Stable-diffusion-webui-video VS DeforumStableDiffusionLocal

Thanks! That wavy effect is kind of the default when you use this script.

This is seed travel, using https://github.com/memes-forever/Stable-diffusion-webui-video

Oh yeah! Once you installed Automatic1111 and this video script, you can set the seconds value to anything you like. Careful, the longer your result the longer it takes to calculate. (I have a pretty powerful PC and seconds turn into minutes.)

I am using Automatic1111's UI and I have previously ran https://github.com/memes-forever/Stable-diffusion-webui-video that video script just fine... but the last few days I have been having trouble getting it to finish the project... It makes the images fine, but when processing the video it fails for me...

there is also this script for the automatic1111 fork which kind of does something similar from the very simple test i did

i found this some hours ago, it works! https://github.com/memes-forever/Stable-diffusion-webui-video

I'm working on a video explaining my workflow and will share all my code when that's ready - I wrote a bunch of custom python scripts to help me make these videos. I use Stable Diffusion's V2 model and another amazing open-source extension called "Deforum Diffusion" which bundles a depth model and some helpers for generating animations https://github.com/HelixNGC7293/DeforumStableDiffusionLocal All my scripts deal with Deforum/stable diffusion, FFMPEG, subtitling, import/export from Audacity and ways to retry/tweak prompts that don't work out well.

Sorta but it's probably not what you are looking for. There's a plugin that will let you make animations from text prompts but it does not understand the video as a whole but only the current frame and the previous frame. What that means is you get a lot of videos that are just X morphing into Y instead of something with motion like a car driving across a road.

HOW TO DO IT

Nice, then you're probably looking for this

I am trying to run https://github.com/HelixNGC7293/DeforumStableDiffusionLocal on google colab with https://huggingface.co/stabilityai/stable-diffusion-2-1/blob/main/v2-1_768-ema-pruned.ckpt and v2-inference-v.yaml and I get:

For anyone else wanting to make shit like this - it's a free AI program called Deforum Stable Diffusion

Ok so I've read the HelixNG Git once more and apparently I had to download extra stuff like

You can also use Deforum which works really well and has a ton of options. I used it to make this video .