Check out our quickstart Guide! https://education.civitai.com/quickstart-guide-to-stable-video-diffusion/
The base img2vid model was trained to generate 14 frames at 1024x576, uses less VRAM than the...
img2vid-xt model, trained to generate 25 frames at 1024x576.
img2vid-xt-1.1, the latest version, is finetuned to provide enhanced outputs for the following settings;
Width: 1024
Height: 576
Frames: 25
Motion Bucket ID: 127
FPS: 6
Augmentation Level: 0.00
Stable Video Diffusion (SVD) Image-to-Video is a diffusion model that takes in a still image as a conditioning frame, and generates a video from it. Developed by: Stability AI
Description
Finetuned from the previous version, specifically for Motion Bucket 127 outputs.
FAQ
Comments (13)
Vladmandic SD Next says it's compatible, but how to use it?
For those who's asking about A1111 and how to use, there's a fork of A1111 called "Forge" which can use this model. "illyasviel/stable-diffusion-webui-forge"
We all know A1111 is a bit slow on the updates, but support will come eventually. If there's an extension for A1111 idk, I just know the fork "Forge" works.
need more movement
just curious, why xt-v1.1 is much smaller than xt-v1?
Does anyone know how to set the video length to be longer than 4 seconds? Everything works on Forge, but the video is only 4 seconds long.
would be nice to have more info for xt-1.1 settings
what are recommended min-max CFG? steps? S&S combination? 1.0-2.9, 25, euler karras?
Can it do text to video too?
And can it generate other aspect ratios?
for those searching: latest version of forge doesn't have svd anymore, but i've cobbled together a few files from previous versions that can get it working again. Just extract and paste these into the webui directory after installing forge, and replace files when prompted: https://www.mediafire.com/file/72qt13kerg2h8f5
link is broken :(
Check the augmentation level difference between 0 and 0.15
https://youtu.be/dZr3NG4FWg4?si=6IRSZ2IlLIvn3P9H
And different scheduler selection:
I have 4GB of VRAM. It was not so fast as the speed of light, but finally I got the result 🤣
using this on reForge and I cannot get any motion out of this. Just a static image that flickers a bit, or the character slides a little with no other movement. No idea how I'm supposed to get motion out of this.
There seems to be an extreme lack of any kind of help going on. I match what few settings are recommended in the description, and all I get is heavily distorted images. How does this work? Do you include motion in the prompts? What sample steps? What cfg? I can't find any additional information on this.
Details
Files
stableVideoDiffusion_img2vidXt11.safetensors
Mirrors
stableVideoDiffusion_img2vidXt11.safetensors
svd_xt_1_1.safetensors
svd_xt_1_1.safetensors
svd_xt_1_1.safetensors
model.safetensors
svd_xt_1_1.safetensors
svd_xt_1_1.safetensors
svd_xt_1_1.safetensors
svd_xt_1_1.safetensors
svd_xt_1_1.safetensors
svd_xt_1_1.safetensors
svd_xt_1_1.safetensors
svd_xt_1_1.safetensors
svd_xt_1_1.safetensors
svd_xt_1_1.safetensors
svd_xt_1_1.safetensors
svd_xt_1_1.safetensors
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.