Loading
Have a ComfyUI workflow? Get a readable Workflow Doc in your browser.Make a workflow readable.Try it now ↗
This is an Animatediff 3 workflow with a start and an end image. This is done across Sparse Control. And this workflow is made of two steps. First you create a preview video with 8 frames per seconds and in the usual...
Runtime profile
Source description
Description
This is an Animatediff 3 workflow with a start and an end image. This is done across Sparse Control. And this workflow is made of two steps. First you create a preview video with 8 frames per seconds and in the usual fitting size for Animatediff. And then you upscale it with the Supermegabadassupscaler workflow.
The result is of course not as good as from the newer AI video models. And you now need to do some further work in your video editing software. But for Animatediff it is nevertheless impressing.
Howto
Load a start and end image. Adjust prompt, size and length. Produce a preview video. Upscale this preview video in the Supermegabadassupscaler workflow to the final result and frame rate.
AI-generated commentary
AI-generated explanation based on source and configuration details. Suggestions are clearly labeled.
Use this AnimateDiff 3 workflow to create an 8 fps preview video from a start image and an end image with Sparse Control. Adjust the prompt, size, and length; the preview can then be upscaled with the Supermegabadassupscaler workflow.
The described process has two stages: create a preview at a size suited to AnimateDiff, then upscale it with the Supermegabadassupscaler workflow.
Provide a start image and an end image; the prompt, size, and video length are adjustable.
The workflow produces a preview video; the separate upscaling stage is described as producing the final result and frame rate.
The creator lists at least 12 GB of VRAM as a requirement; the base model is SD 1.5.
Listed node packs: ComfyUI-AnimateDiff-Evolved, ComfyUI-VideoHelperSuite, mikey_nodes, comfy_mtb, ComfyUI-Advanced-ControlNet, ComfyUI-KJNodes, ComfyUI_UltimateSDUpscale, ComfyUI-Frame-Interpolation, ComfyUI-Custom-Scripts, and efficiency-nodes-comfyui.
Listed model files include vae-ft-mse-840000-ema-pruned.safetensors, microsoft/Florence-2-large, rife47.pth, and ip-adapter-plus_sd15.safetensors.
The creator’s example uses a 4-second preview at 560 × 320, then upscales it to 1120 × 640.
The creator reports that the UltimateUpscale SD node often runs out of memory on an 8 GB graphics card.
The creator says the result needs further editing in video software and is not as good as newer AI video models.
Does this need to be edited?
Sign in to send an edit request.
Sources
1 sourceSource excerpts
2 excerptsSource context: 316 downloads · Type Workflows · Base model SD 1.5
Time
The preview video in the length of 4 seconds and a size of 560 x 320 took around two minutes. And upscaling then took another two and a half minutes to a size of 1120 x 640.
Requirements
The workflows requires at least 12 gb vram. It uses the UltimateUpscale SD node, which simply runs out of memory most of the times with a card with just 8 gb vram. I was never able to use it with my old card.
Requirements
analogMadness_v70.safetensors
Not resolvedCheckpoint · Unknown
Vision model · Hugging Face · microsoft/Florence-2-large
LoRA · 97.4 MB · Hugging Face · guoyww/animatediff
Checkpoint · 1.56 GB · Hugging Face · guoyww/animatediff
Checkpoint · 1.85 GB · Hugging Face · guoyww/animatediff
VAE · 319 MB · SAFETENSORS · Hugging Face · stabilityai/sd-vae-ft-mse-original
Node pack · Registry
comfyui-advanced-controlnet
FoundNode pack · Registry
comfyui-animatediff-evolved
FoundSource context: 305 downloads · Type Workflows · Base model SD 1.5
Node pack · Registry
comfyui-custom-scripts
FoundNode pack · Registry
comfyui-florence2
PossibleNode pack · Registry
Node pack · Registry
comfyui-videohelpersuite
FoundNode pack · Registry
Node pack · Registry
pr-was-node-suite-comfyui-47064894
Not resolvedNode pack · Registry
