NEW Stable Video Diffusion XT 1.1: Image2Video
Vložit
- čas přidán 22. 05. 2024
- 🎨 Generative AI Art Playground: www.pixeldojo.ai/?...
✨ Support my work on Patreon: / allyourtech
💻My Stable Diffusion PC: kit.co/AllYourTech/stable-dif...
Model download: huggingface.co/stabilityai/st...
ComfyUI Workflow: github.com/purzbeats/svd-comp...
Stability AI recently released Stable Video Diffusion XT 1.1 which can be used to animate any image. In this video I'm going to show you how to download, install, and use this on your own PC.
Stable Video Diffusion (SVD) 1.1 Image-to-Video is a diffusion model that takes in a still image as a conditioning frame, and generates a video from it. - Věda a technologie
🎨 Generative AI Art Playground: www.pixeldojo.ai/?
✨ Support my work on Patreon: www.patreon.com/allyourtech
💻My Stable Diffusion PC: kit.co/AllYourTech/stable-diffusion-build
Great video. Where did the SVD-Comparison-Grid.json file come from?
Thank you so much for this detailed and well explained tutorial. It has been of tremendous help! I have a question. I don't know if I may have missed that detail from the tutorial, but, how do you export the generated video animation as an mp4 file to your computer?
Hi, thank you for helping out newbies like me!!! I am almost done with the installation and everything seems ok and running, but when I run it, the 5 big nodes from "final grid assembly" are in red and when it finishes I have an error within comfyUI, no result. Any idea what could I do to find a solution and getting it running?
You might be missing some nodes. Do you have comfy manager installed? If not grab that and make sure you have any missing nodes installed
@@allyourtechai no missing nodes according to comfy manager, it all looks normal there. I copy/paste and screenshot the error, I put it in my drive if you would be so kind to help me out to find the issue... Thank you so much!!! drive.google.com/drive/folders/1_WARo5zwIaloyc3wdi9Q58dS_wpGot3V?usp=sharing
you need to follow the dimensions it give initiaslly. resize your photos with birme
Love you sir
downloaded all models, no red boxes in Comfy, but bunch of errors once the Ksampler begins...any ideas about why?
Bro comfy has too many settings for me
You did something wrong?
Truly appreciate all the time and effort you spend on your highly educational videos Mr. Cruise.
Keep up all the great work brother and stay blessed ~ 🤍🙏🏼🪽
Haha, thank you my friend!
I presume that it's not good at photographs yet. I got some very creepy results!
Hit and miss for sure!
Use animatediff you can get almost photographic quality up to 120 frames on a 24GB VRAM GPU
Can i install it inside Automatic 1111?
I have only seen it used with comfy
I made a video in which I told how to create my own image with stable diffusion. Please help me or make my model for me and I will give you your photo.
do you have brain ? 🧠
@rubbish9231 😂 he watched a tutorial then demanded someone do all the work
stable diffusion is obsolete by now, they are behind the competition in every aspect, and I say that as stable diffusion user ever since model 1.0
The only redeeming factor of SD is that it's open source, free and without censorship, which is a lot yes, but other than that, SDXL is a huge disappointment and SVD is pretty lame.
Thanks for the video regardless!
Nah, ive seen SD thats way better than almost all other AI generations. Midjourney quality is better but control and versatility, SD is far better.
@amorgan5844 sdxl controlnet is bad, depth is OK, but if you use openpose ot ipadapter on sdxl it's horrible compared to 1.5. All the sdxl openpose models are completely inaccurate, no matter hard you try, clipskip, whatever, none of them are as accurate as the 1.5. So here we are, having 1.5 with great control but not as good prompt accuracy, but on the other hand we have sdxl with better prompt accuracy but horrible control. To me that is one step forward two steps backwards. Don't believe me? Try it for yourself with openpose on sdxl compared to 1.5. Sdxl is a huge disappointment
I just installed and had a go but I am struggling to get real movement of feet etc - this is a video my 10 year old daughter made with modelscope/zeroscope on 448 x 256 - its trained on 30 frames - czcams.com/video/E5Ot03UXDoY/video.html