To go with the guide: https://civarchive.com/articles/5906/guide-unsampling-for-animatediffhotshot-an-inner-reflections-guide
Description
SDXL Version
FAQ
Comments (25)
:O
Used resources of the workflow:
hsxl_temporal_layers.f16.safetensors
https://huggingface.co/hotshotco/Hotshot-XL/tree/main
xinsir controlnet-scribble-sdxl (alternatively you can install it in the model tab of the Comfy Manager)
https://huggingface.co/xinsir/controlnet-scribble-sdxl-1.0/tree/main
Checkpoint recommended:
Anime
https://civitai.com/models/118406?modelVersionId=563988
Realistic
https://civitai.com/models/119229/zavychromaxl
https://civitai.com/models/133005?modelVersionId=456194
Pixelart
https://civitai.com/models/451177?modelVersionId=502362
If the video output is flashing
1. Add Custom Sigmas node to the workflow and connect it to the FlipSigmas input (disconnect the KSamplerSelect & AlignYourStepsScheduler).
2. Click the "?" on the top right corner of the Custom Sigmas node and copy the SDXL numbers and replace what's inside the node.
3. Remove the last number (0.029).
Gratzi Catz!
Can you add a screen shot of this? I've been sticking to SD1.5 due to the flashing issue.
@bprzypek866 Show in the stream replay at this minute mark https://youtu.be/-E-DNhFumW8?t=1795
"Error occurred when executing SamplerCustom: 0" Could anybody help me with this error?
Thats a very Wierd error - not one I have seen.
@Inner_Reflections_AI I guess the problem was with the Hotshot package. Reinstallation helped ^^
Error occurred when executing SamplerCustom: 0. It Only when control net is On.
@sd_storm Sorry Its a wierd error make sure you have all the files I guess - post the full issue on banadoco if you can't figure it out.
I saw you on the livestream and checked out this workflow. Dude Finally! It's awesome. I'll share some stuff I've come up with soon. I've been on a quest since Waking Life to automate rotoscoping. The time is now, and you have contributed immensely to that. Thank you!
Coolio!!
my error:
Error occurred when executing VHS_LoadVideoPath:
could not broadcast input array from shape (1080,1920,3) into shape (1920,1080,3)
I think you are trying to force size. Try not doing that.
its disable, its worked only for 1080x1920. thats error when use video 1920x1080.
@Inner_Reflections_AI its worked now when I convert size video to HD , thanks for your work!
Tried to add adetailer at the end of the flow and it brought back flickering to the face... any s suggestions to adding a face detailer ? where in the process ?
I haven't explored addetailer too much. You could try to render just a portion of the video - composit it back in and then upscale at low denoise to blend.
I tried using a mask to add the style just to the subject, but it ended up looking like noise. Have you tried doing this before?
how to upscale
This workflow keeps saying I'm missing a node from Efficiency nodes v2. from jags111. The point is I do have those nodes; tried to update, reinstall... all the time I reload the workflow I receive the message missing those node types. Print
Yeah I am not using any of those nodes as far as I know.
This workflow is short, elegant and brilliant, it gives such beautiful results, I want to thank you for it and I look forward to your next workflows ! maybe you can share your amazing comic book video process? (workflow and prompts)
Thanks dud
【comfyui_controlnet_aux】Conflicted Nodes (5)
AnimalPosePreprocessor [ComfyUI-tbox]
DWPreprocessor [ComfyUI-tbox]
DensePosePreprocessor [ComfyUI-tbox]
LineArtPreprocessor [ComfyUI-tbox]
LineartStandardPreprocessor [ComfyUI-tbox]