基于 instantX-research/diffusers_sd3_control: ? Diffusers: State-of-the-art diffusion models for image and audio generation in PyTorch and FLAX. (github.com)
和
ZHO-ZHO-ZHO/ComfyUI-SD3-Medium-CN-Diffusers: ComfyUI SD3-Medium ControlNet(Diffusers) (github.com)
改的,这两个repo都有些问题,中间看提交记录,有好些人在参与改动,为了能跑的了,我分别fork了这两个分支, 在这里:
prodogape/ComfyUI-SD3-Medium-CN-Diffusers: ComfyUI SD3-Medium ControlNet(Diffusers) (github.com)
prodogape/diffusers_sd3_control: refactor name from "diffusers" to "diffusers_sd3_control" (github.com)
1. 先说怎么装
打开第一个橘色背景的repo:
prodogape/ComfyUI-SD3-Medium-CN-Diffusers: ComfyUI SD3-Medium ControlNet(Diffusers) (github.com)
按照Install这里的步骤安装
时间有限只写了手动安装的步骤,分4步:
(1)git 克隆 comfyui的节点
(2)下载hf的两个仓库的模型(一个controlnet一个sd3)
(3)修改sd3模型文件夹下的文件model_index.json,因为要安装的特殊版本的diffusers我改名为diffusers_sd3_control了
(4)安装diffusers_sd3_control
2. 工作流及效果
这个版本的实现应该还是有些问题的,可能跟control的百分比和control了不太对的部分有关
3. 上图的工作流(非API版本工作流)
{ "last_node_id": 82, "last_link_id": 139, "nodes": [ { "id": 77, "type": "CannyEdgePreprocessor", "pos": [ -1756, -33 ], "size": { "0": 315, "1": 106 }, "flags": {}, "order": 2, "mode": 0, "inputs": [ { "name": "image", "type": "IMAGE", "link": 133 } ], "outputs": [ { "name": "IMAGE", "type": "IMAGE", "links": [ 134, 139 ], "shape": 3, "slot_index": 0 } ], "properties": { "Node name for S&R": "CannyEdgePreprocessor" }, "widgets_values": [ 100, 200, 512 ] }, { "id": 82, "type": "PreviewImage", "pos": [ -1753, 110 ], "size": { "0": 299.701171875, "1": 326.380859375 }, "flags": {}, "order": 4, "mode": 0, "inputs": [ { "name": "images", "type": "IMAGE", "link": 139 } ], "properties": { "Node name for S&R": "PreviewImage" } }, { "id": 80, "type": "PreviewImage", "pos": [ -970, -40 ], "size": { "0": 438.1802978515625, "1": 471.4967041015625 }, "flags": {}, "order": 5, "mode": 0, "inputs": [ { "name": "images", "type": "IMAGE", "link": 137 } ], "properties": { "Node name for S&R": "PreviewImage" } }, { "id": 74, "type": "LoadImage", "pos": [ -2113, -24 ], "size": { "0": 349.269775390625, "1": 459.3564147949219 }, "flags": {}, "order": 0, "mode": 0, "outputs": [ { "name": "IMAGE", "type": "IMAGE", "links": [ 133 ], "shape": 3, "slot_index": 0 }, { "name": "MASK", "type": "MASK", "links": null, "shape": 3 } ], "properties": { "Node name for S&R": "LoadImage" }, "widgets_values": [ "IMG_20240617_070212_131.jpeg", "image" ] }, { "id": 76, "type": "SD3MCN_BaseModelLoader_Zho", "pos": [ -1424, -42 ], "size": { "0": 428.3705749511719, "1": 95.87933349609375 }, "flags": {}, "order": 1, "mode": 0, "outputs": [ { "name": "pipe", "type": "SD3MMODEL", "links": [ 132 ], "shape": 3, "slot_index": 0 } ], "properties": { "Node name for S&R": "SD3MCN_BaseModelLoader_Zho" }, "widgets_values": [ "stabilityai/stable-diffusion-3-medium-diffusers", "InstantX/SD3-Controlnet-Canny" ] }, { "id": 78, "type": "SD3MCN_Generation_Zho", "pos": [ -1436, 94 ], "size": { "0": 444.2547607421875, "1": 344.82293701171875 }, "flags": {}, "order": 3, "mode": 0, "inputs": [ { "name": "pipe", "type": "SD3MMODEL", "link": 132 }, { "name": "image", "type": "IMAGE", "link": 134 } ], "outputs": [ { "name": "IMAGE", "type": "IMAGE", "links": [ 137 ], "shape": 3, "slot_index": 0 } ], "properties": { "Node name for S&R": "SD3MCN_Generation_Zho" }, "widgets_values": [ "close-up of a young girl, masterpiece, best quality, photo, realistic, very aesthetic", "worst quality, low quality, toon", 512, 512, 40, 9, 1, 542410872807786, "randomize" ] } ], "links": [ [ 132, 76, 0, 78, 0, "SD3MMODEL" ], [ 133, 74, 0, 77, 0, "IMAGE" ], [ 134, 77, 0, 78, 1, "IMAGE" ], [ 137, 78, 0, 80, 0, "IMAGE" ], [ 139, 77, 0, 82, 0, "IMAGE" ] ], "groups": [], "config": {}, "extra": { "workspace_info": { "id": "-gBBQ2xIYibjLVAhj5iML" }, "ds": { "scale": 1.3513057093104035, "offset": { "0": 2165.8774539604974, "1": 287.4375476041587 } } }, "version": 0.4 }
comfyuimediumgitgithubcontrolnettpummo工作流diffusionsemjsonapi文件夹2024erppytorchcto