Showing a basic example of how to interpolate between poses in comfyui!
Used some re-rerouting nodes to make it easier to copy and paste the open pose groups. This way you can essentially do keyframing with different open pose images.
3 keyframes v1.1
Now with a demonstration of how to mix keyframes with prompt scheduling! Uses AnimateDiff-Evolved nodes to make this work.
Did you find these workflows helpful? Feel free to leave me a donation: https://www.buymeacoffee.com/bjew
FAQ:
Q: Which AnimateDiff nodes should I use?
A: The 3 keyframes v1.1 demonstration uses the newer AnimateDiff-Evolved: Kosinkadink/ComfyUI-AnimateDiff-Evolved: Improved AnimateDiff for ComfyUI (github.com)
This is to enable the prompt scheduling. The older versions uses the original nodes: ArtVentureX/comfyui-animatediff: AnimateDiff for ComfyUI (github.com)
It's suggested to not have both enabled at the same time as it can cause errors. Try uninstalling one before installing the other if you run into errors.
Q: I see this error when running the workflow: "Error occurred when executing LatentKeyframeTiming: strength_to must be greater than or equal to strength_from"
A: Update ComfyUI-Advanced-ControlNet to a newer version. For those familar with git commands, the workflow is using commit 7213bc69f5d48fe9b40da8017e7762d73f9cc845.
Description
Demostrating how it can be used with prompt scheduling. Using the newer animatediff nodes.
FAQ
Comments (20)
Awesome workflow!
Can you (or someone) please provide a workflow that includes hires and or face detailer? Thanks!
The github page for the animatediff-evolve has some examples of those:
Kosinkadink/ComfyUI-AnimateDiff-Evolved: Improved AnimateDiff for ComfyUI (github.com)
The basic idea is that after the ksampler node you can pass that into another one for hi-res or face detailer. Then from there combine it to a video.
@a1lazydogย Ok, I will have a look. Thanks.
Upload your Start , midle,end, image
Hello! Thank you for sharing your workflow - it looks most impressive and I look forward to making some good animations with it.
I am havng a problem though... I get this error when I run the workflow. Has this occured during your testing?
"Error occurred when executing ImpactKSamplerBasicPipe: 'NoneType' object has no attribute 'shape'"
I would much appreciate any help you can afford!
Hello! Thank you for sharing your workflow - it looks most impressive and I look forward to making some good animations with it.
I am havng a problem though... I get this error when I run the workflow. Has this occured during your testing?
Error occurred when executing ImpactKSamplerBasicPipe:
index 10 is out of bounds for dimension 0 with size 1
So there's a bunch of custom nodes and stuff you can't find in ComfyUI Manager or on Civitai, so I can't get started from what's shown here.
In particular, AnimateDiff doesn't even have some of the animating nodes that you're using anymore and largely seems to have been deleted, telling me to use VideoHelperSuite.
So the entire 'Animation' group is red.
You want
AnimatedDiff-Evolved
It's in manager. I don't know the state of the original.
The only problem with ComfyUI-Manager is that old / dead repos never seem to get removed from the list (there should at least be an "outdated" list. Many repots haven't had any files changed in 4 months which I'd consider dead with how often the ABIs of comfy, torch, diffusers, and everything else introduce breaking changes to add new features.
Hi. Thanks for sharing. I'm getting an error when trying to run this workflow:
"AttributeError: 'NoneType' object has no attribute 'lower'"
Any idea how to fix this? Thanks in advance.
This line precedes the error: "File "C:\ComfyUI_windows_portable_nvidia_cu121_or_cpu\ComfyUI_windows_portable\ComfyUI\comfy\utils.py", line 12, in load_torch_file if ckpt.lower().endswith(".safetensors"):"
same problem to me. Can't find solution :(
Yeah your controls are missing. Change them.
Ok, I am a real comfyui novice and not really that programing savue. What does "Type 'ControlLLLiteAdvanced' must be used with Apply Advanced ControlNet ๐๐ ๐ ๐ node (with model_optional passed in); otherwise, it will not work." mean? I love your two frame workflow and would really like to get the 3 frame one working.
Try RunDiffusion. It loaded with "NO ERRORS"
Error occurred when executing ControlNetLoaderAdvanced: 'ControlNet' object has no attribute 'latent_format'
Runs perfect in RunDiffusion.Thanks. What's your Coffee Link?
It runs for me until "KSampler(pipe) . Then I get this error: ImpactKSamplerBasicPipe
mat1 and mat2 shapes cannot be multiplied (2464x2048 and 768x320). Any idea what I have to change?
I used the example pics which are all square
I used the wrong controlnet model. Now it works.
