Mickmumpitz
patreon
[NEW VIDEO & FREE WORKFLOW] AI COMPOSITING V01
🕑 Added 2024-08-09 13:34:26 +0000 UTCAI is the future of VFX! In this deep dive, I'll show you a free workflow for compositing any video on any background for your movie projects.
Easily composite any video on any background with one click! I developed a free AI-powered ComfyUI workflow to automatically cut out the subject, match lighting, fix edges, and add additional VFX. Just import video and background, wait a few minutes, and get your final shot.
Fine-tune and control the result further with traditional VFX tools like Blender and After Effects, and other mind-blowing AI tools. Whether you're creating a music video or sci-fi short, this workflow can transform the way you make movies!
Download the FREE version of this workflow below.
You can find the INSTALLATION GUIDE here: https://docs.google.com/document/d/1jErR3WXj-iLhyQuA4XBPmvvRxo_MQ8x6oU7twTSzOhE/edit?usp=sharing
You can find the exclusive ADVANCED workflow here: [Already uploaded, Link coming soon]
This is the first version of the workflow and I have lots of ideas about what I want to add.
So please let me know what features you would like to see or if you have any ideas!
Recommended Edge Fix / BG Fix Settings
For Edge Fix, light Diffusion:
Use this to keep the subject and background image as close as possible to the original and only correct the edges and reduce flickering.
steps: 6
cfg: 1.5-2
sampler: dpmpp2m
scheduler: sgm_uniform
denoise: 0.65
Animate Diff Model: v3_sd15_mm
ControlNet: v11f1e_sd15_tile
strength: 0.65
For BG Fix, higher Diffusion:
This allows you to integrate your subject more seamlessly into the image. This changes the style and makes it more AI-like. But it can look really cool! Use the BG mask to apply it to the background only, or use the Reactor Face Swap to correct your face if your subject looks too different.
steps: 8
cfg: 1.5-2
sampler: lcm
scheduler: sgm_uniform
denoise: 0.65
Animate Diff Model: AnimateLCM_sd15_t2v
ControlNet: v11f1e_sd15_tile
strength: 0.45-0.8 (Lower values will give your shot a more abstract AI look)
![[NEW VIDEO & FREE WORKFLOW] AI COMPOSITING V01](https://img5.patrefans.com/storage/6/kw/fu/563145-019e8569-dafd-7720-9f4e-53e615eeaf36.jpg)
Comments
Lucas Guerois
i'm not running all with mac m1 do you know by what i need to replace cuda controlnet and animatediff ?
James DeFina
where is the advanced workflow??
Mario Hernandez Padilla
As a Videographer and content creator this is a great tool. Any way to update with sdxl models?
Denis
Thank you Scott, it was not shown by default, I had to search for it but found it thanks to your advice! :-)
Scott Ownbey
Denis I just discovered it had been posted directly above this post whole time.
Denis
Hey Mickmumpitz, can you please make the advanced workflow available? π You wrote the workflow is already uploaded, and the link is coming soon, but it's already been a while. π Thanks / Danke!
Soul Merge
Resolved it by replacing MaskBlur+ Node with a fresh one and assigned a new value.
Soul Merge
Hi everyone. Any ideas how to solve: Error occurred when executing MaskBlur+: sigma values should be positive and of the form (min, max). Got [0.0, 0.0]
Josua Santosa Tjenra
Hi broth i got this message -import failed was node suite -import failed Reactor Node for comfyUI -import failed ComfyUI-Image Filter do you know how to fixed ?
Ranska_ X
Hi, I got this message IC-Light: Could not patch calculate_weight - IC-Light: The 'calculate_weight' function does not exist in 'lora' what should I do ? [ok update comfyUI fix it] thanks for this workflow :)
Jared Saunders
going to give it. a go now, ill let you know in discord if it fixes it
Terry Astrial
how do i fit the foreground video image (eg. person at 9:16) to be proportionate and centered in the background video image (16:9)?
Mick Mahler
https://github.com/storyicon/comfyui_segment_anything/issues/83 I just found this relatively new thread about this. Maybe you have these conflicting nodes installed?
Jared Saunders
having an issue with this workflow, the GroundingDinoSegment module is throwing the error "Sam' object has no attribute 'image_size', not sure why, any help?
Zs Studio
hello, i can't see any image preview when I choose the path to image sequences and BG video
Daria Yudina
Hi Mick, I would like to try what goes out on the XL models, but can't find the model anywhere IC-Light XL model. Can you tell me if it even exists?
John Nguyen
Good headline...it really is insane. Soooo good!
