Skip to playerSkip to main content
Transform your Stable Diffusion masterpieces into breathtaking animations! In this step-by-step guide, I'll show you how to use the Animatediff extension for Automatic1111 to bring your AI art to life with mesmerizing loops and character movement.

I cover everything you need to get started, including installation from the extension tab, downloading the essential motion models, and optimizing your VRAM usage with Xformers. You'll learn the secrets behind settings like Context Batch Size, Frame Rate (FPS), and how to create perfect "Closed Loops" for natural-looking animations. I also demonstrate the advanced "Prompt Travel" technique to influence how your animation evolves over time!

Original YouTube Tutorial: https://youtu.be/2vMr8pqPXtc

Source/Model Resources:
* Animatediff Motion Models: https://huggingface.co/guoyww/animatediff/tree/main
* Animatediff automatic1111: https://github.com/continue-revolution/sd-webui-animatediff
* Animatediff Github: https://github.com/guoyww/animatediff/
* Recommended Model: mm_sd_v15_v2.ckpt
* Recommended SD Models: Toon-You (Cartoon) or Realistic Vision V2.0

Video Details:
* Original Publish Date: April 19, 2024
* Focus: Animatediff / AI Animation / Stable Diffusion / Automatic1111
* Testing Hardware: RTX 4060 Ti (16GB VRAM)

Follow lordcaocao2025 on Dailymotion for more advanced AI animation techniques, ComfyUI workflows, and technical research guides!

---
Connect with me:
📺 YouTube: https://www.youtube.com/@CaoCao2025
📱 TikTok: https://www.tiktok.com/@caocao20250
💎 Patreon: https://www.patreon.com/cw/Caocao2025

#Animatediff #StableDiffusion #AIAnimation #Automatic1111 #AITutorial #GenerativeAI #lordcaocao2025
Transcript
00:00Hello guys! Welcome back with me! Ciao Ciao 2025!
00:06Today, we're going to unleash the true potential of your imaginations with AnimatedEve.
00:13An extension that lets you transform your stable diffusion masterpieces into breathtaking animations.
00:20Get ready to add a whole new dimension to your AR journey.
00:26Imagine your AI art bursting with life, characters dancing, landscapes morphing, and stories unfolding in mesmerizing loops.
00:34This is the magic of AnimatedEve for stable diffusion in AutoEven 11.
00:39Today, we'll unleash this animation power step by step.
00:44Before we dive in, let's check compatibility. AnimatedEve currently requires Windows or Linux.
00:51Also, VRAM usage depends on your image size and context batch size.
00:57You can try to reduce image size or context batch size to reduce ROM usage.
01:02Also, use Xformer to reduce VRAM significantly.
01:16Now, let's install the AutomaticEleven AnimatedEve extension.
01:21From AutomaticEleven 11 window, go to Extension tab.
01:26Click Available, click Load From, and the search for AnimatedEve.
01:34After you've found it, click Install.
01:41After it's installed, go to Install tab and the click Apply and Restart UI.
01:53When AutomaticEleven 11 is showing up, check if you have a new menu below that is AnimatedEve.
01:59Check the menu.
02:06You will see the Motion Module is empty.
02:09We need to download the model first.
02:12This model is located at huggingface.co.goiw.
02:17Slash AnimatedEve.
02:19Slash Tree.
02:20Slash Main.
02:22I also put the link on the description.
02:25You need to download the Motion Model.
02:27I suggest you use the latest one that is MMSDV15V2.CKPT.
02:35There are also a bunch of LoRa here, but for this video, we will ignore it for now.
02:40Save the file in uwebui slash extensions slash sdwebuianimatedif slash model.
02:47While we wait, I will explain the function of the setting of the animated div.
02:53Save format is what you want the format of your output.
02:56You need to choose at least one, but you could choose more.
02:59Check txt if you want info text, which will live in the same directory as the output gif.
03:05Number of frames.
03:06Choose how many frame do you want.
03:08If you enter zero, the number of frames will be your context batch size described below.
03:13If you enter something smaller than your context batch size other than zero, you will get the first number of
03:19frames frames as your output gif from your whole generation.
03:23All following frames will not appear in your generated gif, but will be saved as PNG as usual.
03:28Do not set number of frames to be something smaller than context batch size other than zero.
03:34FPS's frames per second, which is how many frames images, are shown every second.
03:40If sixteen frames are generated at eight frames per second, your gif's duration is two seconds.
03:47Display loop number.
03:50This decide how many times the GIF is looped.
03:54A value of zero means the GIF never stops playing.
03:59Context batch size is how many frames will be passed into the motion module at once.
04:05The SDUN 0.5 motion modules are trained with sixteen frames, so it will give the best results when the
04:11number of frames is set to sixteen.
04:14Closed loop.
04:15Decide to make the last frame the same as the first frame.
04:19Option means absolutely no closed loop.
04:22Minus P means that the extension will try to reduce the number of closed loop context.
04:28R plus P means that the extension will try to reduce the number of closed loop context.
04:34A means that the extension will aggressively try to make the last frame the same as the first frame.
04:41The prompt travel will be interpolated to be a closed loop.
04:46Stride mean max motion stride as a power of two.
04:51Overlap as number of frames to overlap in context.
04:56Frame interpolation.
04:57Interpolate between frames.
05:00Video source and video path will be talking later video.
05:04You need control net to do this.
05:07Move a motion video to keep you, just don't use this.
05:10It gonna make the process really slow.
05:16How to use animated diff.
05:18For testing, I suggest you use Tunyu Beta 6 model for cartoon model or realistic division 2 for realistic picture.
05:28Let's make an example using Tunyu Beta 6 model.
05:32What you need to do first is to make sure you check the enable animated diff.
05:38Let's leave everything at default for now and make a prompt.
05:45For negative prompt, let's use Worst Quality, Low Quality, Letterboxd.
05:50For positive prompt, let's try Flying Bird, Blue Scott.
05:55For sample method, I use DPM Plus, Plus A Kara.
06:01The other is a default value, click Generate.
06:05It's going to take a while.
06:07Let's wait.
06:09Okay, we'll make a grid something like a sprite in two games.
06:14Now, you will see the picture is moving.
06:17Generate for 1 minutes 24 seconds for this GIF file.
06:22But as you can see, the loop is not natural because the bird in the last frame suddenly jumped to
06:27the front.
06:28If you want it to be a closed loop, just select the loop to A and then we use the
06:33same seed.
06:35Now, let's click Generate.
06:37Now, as you can see, the bird moving in closed loop.
06:42Of course, with very low frame.
06:44This generate a very strange animation.
06:49If you don't want close loop, simply click now.
06:52If you want longer animation, you could manually put number of frames manually.
06:58You should put in above context and it should be able to be divided by your FPs.
07:04So, I put 48 for FPs 8.
07:08It means it is going to be a 6 second animation.
07:12Of course, longer animation means longer processing time.
07:17Okay, that's the result.
07:19It is pretty long, but nothing special.
07:21So, let's make another example.
07:31This is another example.
07:33If we draw characters, the water is moving great.
07:37For now, we can't control the flow of the animation.
07:41But we can actually use prompt traveling to somehow influence the animation.
07:47You could put a frame number, then column.
07:51After that, put your first prompt there.
07:54Then you press enter and add a new frame number with column and with prompt.
08:00You could repeat this as many times as you want.
08:04As long as the frames exist.
08:07Here, in this example, I put a girl walking for the first 0-15 frame.
08:14And make her running for the 16-31 frame and make her flying in the rest of the frame.
08:20The result is not that good, but that just example.
08:23Also, you could use URI's fix for the image if you want.
08:26I'm not going to use it for now since it takes a long time just to process the regular image.
08:31This example, I use Realism 50 model.
08:34It looking great.
08:51You could use it for the first 0-15 frame.
08:52Okay, guys.
08:53This is another example.
09:08All right, guys.
09:10For now, I end this video.
09:12There are many tricks that we could use with Animated-IF.
09:16But we could talk about it on the next video.
09:18Thanks for watching and have a nice day.
Comments

Recommended