146 "",
147 1270036642,
148 "randomize",
149 "false"
150 ],
151 "color": "#432",
152 "bgcolor": "#653"
153 }
154 ],
155 "links": [
156 [
157 31,
158 2,
159 0,
160 16,
161 0,
162 "IMAGE"
163 ],
164 [
165 32,
166 16,
167 0,
168 4,
169 0,
170 "IMAGE"
171 ]
172 ],
173 "groups": [],
174 "config": [],
175 "extra": {
176 "ds": {
177 "scale": 0.8180279679683498,
178 "offset": [
179 3683.018988118275,
180 -4404.051980771512
181 ]
182 },
183 "frontendVersion": "1.48.7",
184 "VHS_latentpreview": false,
185 "VHS_latentpreviewrate": 0,
186 "VHS_MetadataImage": true,
187 "VHS_KeepIntermediate": true
188 },
189 "version": 0.4
190}