sebastiankamph
patreon
AnimateAnyone Patreon supporter workflow
🕑 Added 2024-01-29 12:41:57 +0000 UTCWorkflow from the video guide on AnimateAnyone here: https://youtu.be/MGvx37ccCOM
https://github.com/MrForExample/ComfyUI-AnimateAnyone-Evolved
Install (You can also use ComfyUI Manager)
Clone this repo into the Your ComfyUI root directory\ComfyUI\custom_nodes\ and install dependent Python packages:cd Your_ComfyUI_root_directory\ComfyUI\custom_nodes\
git clone https://github.com/MrForExample/ComfyUI-AnimateAnyone-Evolved.git
pip install -r requirements.txt
# If you got error regards diffusers then run:
pip install --force-reinstall diffusers>=0.26.1Download pre-trained models (new links):https://huggingface.co/stable-diffusion-v1-5/stable-diffusion-v1-5/tree/main/unet
&
Moore-AnimateAnyone Pre-trained Models
Above models need to be put under folder pretrained_weights as follow:
./pretrained_weights/
|-- denoising_unet.pth
|-- motion_module.pth
|-- pose_guider.pth
|-- reference_unet.pth
`-- stable-diffusion-v1-5
|-- feature_extractor
| `-- preprocessor_config.json
|-- model_index.json
|-- unet
| |-- config.json
| `-- diffusion_pytorch_model.bin
`-- v1-inference.yaml
Download clip image encoder (e.g. sd-image-variations-diffusers ) and put it under Your_ComfyUI_root_directory\ComfyUI\models\clip_vision
Download vae (e.g. sd-vae-ft-mse) and put it under Your_ComfyUI_root_directory\ComfyUI\models\vae
This workflow includes a facedetailer and also an extra workflow to extract openpose videos from your input videos.
If you are encountering this error, you can check what nodes are red, and reload them by manually adding them again
TROUBLESHOOTING:
UPDATE: User JayShay in Discord suggested that the portable install of Comfy had issues with AnimateAnyone, and that installing ComfyUI fresh through git clone helped him and others.
A lot of users seem to have issues with conflicting nodes. I would recommend installing this on a fresh Comfy to not break your old one (unless you know what you're doing). The fix would be to just remove the newly installed custom nodes again.
Some users have also reported that updating their pytorch to match their cuda version has helped. Here's how a user with cuda 11.8 solved his issue:
pip install torch torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/cu118
If you have errors from missing custom nodes. Try a fresh install of Comfy, make sure it's the latest version. You can also try manually installing AnimateAnyone (can try git clone from github into custom nodes folder).
Download below!
Comments
Sebastian Kamph
You can't mix sdxl and 1.5. Stick with 1.5 models
Sergiu Falcusan
Found solution here: https://github.com/MrForExample/ComfyUI-AnimateAnyone-Evolved/issues/17
Sergiu Falcusan
I've reloaded all the red nodes, however I don't know which settings I need to use. I get this error: Error occurred when executing [AnimateAnyone] Animate Anyone Sampler: mat1 and mat2 shapes cannot be multiplied (2x1024 and 768x320)
DJ
What do you mean you renamed the modules with 3D? What did you rename the models to?
Sergio Valsecchi
@Sebastian, I succeeded and thank you for the guidance. In my case, after various attempts, including installing comfy from scratch, I tried to install following this sequence: comfy from scratch first module installed AnimateAnyone then manager then all the missing ones + ComfyUI-AnimateDiff-Evolved, by cloning it from https://github.com/Kosinkadink/ComfyUI-AnimateDiff-Evolved then I renamed the remaining modules with 3d replacing the flows with the module that was giving an error, there are 5/6 Upon restart, everything worked!
Stefan Burger
No it is not ... right click on the red nodes, write down or remember their name, then double click on the canvas and search for this name and add them again, then connect them as the red ones were and after delete the red ones that could not load.
Sergio Valsecchi
I've tried in every way: A new and clean Comfy setup, and it just won't load for me. [ComfyUI-3D] Animate Anyone Sampler [ComfyUI-3D] Load UNet2D ConditionModel [ComfyUI-3D] Load UNet3D ConditionModel [ComfyUI-3D] Pose Guider Encode [ComfyUI-3D] Load Pose Guider Could it be due to the video card? I have an RTX 2080 ti 12 GB.
SeCha
How do you reinstall pytorch
Sebastian Kamph
Most of the time, you can find the missing custom nodes from "Install missing custom nodes". If this isn't enough (make sure you have latest versions), it's on to google and github to find those little niche nodes ;(. All the nodes should be available through this guide though.
Sam B
I am missing all the nodes. How do I know which ones to install? I installed the animateanyone evolved from the manager but almost every node is red and undefined
Destin Peters
nope... not just you. Broke a fresh conda env and a stand alone.
Dias Nursu
I tried, it didn't fix the problem...
Sebastian Kamph
A lot of users seem to be having this issue. And looking at that github issue post, the author seems unsure how to solve currently.
Mehmet Arsak
Statt auf meinen Post zu antworten wird er gelöscht. Nochmal, das Problem mit den Nodes ist bekannt, eine echte Lösung gibt es momentan nicht: https://github.com/MrForExample/ComfyUI-AnimateAnyone-Evolved/issues/4
Mehmet Arsak
portable and regular 1 not working, also its not possible to install 3D image pack, issue cant be fixed with a regular method: https://github.com/MrForExample/ComfyUI-AnimateAnyone-Evolved/issues/4
Sebastian Kamph
See update in post. JayShay and others got it working through installing Comfy from a git clone and not downloading the portable one.
Sebastian Kamph
Hey! That's really what Comfy is about. It's quite time consuming to get running. The ease of use part, is you can copy/paste workflows and adapt once you know what's going on. But no, I don't think it's a better user experience unless you're very tech minded and want to fiddle with the internal workings.
Mehmet Arsak
Same problem here, s the workflow I paid for is useless!!
Benjamin Benichou
Ok nothing changed still the same issue with loading and at the Node level I have an error message: (IMPORT FAILED) ComfyUI-Moore-AnimateAnyone
JayShay
This is my first time even running comfy UI stable diffusion locally But I was interested in this animate anyone Come over to the discord. We're in the help channel. I figured out how to get it loaded
Benjamin Benichou
Yes but I noticed that from the download weights command, some content is missing from this: ./pretrained_weights/ |-- DWPose | |-- dw-ll_ucoco_384.onnx | `-- yolox_l.onnx |-- image_encoder | |-- config.json | `-- pytorch_model.bin |-- denoising_unet.pth |-- motion_module.pth |-- pose_guider.pth |-- reference_unet.pth |-- sd-vae-ft-mse | |-- config.json | |-- diffusion_pytorch_model.bin | `-- diffusion_pytorch_model.safetensors `-- stable-diffusion-v1-5 |-- feature_extractor | `-- preprocessor_config.json |-- model_index.json |-- unet | |-- config.json | `-- diffusion_pytorch_model.bin `-- v1-inference.yaml So I'm trying to complete it manually and I'll let you know if this works.
Sebastian Kamph
Sorry to hear about that. I recommend trying the troubleshooting steps above. If nothing works, you'd have to wait for more updates. Did you already try reinstalling pytorch?
Benjamin Benichou
Hi John! "new_ComfyUI_windows_portable_nvidia_cu121_or_cpu" setup locally and I have a 3090 RTX. I'm not expert but I have been able to setup and run workflow previously, install custom nodes, etc. I even reinstalled everything to start fresh but still stuck at the: When loading the graph, the following node types were not found: [AnimateAnyone] Load Pose Guider [AnimateAnyone] Animate Anyone Sampler [AnimateAnyone] Load UNet2D ConditionModel [AnimateAnyone] Load UNet3D ConditionModel [AnimateAnyone] Pose Guider Encode Nodes that have failed to load will show as red on the graph.
JayShay
hi Benjamin. Are you using ComfyUI Local or Regular install?
Benjamin Benichou
I'm getting the same: When loading the graph, the following node types were not found: [ComfyUI-3D] Animate Anyone Sampler [ComfyUI-3D] Load UNet2D ConditionModel [ComfyUI-3D] Load UNet3D ConditionModel [ComfyUI-3D] Pose Guider Encode [ComfyUI-3D] Load Pose Guider Nodes that have failed to load will show as red on the graph.
Stefan Steeger
I have the same Issue
thekriskooper
And this too: And in ComfyUI Manager getting (IMPORT FAILED) ComfyUI-AnimateAnyone-Evolved. Anybody have a fix for this?
thekriskooper
Me too... Hi, I've tried everything but am also getting the following error. When loading the graph, the following node types were not found: [ComfyUI-3D] Load UNet2D ConditionModel [ComfyUI-3D] Load UNet3D ConditionModel [ComfyUI-3D] Pose Guider Encode [ComfyUI-3D] Animate Anyone Sampler [ComfyUI-3D] Load Pose Guider Nodes that have failed to load will show as red on the graph.
JayShay
And in ComfyUI Manager getting (IMPORT FAILED) ComfyUI-AnimateAnyone-Evolved. Anybody have a fix for this?
JayShay
Hi, I've tried everything but am also getting the following error. When loading the graph, the following node types were not found: [ComfyUI-3D] Load UNet2D ConditionModel [ComfyUI-3D] Load UNet3D ConditionModel [ComfyUI-3D] Pose Guider Encode [ComfyUI-3D] Animate Anyone Sampler [ComfyUI-3D] Load Pose Guider Nodes that have failed to load will show as red on the graph.
tdog
tried a fresh install, its now much worse... I don't know if it's me or what, but I'm getting discouraged from using ComfyUI
Fridabee
Hey Seb, I've been using SD for almost two years, but I never switched to ComfyUI because I find the logic spaghetti...NOT comfy at all! I tried it but it doesn't make sense to me. It is complex, takes longer than SD 1.5 to get something if anything out of it. I don't see how having to build the software's logic sequence from ground up every time is an improvement as an interface. I feel it takes a lot of the fun out of prompting. Am I missing out on something? Do you have a video that addreses these concerns? thank you
Philip Bruinette
I am getting the same error, No module named 'diff_gaussian_rasterization' so it won't import comfyUI-3d. How do I fix this so that it works. I've downloaded that submodule but I don't know what to do with it
Sebastian Kamph
Currently no
Sebastian Kamph
Hey, talked to TD devs. Here's a fix: Here is the forked extension with customisation: https://github.com/thinkdiffusion/ComfyUI-AnimateAnyone-Evolved.git Remove your previously installed custom-node from comfy > custom-nodes > ComfyUI-AnimateAnyone-Evolved Install Custom node from url above
Oskar Ekehäll
Same issue here, ComfyUI-3d won't work after install. (Fresh ThinkDiffusion instance).
Black Swan
I had the same error, sharing what worked for me: I disabled animate anyone, restarted my UI and instead of loading the workflow and installing missing custom nodes I reinstalled Animate Anyone through the install custom nodes button in the manager, then restarted UI again and when I loaded the workflow everything was good.
Chuan Lim
— I had the same with another " diffusers " based plugin .. People are releasing a lot of shitty plug ins at the moment. That just package the complete " diffusers " set up and u - nets from latest papers. So the image controls are inside that new instance of " diffusers " and not read from " comfy ui " at all ..
Chuan Lim
The " comfyui-3d-pack " is missing " diff_gaussian_rasterization " from its dependencies. Though calls it inside the " main_3dgs_renderer.py " script. The add - on will never load without it .. — Eg. Its a submodule from this repo : [ https://github.com/graphdeco-inria/gaussian-splatting ]
David Binkley
Go into the manager > Add custom nodes and search for ComfyUI-3d and install those nodes.
indievish
do you have this workflow with sdxl?
polaraju p
what i do ?
polaraju p
When loading the graph, the following node types were not found: [ComfyUI-3D] Animate Anyone Sampler [ComfyUI-3D] Load UNet2D ConditionModel [ComfyUI-3D] Load UNet3D ConditionModel [ComfyUI-3D] Pose Guider Encode [ComfyUI-3D] Load Pose Guider Nodes that have failed to load will show as red on the graph.
Sebastian Kamph
Happy to hear it!
Bob Doyle - VR Films
The width and height adjustments do not seem to have an impact on the output size. No matter what I put in there, it renders a portrait video, but the source is landscape. The Openpose video and source image are all the right dimensions, but changing the values has no effect on the final video dimensions, so it's all squished. Suggestions? When I do a portrait video, it works fine!
Bob Doyle - VR Films
I actually found a simpler fix! The problem seemed to be that somehow pytorch was no longer compatible with my cude 11.8 environment. I just reinstalled pytorch with the following and everything worked again: pip install torch torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/cu118
Sebastian Kamph
You could just go into your custom nodes folder and delete the nodes you installed. Comfy has a lot of conflicting nodes, sometimes they break. However, even if you would do a fresh install, you can just copy/paste files from your older install, like config, path.yaml file, models etc. I probably make a fresh one every two weeks because something breaks it too much.
Bob Doyle - VR Films
Yikes. Installing this killed my ComfyUI installation, now getting "AssertionError: Torch not compiled with CUDA enabled" and everything I'm reading says to do a "fresh Comfy install." So I want to make sure I understand this. Does this mean that all the nodes associated with my current installation would all have to be reinstalled? Like I'm truly starting from scratch? I have to redirect my models folder and all that like before, just to get operational again? I'm hoping there is a way to undo this installation or something because I can't fathom having to rebuild my entire comfy environment from scratch again just because of this. What can I do aside from starting all over? Or am I misunderstanding what a "Fresh comfyui install" is. From the video, it looks like you're starting all over.
