什么是 inpainting?
inpainting 是 “只重绘图像的一部分” 的手法。
但是,其实看里面可以分为以下 2 种模式。
- 类型 A: 只对掩膜(Mask)部分做 image2image
- 类型 B: 一边看周围的信息,一边自然地填补掩膜部分
一般来说,这俩不加区别,但因此经常看到混乱的初学者。
先作为别的东西分开考虑吧。
掩膜的制作方法和掩膜编辑的详情,请参照别页的 掩膜操作、使用 AI 生成掩膜。
类型 A: 只对掩膜部分 image2image
是只对掩盖的部分,以和通常的 image2image 同样的调子重绘的方法。
适合想要稍微改变面部表情、改变画风、稍微修正细微部分的时候。
工作流
在这个工作流中,使用 SetLatentNoiseMask 节点指定“在哪里添加噪声”。

{
"id": "8b9f7796-0873-4025-be3c-0f997f67f866",
"revision": 0,
"last_node_id": 17,
"last_link_id": 20,
"nodes": [
{
"id": 10,
"type": "VAELoader",
"pos": [
150.35849892963756,
608.4685752753562
],
"size": [
231.48760330578511,
58
],
"flags": {},
"order": 0,
"mode": 0,
"inputs": [],
"outputs": [
{
"name": "VAE",
"type": "VAE",
"links": [
10,
13
]
}
],
"properties": {
"cnr_id": "comfy-core",
"ver": "0.3.76",
"Node name for S&R": "VAELoader"
},
"widgets_values": [
"vae-ft-mse-840000-ema-pruned.safetensors"
]
},
{
"id": 12,
"type": "VAEEncode",
"pos": [
427.11516235234546,
674.510187570697
],
"size": [
140,
46
],
"flags": {},
"order": 5,
"mode": 0,
"inputs": [
{
"name": "pixels",
"type": "IMAGE",
"link": 11
},
{
"name": "vae",
"type": "VAE",
"link": 13
}
],
"outputs": [
{
"name": "LATENT",
"type": "LATENT",
"links": [
14
]
}
],
"properties": {
"cnr_id": "comfy-core",
"ver": "0.3.76",
"Node name for S&R": "VAEEncode"
},
"color": "#322",
"bgcolor": "#533"
},
{
"id": 7,
"type": "CLIPTextEncode",
"pos": [
416.1970166015625,
392.37848510742185
],
"size": [
410.75801513671877,
158.82607910156253
],
"flags": {},
"order": 4,
"mode": 0,
"inputs": [
{
"name": "clip",
"type": "CLIP",
"link": 5
}
],
"outputs": [
{
"name": "CONDITIONING",
"type": "CONDITIONING",
"slot_index": 0,
"links": [
6
]
}
],
"properties": {
"cnr_id": "comfy-core",
"ver": "0.3.33",
"Node name for S&R": "CLIPTextEncode"
},
"widgets_values": [
"text, watermark, worst quality"
]
},
{
"id": 9,
"type": "SaveImage",
"pos": [
1451,
189
],
"size": [
503.2876035004722,
576.239673217884
],
"flags": {},
"order": 9,
"mode": 0,
"inputs": [
{
"name": "images",
"type": "IMAGE",
"link": 9
}
],
"outputs": [],
"properties": {
"cnr_id": "comfy-core",
"ver": "0.3.33"
},
"widgets_values": [
"ComfyUI"
]
},
{
"id": 8,
"type": "VAEDecode",
"pos": [
1209,
188
],
"size": [
203.39999999999986,
46
],
"flags": {},
"order": 8,
"mode": 0,
"inputs": [
{
"name": "samples",
"type": "LATENT",
"link": 7
},
{
"name": "vae",
"type": "VAE",
"link": 10
}
],
"outputs": [
{
"name": "IMAGE",
"type": "IMAGE",
"slot_index": 0,
"links": [
9
]
}
],
"properties": {
"cnr_id": "comfy-core",
"ver": "0.3.33",
"Node name for S&R": "VAEDecode"
},
"widgets_values": []
},
{
"id": 4,
"type": "CheckpointLoaderSimple",
"pos": [
39.900463636363625,
349.9095500000008
],
"size": [
315,
98
],
"flags": {},
"order": 1,
"mode": 0,
"inputs": [],
"outputs": [
{
"name": "MODEL",
"type": "MODEL",
"slot_index": 0,
"links": [
1
]
},
{
"name": "CLIP",
"type": "CLIP",
"slot_index": 1,
"links": [
3,
5
]
},
{
"name": "VAE",
"type": "VAE",
"slot_index": 2,
"links": []
}
],
"properties": {
"cnr_id": "comfy-core",
"ver": "0.3.33",
"Node name for S&R": "CheckpointLoaderSimple"
},
"widgets_values": [
"v1-5-pruned-emaonly-fp16.safetensors"
]
},
{
"id": 6,
"type": "CLIPTextEncode",
"pos": [
415,
186
],
"size": [
411.95503173828126,
151.0030493164063
],
"flags": {},
"order": 3,
"mode": 0,
"inputs": [
{
"name": "clip",
"type": "CLIP",
"link": 3
}
],
"outputs": [
{
"name": "CONDITIONING",
"type": "CONDITIONING",
"slot_index": 0,
"links": [
4
]
}
],
"properties": {
"cnr_id": "comfy-core",
"ver": "0.3.33",
"Node name for S&R": "CLIPTextEncode"
},
"widgets_values": [
"high quality, RAW photo of a woman with red hair, Punch perm"
]
},
{
"id": 11,
"type": "LoadImage",
"pos": [
39.85257865587715,
722.7009982393679
],
"size": [
341.9935235795455,
473.93255454545454
],
"flags": {},
"order": 2,
"mode": 0,
"inputs": [],
"outputs": [
{
"name": "IMAGE",
"type": "IMAGE",
"links": [
11
]
},
{
"name": "MASK",
"type": "MASK",
"links": [
15
]
}
],
"properties": {
"cnr_id": "comfy-core",
"ver": "0.3.76",
"Node name for S&R": "LoadImage",
"image": "clipspace/clipspace-painted-masked-1765071052914.png [input]"
},
"widgets_values": [
"clipspace/clipspace-painted-masked-1765071052914.png [input]",
"image"
]
},
{
"id": 3,
"type": "KSampler",
"pos": [
863,
186
],
"size": [
315,
262
],
"flags": {},
"order": 7,
"mode": 0,
"inputs": [
{
"name": "model",
"type": "MODEL",
"link": 1
},
{
"name": "positive",
"type": "CONDITIONING",
"link": 4
},
{
"name": "negative",
"type": "CONDITIONING",
"link": 6
},
{
"name": "latent_image",
"type": "LATENT",
"link": 16
}
],
"outputs": [
{
"name": "LATENT",
"type": "LATENT",
"slot_index": 0,
"links": [
7
]
}
],
"properties": {
"cnr_id": "comfy-core",
"ver": "0.3.33",
"Node name for S&R": "KSampler"
},
"widgets_values": [
45678,
"fixed",
20,
8,
"euler",
"normal",
0.8
]
},
{
"id": 13,
"type": "SetLatentNoiseMask",
"pos": [
607.2212780548256,
721.4635076308024
],
"size": [
208.02038352272726,
46
],
"flags": {},
"order": 6,
"mode": 0,
"inputs": [
{
"name": "samples",
"type": "LATENT",
"link": 14
},
{
"name": "mask",
"type": "MASK",
"link": 15
}
],
"outputs": [
{
"name": "LATENT",
"type": "LATENT",
"links": [
16
]
}
],
"properties": {
"cnr_id": "comfy-core",
"ver": "0.3.76",
"Node name for S&R": "SetLatentNoiseMask"
},
"color": "#232",
"bgcolor": "#353"
}
],
"links": [
[
1,
4,
0,
3,
0,
"MODEL"
],
[
3,
4,
1,
6,
0,
"CLIP"
],
[
4,
6,
0,
3,
1,
"CONDITIONING"
],
[
5,
4,
1,
7,
0,
"CLIP"
],
[
6,
7,
0,
3,
2,
"CONDITIONING"
],
[
7,
3,
0,
8,
0,
"LATENT"
],
[
9,
8,
0,
9,
0,
"IMAGE"
],
[
10,
10,
0,
8,
1,
"VAE"
],
[
11,
11,
0,
12,
0,
"IMAGE"
],
[
13,
10,
0,
12,
1,
"VAE"
],
[
14,
12,
0,
13,
0,
"LATENT"
],
[
15,
11,
1,
13,
1,
"MASK"
],
[
16,
13,
0,
3,
3,
"LATENT"
]
],
"groups": [],
"config": {},
"extra": {
"ds": {
"scale": 0.620921323059155,
"offset": [
332.32361134412287,
153.96599000000006
]
},
"frontendVersion": "1.34.6",
"VHS_latentpreview": false,
"VHS_latentpreviewrate": 0,
"VHS_MetadataImage": true,
"VHS_KeepIntermediate": true
},
"version": 0.4
}
- 基础是 image2image 的工作流。
- 🟥 在
VAE Encode节点将原图转换为 latent - 🟩 在
Set Latent Noise Mask节点组合 latent 和掩膜
这种手法的问题点
试着将上面的工作流的 denoise 设为 1.00 看看。

哇哦,生成了恐怖图像 (;・∀・)
这种手法,终究是“只把掩膜部分作为画布的 image2image”。
提高 denoise 的话,在掩膜部分会变成几乎 接近 text2image 的举动。
因为在提示词里写了“红色的爆炸头女性”,所以和原图无关,新画出了女性呢。
有没有一边看整体氛围,一边画被掩盖部分的方法呢?
类型 B: 一边看周围一边填补掩膜
在看了图像整体的基础上,“为了和周围自然连接而重绘掩膜部分”的类型。
刚才,只是“用掩膜物理地剪切 image2image 的适用范围”。
在这个类型中,将掩膜区域本身作为一种 Conditioning 对待,直接将“希望能只重绘这个范围”这样的条件传递给模型。
在此之上,虽然有各种实现方法,但在 SD1.5 只要掌握以下 2 个系统就足够了。
- 使用 inpainting 专用模型
- 用 ControlNet inpaint 让普通模型支持 inpaint
inpainting 模型
是面向“一边看周围一边填补”的任务调整了 SD1.5 的 Checkpoint。
模型的下载
- stable-diffusion-v1-5/sd-v1-5-inpainting.ckpt
-
📂ComfyUI/ └── 📂models/ └── 📂checkpoints/ └── sd-v1-5-inpainting.ckpt
工作流

{
"id": "8b9f7796-0873-4025-be3c-0f997f67f866",
"revision": 0,
"last_node_id": 21,
"last_link_id": 30,
"nodes": [
{
"id": 9,
"type": "SaveImage",
"pos": [
1773.2999999999997,
213.20000000000002
],
"size": [
503.2876035004722,
576.239673217884
],
"flags": {},
"order": 8,
"mode": 0,
"inputs": [
{
"name": "images",
"type": "IMAGE",
"link": 9
}
],
"outputs": [],
"properties": {
"cnr_id": "comfy-core",
"ver": "0.3.33"
},
"widgets_values": [
"ComfyUI"
]
},
{
"id": 8,
"type": "VAEDecode",
"pos": [
1531.2999999999997,
212.20000000000002
],
"size": [
203.39999999999986,
46
],
"flags": {},
"order": 7,
"mode": 0,
"inputs": [
{
"name": "samples",
"type": "LATENT",
"link": 7
},
{
"name": "vae",
"type": "VAE",
"link": 10
}
],
"outputs": [
{
"name": "IMAGE",
"type": "IMAGE",
"slot_index": 0,
"links": [
9
]
}
],
"properties": {
"cnr_id": "comfy-core",
"ver": "0.3.33",
"Node name for S&R": "VAEDecode"
},
"widgets_values": []
},
{
"id": 6,
"type": "CLIPTextEncode",
"pos": [
415,
186
],
"size": [
411.95503173828126,
151.0030493164063
],
"flags": {},
"order": 3,
"mode": 0,
"inputs": [
{
"name": "clip",
"type": "CLIP",
"link": 3
}
],
"outputs": [
{
"name": "CONDITIONING",
"type": "CONDITIONING",
"slot_index": 0,
"links": [
25
]
}
],
"properties": {
"cnr_id": "comfy-core",
"ver": "0.3.33",
"Node name for S&R": "CLIPTextEncode"
},
"widgets_values": [
"high quality, RAW photo of a woman with red hair, Punch perm"
]
},
{
"id": 7,
"type": "CLIPTextEncode",
"pos": [
416.1970166015625,
392.37848510742185
],
"size": [
410.75801513671877,
158.82607910156253
],
"flags": {},
"order": 4,
"mode": 0,
"inputs": [
{
"name": "clip",
"type": "CLIP",
"link": 5
}
],
"outputs": [
{
"name": "CONDITIONING",
"type": "CONDITIONING",
"slot_index": 0,
"links": [
26
]
}
],
"properties": {
"cnr_id": "comfy-core",
"ver": "0.3.33",
"Node name for S&R": "CLIPTextEncode"
},
"widgets_values": [
"text, watermark, worst quality"
]
},
{
"id": 11,
"type": "LoadImage",
"pos": [
479.08257865587706,
719.0709982393678
],
"size": [
341.9935235795455,
473.93255454545454
],
"flags": {},
"order": 0,
"mode": 0,
"inputs": [],
"outputs": [
{
"name": "IMAGE",
"type": "IMAGE",
"links": [
29
]
},
{
"name": "MASK",
"type": "MASK",
"links": [
30
]
}
],
"properties": {
"cnr_id": "comfy-core",
"ver": "0.3.76",
"Node name for S&R": "LoadImage",
"image": "clipspace/clipspace-painted-masked-1765071052914.png [input]"
},
"widgets_values": [
"clipspace/clipspace-painted-masked-1765071052914.png [input]",
"image"
]
},
{
"id": 10,
"type": "VAELoader",
"pos": [
589.5884989296375,
604.8385752753561
],
"size": [
231.48760330578511,
58
],
"flags": {},
"order": 1,
"mode": 0,
"inputs": [],
"outputs": [
{
"name": "VAE",
"type": "VAE",
"links": [
10,
27
]
}
],
"properties": {
"cnr_id": "comfy-core",
"ver": "0.3.76",
"Node name for S&R": "VAELoader"
},
"widgets_values": [
"vae-ft-mse-840000-ema-pruned.safetensors"
]
},
{
"id": 3,
"type": "KSampler",
"pos": [
1185.2999999999995,
210.20000000000002
],
"size": [
315,
262
],
"flags": {},
"order": 6,
"mode": 0,
"inputs": [
{
"name": "model",
"type": "MODEL",
"link": 1
},
{
"name": "positive",
"type": "CONDITIONING",
"link": 23
},
{
"name": "negative",
"type": "CONDITIONING",
"link": 24
},
{
"name": "latent_image",
"type": "LATENT",
"link": 28
}
],
"outputs": [
{
"name": "LATENT",
"type": "LATENT",
"slot_index": 0,
"links": [
7
]
}
],
"properties": {
"cnr_id": "comfy-core",
"ver": "0.3.33",
"Node name for S&R": "KSampler"
},
"widgets_values": [
45678,
"fixed",
20,
8,
"euler",
"normal",
1
]
},
{
"id": 20,
"type": "InpaintModelConditioning",
"pos": [
885.5197492645148,
229.90120706804433
],
"size": [
258.49395533092,
138
],
"flags": {},
"order": 5,
"mode": 0,
"inputs": [
{
"name": "positive",
"type": "CONDITIONING",
"link": 25
},
{
"name": "negative",
"type": "CONDITIONING",
"link": 26
},
{
"name": "vae",
"type": "VAE",
"link": 27
},
{
"name": "pixels",
"type": "IMAGE",
"link": 29
},
{
"name": "mask",
"type": "MASK",
"link": 30
}
],
"outputs": [
{
"name": "positive",
"type": "CONDITIONING",
"links": [
23
]
},
{
"name": "negative",
"type": "CONDITIONING",
"links": [
24
]
},
{
"name": "latent",
"type": "LATENT",
"links": [
28
]
}
],
"properties": {
"cnr_id": "comfy-core",
"ver": "0.3.76",
"Node name for S&R": "InpaintModelConditioning"
},
"widgets_values": [
true
],
"color": "#232",
"bgcolor": "#353"
},
{
"id": 4,
"type": "CheckpointLoaderSimple",
"pos": [
39.900463636363625,
349.9095500000008
],
"size": [
315,
98
],
"flags": {},
"order": 2,
"mode": 0,
"inputs": [],
"outputs": [
{
"name": "MODEL",
"type": "MODEL",
"slot_index": 0,
"links": [
1
]
},
{
"name": "CLIP",
"type": "CLIP",
"slot_index": 1,
"links": [
3,
5
]
},
{
"name": "VAE",
"type": "VAE",
"slot_index": 2,
"links": []
}
],
"properties": {
"cnr_id": "comfy-core",
"ver": "0.3.33",
"Node name for S&R": "CheckpointLoaderSimple"
},
"widgets_values": [
"sd-v1-5-inpainting.ckpt"
],
"color": "#323",
"bgcolor": "#535"
}
],
"links": [
[
1,
4,
0,
3,
0,
"MODEL"
],
[
3,
4,
1,
6,
0,
"CLIP"
],
[
5,
4,
1,
7,
0,
"CLIP"
],
[
7,
3,
0,
8,
0,
"LATENT"
],
[
9,
8,
0,
9,
0,
"IMAGE"
],
[
10,
10,
0,
8,
1,
"VAE"
],
[
23,
20,
0,
3,
1,
"CONDITIONING"
],
[
24,
20,
1,
3,
2,
"CONDITIONING"
],
[
25,
6,
0,
20,
0,
"CONDITIONING"
],
[
26,
7,
0,
20,
1,
"CONDITIONING"
],
[
27,
10,
0,
20,
2,
"VAE"
],
[
28,
20,
2,
3,
3,
"LATENT"
],
[
29,
11,
0,
20,
3,
"IMAGE"
],
[
30,
11,
1,
20,
4,
"MASK"
]
],
"groups": [],
"config": {},
"extra": {
"ds": {
"scale": 0.6830134553650705,
"offset": [
58.63543636363637,
-84.5359
]
},
"frontendVersion": "1.34.6",
"VHS_latentpreview": false,
"VHS_latentpreviewrate": 0,
"VHS_MetadataImage": true,
"VHS_KeepIntermediate": true
},
"version": 0.4
}
- 🟪 加载 inpainting 模型。
- 🟩 将
VAE Encode、Set Latent Noise Mask替换为InpaintModelConditioning节点。- 输入的参数几乎相同。
- 只有
noise_mask参数需要稍微注意。true- 与
Set Latent Noise Mask时同样,强制只重绘掩膜内部。通常这个设置没问题。
- 与
false- 在一部分模型中,设为
true会破损。作为那时的退路请尝试false。
- 在一部分模型中,设为
在上面的例子中,即使将 denoise 设为 1.00,也能明白为了让图像整体看起来自然而重绘了女性的头发。
与类型 A 不同,变成了“一边看与周围的整合性一边填补掩膜部分”的举动呢。
ControlNet inpaint
inpainting 模型的缺点是,必须使用 inpainting 模型。 有时也想把微调了 Stable Diffusion 1.5 的模型,直接用于 inpainting 吧。
那时,ControlNet inpaint 会派上用场。
关于 ControlNet,将在另外的页面进行说明。
自定义节点
ControlNet 模型的下载
- comfyanonymous/control_v11p_sd15_inpaint_fp16.safetensors
-
📂ComfyUI/ └── 📂models/ └── 📂controlnet/ └── control_v11p_sd15_inpaint_fp16.safetensors
工作流

{
"id": "8b9f7796-0873-4025-be3c-0f997f67f866",
"revision": 0,
"last_node_id": 30,
"last_link_id": 53,
"nodes": [
{
"id": 9,
"type": "SaveImage",
"pos": [
1757.2999999999997,
214.20000000000002
],
"size": [
503.2876035004722,
576.239673217884
],
"flags": {},
"order": 12,
"mode": 0,
"inputs": [
{
"name": "images",
"type": "IMAGE",
"link": 9
}
],
"outputs": [],
"properties": {
"cnr_id": "comfy-core",
"ver": "0.3.33"
},
"widgets_values": [
"ComfyUI"
]
},
{
"id": 8,
"type": "VAEDecode",
"pos": [
1515.2999999999997,
213.20000000000002
],
"size": [
203.39999999999986,
46
],
"flags": {},
"order": 11,
"mode": 0,
"inputs": [
{
"name": "samples",
"type": "LATENT",
"link": 7
},
{
"name": "vae",
"type": "VAE",
"link": 10
}
],
"outputs": [
{
"name": "IMAGE",
"type": "IMAGE",
"slot_index": 0,
"links": [
9
]
}
],
"properties": {
"cnr_id": "comfy-core",
"ver": "0.3.33",
"Node name for S&R": "VAEDecode"
},
"widgets_values": []
},
{
"id": 7,
"type": "CLIPTextEncode",
"pos": [
420.98808709870343,
387.97848510742176
],
"size": [
410.75801513671877,
158.82607910156253
],
"flags": {},
"order": 7,
"mode": 0,
"inputs": [
{
"name": "clip",
"type": "CLIP",
"link": 5
}
],
"outputs": [
{
"name": "CONDITIONING",
"type": "CONDITIONING",
"slot_index": 0,
"links": [
33
]
}
],
"properties": {
"cnr_id": "comfy-core",
"ver": "0.3.33",
"Node name for S&R": "CLIPTextEncode"
},
"widgets_values": [
"text, watermark, worst quality"
]
},
{
"id": 22,
"type": "ControlNetApplyAdvanced",
"pos": [
901.0437048084135,
232.48767017280198
],
"size": [
235.54470323065357,
186
],
"flags": {},
"order": 9,
"mode": 0,
"inputs": [
{
"name": "positive",
"type": "CONDITIONING",
"link": 32
},
{
"name": "negative",
"type": "CONDITIONING",
"link": 33
},
{
"name": "control_net",
"type": "CONTROL_NET",
"link": 39
},
{
"name": "image",
"type": "IMAGE",
"link": 31
},
{
"name": "vae",
"shape": 7,
"type": "VAE",
"link": 34
}
],
"outputs": [
{
"name": "positive",
"type": "CONDITIONING",
"links": [
37
]
},
{
"name": "negative",
"type": "CONDITIONING",
"links": [
38
]
}
],
"properties": {
"cnr_id": "comfy-core",
"ver": "0.3.76",
"Node name for S&R": "ControlNetApplyAdvanced"
},
"widgets_values": [
0.8,
0,
1
],
"color": "#232",
"bgcolor": "#353"
},
{
"id": 6,
"type": "CLIPTextEncode",
"pos": [
419.79107049714094,
181.60000000000002
],
"size": [
411.95503173828126,
151.0030493164063
],
"flags": {},
"order": 6,
"mode": 0,
"inputs": [
{
"name": "clip",
"type": "CLIP",
"link": 3
}
],
"outputs": [
{
"name": "CONDITIONING",
"type": "CONDITIONING",
"slot_index": 0,
"links": [
32
]
}
],
"properties": {
"cnr_id": "comfy-core",
"ver": "0.3.33",
"Node name for S&R": "CLIPTextEncode"
},
"widgets_values": [
"high quality, RAW photo of a woman with red hair, Punch perm"
]
},
{
"id": 26,
"type": "ControlNetLoader",
"pos": [
573.8122179379016,
601.5646590492796
],
"size": [
257.93388429752065,
58
],
"flags": {},
"order": 0,
"mode": 0,
"inputs": [],
"outputs": [
{
"name": "CONTROL_NET",
"type": "CONTROL_NET",
"links": [
39
]
}
],
"properties": {
"cnr_id": "comfy-core",
"ver": "0.3.76",
"Node name for S&R": "ControlNetLoader"
},
"widgets_values": [
"control_v11p_sd15_inpaint_fp16.safetensors"
],
"color": "#232",
"bgcolor": "#353"
},
{
"id": 24,
"type": "InpaintPreprocessor",
"pos": [
593.6777428604222,
728.2127272727273
],
"size": [
238.068359375,
78
],
"flags": {},
"order": 5,
"mode": 0,
"inputs": [
{
"name": "image",
"type": "IMAGE",
"link": 35
},
{
"name": "mask",
"type": "MASK",
"link": 36
}
],
"outputs": [
{
"name": "IMAGE",
"type": "IMAGE",
"links": [
31
]
}
],
"properties": {
"cnr_id": "comfyui_controlnet_aux",
"ver": "12f35647f0d510e03b45a47fb420fe1245a575df",
"Node name for S&R": "InpaintPreprocessor"
},
"widgets_values": [
false
],
"color": "#432",
"bgcolor": "#653"
},
{
"id": 3,
"type": "KSampler",
"pos": [
1169.2999999999995,
211.20000000000002
],
"size": [
315,
262
],
"flags": {},
"order": 10,
"mode": 0,
"inputs": [
{
"name": "model",
"type": "MODEL",
"link": 1
},
{
"name": "positive",
"type": "CONDITIONING",
"link": 37
},
{
"name": "negative",
"type": "CONDITIONING",
"link": 38
},
{
"name": "latent_image",
"type": "LATENT",
"link": 53
}
],
"outputs": [
{
"name": "LATENT",
"type": "LATENT",
"slot_index": 0,
"links": [
7
]
}
],
"properties": {
"cnr_id": "comfy-core",
"ver": "0.3.33",
"Node name for S&R": "KSampler"
},
"widgets_values": [
12345,
"fixed",
20,
8,
"euler",
"normal",
1
]
},
{
"id": 10,
"type": "VAELoader",
"pos": [
395.1303171114555,
1012.947966184447
],
"size": [
231.48760330578511,
58
],
"flags": {},
"order": 1,
"mode": 0,
"inputs": [],
"outputs": [
{
"name": "VAE",
"type": "VAE",
"links": [
10,
34,
41
]
}
],
"properties": {
"cnr_id": "comfy-core",
"ver": "0.3.76",
"Node name for S&R": "VAELoader"
},
"widgets_values": [
"vae-ft-mse-840000-ema-pruned.safetensors"
]
},
{
"id": 27,
"type": "VAEEncode",
"pos": [
486.7770161500693,
868.15734837917
],
"size": [
140,
46
],
"flags": {},
"order": 4,
"mode": 0,
"inputs": [
{
"name": "pixels",
"type": "IMAGE",
"link": 43
},
{
"name": "vae",
"type": "VAE",
"link": 41
}
],
"outputs": [
{
"name": "LATENT",
"type": "LATENT",
"links": [
51
]
}
],
"properties": {
"cnr_id": "comfy-core",
"ver": "0.3.76",
"Node name for S&R": "VAEEncode"
},
"color": "#322",
"bgcolor": "#533"
},
{
"id": 30,
"type": "SetLatentNoiseMask",
"pos": [
654.6523241448506,
916.4969147623518
],
"size": [
180.74765625,
46
],
"flags": {},
"order": 8,
"mode": 0,
"inputs": [
{
"name": "samples",
"type": "LATENT",
"link": 51
},
{
"name": "mask",
"type": "MASK",
"link": 52
}
],
"outputs": [
{
"name": "LATENT",
"type": "LATENT",
"links": [
53
]
}
],
"properties": {
"cnr_id": "comfy-core",
"ver": "0.3.76",
"Node name for S&R": "SetLatentNoiseMask"
},
"color": "#322",
"bgcolor": "#533"
},
{
"id": 11,
"type": "LoadImage",
"pos": [
17.462760474058893,
915.0939090909087
],
"size": [
341.9935235795455,
473.93255454545454
],
"flags": {},
"order": 2,
"mode": 0,
"inputs": [],
"outputs": [
{
"name": "IMAGE",
"type": "IMAGE",
"links": [
35,
43
]
},
{
"name": "MASK",
"type": "MASK",
"links": [
36,
52
]
}
],
"properties": {
"cnr_id": "comfy-core",
"ver": "0.3.76",
"Node name for S&R": "LoadImage",
"image": "clipspace/clipspace-painted-masked-1765071052914.png [input]"
},
"widgets_values": [
"clipspace/clipspace-painted-masked-1765071052914.png [input]",
"image"
]
},
{
"id": 4,
"type": "CheckpointLoaderSimple",
"pos": [
59.95686363636362,
333.7249500000008
],
"size": [
315,
98
],
"flags": {},
"order": 3,
"mode": 0,
"inputs": [],
"outputs": [
{
"name": "MODEL",
"type": "MODEL",
"slot_index": 0,
"links": [
1
]
},
{
"name": "CLIP",
"type": "CLIP",
"slot_index": 1,
"links": [
3,
5
]
},
{
"name": "VAE",
"type": "VAE",
"slot_index": 2,
"links": []
}
],
"properties": {
"cnr_id": "comfy-core",
"ver": "0.3.33",
"Node name for S&R": "CheckpointLoaderSimple"
},
"widgets_values": [
"v1-5-pruned-emaonly-fp16.safetensors"
]
}
],
"links": [
[
1,
4,
0,
3,
0,
"MODEL"
],
[
3,
4,
1,
6,
0,
"CLIP"
],
[
5,
4,
1,
7,
0,
"CLIP"
],
[
7,
3,
0,
8,
0,
"LATENT"
],
[
9,
8,
0,
9,
0,
"IMAGE"
],
[
10,
10,
0,
8,
1,
"VAE"
],
[
31,
24,
0,
22,
3,
"IMAGE"
],
[
32,
6,
0,
22,
0,
"CONDITIONING"
],
[
33,
7,
0,
22,
1,
"CONDITIONING"
],
[
34,
10,
0,
22,
4,
"VAE"
],
[
35,
11,
0,
24,
0,
"IMAGE"
],
[
36,
11,
1,
24,
1,
"MASK"
],
[
37,
22,
0,
3,
1,
"CONDITIONING"
],
[
38,
22,
1,
3,
2,
"CONDITIONING"
],
[
39,
26,
0,
22,
2,
"CONTROL_NET"
],
[
41,
10,
0,
27,
1,
"VAE"
],
[
43,
11,
0,
27,
0,
"IMAGE"
],
[
51,
27,
0,
30,
0,
"LATENT"
],
[
52,
11,
1,
30,
1,
"MASK"
],
[
53,
30,
0,
3,
3,
"LATENT"
]
],
"groups": [],
"config": {},
"extra": {
"ds": {
"scale": 0.683013455365071,
"offset": [
82.5372395259411,
-81.60000000000002
]
},
"frontendVersion": "1.34.6",
"VHS_latentpreview": false,
"VHS_latentpreviewrate": 0,
"VHS_MetadataImage": true,
"VHS_KeepIntermediate": true
},
"version": 0.4
}
- 加载喜欢的 SD1.5 checkpoint(+LoRA)
- 🟨 将图像和掩膜输入
Inpaint Preprocessor,转换为 ControlNet 用的图像- 实际上,只是把掩膜部分涂黑了。
- 🟩 向
Apply ControlNet节点输入 ControlNet 模型・图像・VAE - 🟥 组入在上面做的使用了
Set Latent Noise Mask的 inpainting
连结到 SDXL / Flux 等
本页面是专注于 SD1.5 的,但此外也存在几个 inpainting 手段。
- Fooocus inpaint(面向 SDXL 的 inpaint 模型)
- Flux.fill(Flux 系的填充功能)
- Lanpaint(图像编辑・inpaint 系工具)
这些预定另行处理。