TXT2VID - Simple SD1.5 AnimateDiff Workflow
created a year ago
lora
animatediff
video
upscale
19 nodes

4.6k

1.1k

Outputs
example video output 3 for workflow.
Description

Globby J's TXT2VID Simple SD1.5 AnimateDiff Workflow

DESCRIPTION:

This is a very simple workflow designed for use with SD 1.5 and AnimateDiff in order to produce short text to video (gif/mp4/etc...) results.

Default configuration of this workflow produces a short gif/mp4 (just over 3 seconds) with fairly good temporal consistencies with the right prompts. It excels at producing portrait clips with subtle movement, as shared on this page.

If you want a way to upscale these videos without modifying the workflow to upscale within comfy, It may save more time to use a tool like Video2X. You can find the link here:
https://github.com/k4yt3x/video2x

RENDER TIME:

I have a 3080Ti. Videos just over 3 seconds take about 110 seconds for me to produce with this workflow. This is close to the amount of time it takes my GPU to produce an image with SDXL and run it through my refining/2x Upscale process. It is surprisingly fast.

CHANGELOG:

v3 - Added an upscale and a bypasser switch to go along with it. Easily capable of crisp 1024x1024 outputs.

Enjoy!

Built-in nodes
Custom nodes
VAELoader
LoraLoader
ADE_EmptyLatentImageLarge
ADE_AnimateDiffUniformContextOptions
ADE_AnimateDiffLoaderWithContext
ADE_AnimateDiffLoRALoader
KSampler
CLIPSetLastLayer
VAEDecode
VHS_VideoCombine
BNK_CLIPTextEncodeAdvanced
FILM VFI
CheckpointLoaderSimple
UpscaleModelLoader
UltimateSDUpscale
3
0
0
0