How To Make A.I. Animations with AnimateDiff + A1111 | FULL TUTORIAL

since my last animate diff tutorial within automatic 1111 control net underwent an update that actually broke everything that I showcased in my previous tutorial now I spent days looking for a fix for this myself as well as many of you kept getting the attribute error with the IP adapter huge shout out to the Reddit user known as inma for creating a fix for this by creating a separate control net and animate diff that work cohesive ly to prevent these errors from happening so in this tutorial I'm going to go over the installation of these extensions installing the models for these extensions and I'm going to show you three ways to generate animations common errors you might face the perfect settings to use and just an overall workflow to get you ready to conquer generative AI art now without further Ado let's Jump Right In so let's get the updates to control net and animate diff so the first link in the description will bring you to the control net for animate diff now if you've never installed an extension for automatic 1111 before super easy just click on the green code button here copy the URL then in automatic 1111 click on extensions install from URL and then post the link in this little section here then click install second thing we need is the updated animate diff extension to work with our control net so that will be the second link in the description should bring you to this and again same thing just go to the code button copy paste in the URL now after you have installed these make sure you disable the original control net and the original animate diff if you have them active and enabled while you have these new ones enabled it's going to cause some issues so make sure you just uncheck the boxes and make sure the the ones that we installed are active then check for updates apply and restart UI you might even benefit from closing and reopening the command prompt that runs the automatic 11-11 user interface next we need our motion model so the third link in the description will bring you to this hugging face page here here you'll be able to find the motion models let's download this version two checkpoint it's the latest and greatest it's a pretty hefty file size let's click this download button here so now that you have that downloaded let's add that folder to the motion models folder within animate diff navigate to your stable diffusion web UI folder then go to extensions then animate diff for a control net and then just find the model folder and then just paste it in there lastly make sure that your control net has all the models that we need which is the tile model but the tile resample repr processor now they should autop populate down here once you select the tile blur option if they don't if you do not have this model the fourth Link in the Des description will bring you to this hugging face page and you can download that file right here and when you download that file go to the stable diffusion webui folder yet again go to extensions control net for animate diff models and then throw the path file in here you can see mine right there so within automatic 1111 if you open the animate diff extension and you see the motion model here and you scroll down to the control net model activate tile blur and you see the tile resample pre-process processor and the tile model we are ready to start animating the first method I'm going to show you is text to video so for this we're going to need a prompt to generate the video now I don't want to take 5 to 10 minutes to come up with a complex highquality prompt so I'm going to be using the Tyron prom generator so I want for this let's just do a woman wearing a red dress and then just click this to copy it now I'm just going to paste that here so now that we have our prompt just going to enable animate diff let's do 20 frame animation 10 frames a second it's going to be a 2 second animation now since we're just doing text to video we don't need our control net just yet so let's go ahead and generate so I'm glad this happen this is a common problem with anime diff if you are experiencing this it is because our prompt is too long you want to keep your prompt below 50 tokens and we are just above that threshold so let's go back to the prompt generator and then still do woman wearing a red dress let's do 20 words now we're at 32 tokens this should be much better if you are still having issues with the GIF changing to something completely different halfway through there's a setting that'll fix that so go to settings go to optimizations then make sure the pad prompt SL negative prompt to be the same length option is checked and this should fix the issue now the second way of animating with animate diff that I want to show you is image to video so for this I'm going to use the same prompt but I'm going to generate an image so obviously going to disable animate diff and generate so as you're most likely aware you can use this method with any image and this is where we introduce control net so enable control net Pixel Perfect drag your image into the control net here then enable animate diff so since I want the animation to still be of a woman wearing a red dress we are going to keep the prompt exactly the same so the end goal is to just animate this she's not going to morph into anything and it's not going to transition into anything different and then just make sure that you have tile Blair pre-processor and model all set to these parameters then hit generate and boom we have our animation based off of our image now there's a lot of flickering and some inconsistencies in this this is kind of typical I will show you the way I upscale this and add a lot more quality to it in this video now the last technique I'm going to show you is image to image video or image to image to video essentially so for this we're going to be using two different images so for this technique we are essentially going to transition from this image to this image here which really the only difference is the color of the dress that she's wearing we are going to animate the difference between the two images so to do this we're going to introduce pissing me off more sorry about that fruit flies brother so for this we're going to be introducing a second control net it's going to enable it Pixel Perfect tile blur the exact same settings we use on the first control net just make sure that in control Net Zero you have the image that you want to start from and then the second control net you have the image you want to end at and we're going to enable anim diff and the only thing I changed to generate this image was I changed Crimson to White so I'm going to keep that for this animation because I want the animation to end at a woman wearing a white dress and it's going to start at a woman wearing a crimson dress so with all those parameters set it's generate and I just ran into this error here I I don't really know why this error exists I don't really know how to fix it aside from just restarting the web UI completely so I'm going to do that yep and just like that completely restarting the user interface fixes the issue and now we have our animation that transitions from a red dress to a white dress now there there are a lot of issues with this animation I mean like with any kind of generative art process it takes a lot of trial and error to get the result that you're looking for so this is first try so we can also fine-tune our prompt a little bit as well I'm going to be a little more descriptive in my end prompt instead of her just transitioning into a white dress I'm going going to completely change the background just so there's more going on so let's do a woman in a white dress on the beach that's a great prompt I forgot to add a limit so let's redo that now that's actually really good again just going to copy that and if you want to get your hands on this prompt generator check the link in the description and use code YouTube 25 for 25% off there's also a 7-Day free trial included in that too so you can try it out for 7 days before you actually want to pay for it so going back into stable diffusion let paste the new prompt and really that's the only thing then just disable the control Nets actually I want to keep the woman as closely resembled to this image as possible so I'm going to use this control net but change it to reference and then disable animate diff that's close enough this should be a good example so then we're just going to load these images in the control net just re enable that and change this to tile blur then enable animate diff regenerate well I got that weird error again but restarted it reloaded everything and got this animation so this actually looks really good there's not really anything that's wildly inconsistent but this will work this this looks pretty good so now I'm going to show you how to overall improve the quality of these so I'm going to save this so this is one of my favorite AI tools Bar None you may have heard about it you may even use it but it's called topaz video AI so I'm just going to load the gift image into here and you can see again it just looks really looks really choppy so there are a number of things I could have done within automatic 1111 to get a little bit better animation such as using a detailer a negative prompt would have been very good as well but overall this doesn't look too bad so I want to upscale this and make it much smoother which we can do very easily in topaz and I'm going to show you the perfect settings for that so immediate mediately you can see that the input video is showing 512 x 512 at 10 frames per second which is exactly what this is that's what we generated it as I want it to be upscaled by two so we can either do a 2 by upscale or do a custom resolution here so 2 by upscale it's going to show you the dimensions 1024x 1024 then I want a 60 FPS frame rate make it buttery smooth then frame interpolation I use the Apollo AI model the protus AI model for AI enhancement then for stabilization reduce jittery motions will reduce the flashiness in the video sometimes it can remove some stuff in the animation that we don't want removed so use this one with caution but when used correctly it makes a huge difference on the quality motion de blur we're not going to use enhancement we're going to use the progressive style then leave everything else the same and then I like an MP4 format so with those set going to export it so right away in this side by side you can you can see the stabilization is kind of making everything a little too blurry so I'm going to disable that and regenerate and now that looks much better compared to this it's a lot more smooth a lot less flickery so you can definitely see the potential in this when you have a video that's actually high quality cuz your eyes are all up and I love this product so much I actually reached out to Topaz Labs personally to get a referral link because everyone who uses generative AI to generate animations will benefit heavily from using this software and I want to talk about it and I want to help them out so if you're interested in using this software I do have a referral Link in the description and if you use it it'll greatly help out my channel not required however if you're serious about generative art and animations you will greatly benefit from topaz video so that's the end of this video hopefully you were able to get a good understanding of animate diff using it with control net and combining it to create animations that absolutely blow your mind now I have a ton of ideas and techniques that I want to showcase not to mention the fact that new Innovations and techniques are popping up every single day in this space so if you don't want to miss out on all that make sure you subscribe and like the video if you need help have any questions or just want to talk to me personally make sure you join the Tyrant Empire link is in the description you can get access to the private Tyrant Empire Community be surrounded by like-minded individuals who are into generative AI art as well as like-minded individuals who are on a mission to conquer their lives in every aspect so with that said that is all for now till next time guys keep conquering

Channel: Ty The Tyrant

Views: 33,189

Keywords: tyrant, empire, tyrantempire, rampayj, ai, artificialintelligence, art, midjourney, stablediffusion, automatic1111, leonardoai, dalle, chatgpt, fitness, selfimprovement, selfhelp, learning, gaming, apexlegends, leagueoflegends, callofduty, dating, women, redpill, entreprenuer, freelance, freelancing, money, business, memes, streaming, tyrin, music, musicproduction, dj, howto, makemoney, america, usa, lamborghini, ferrari, bugatti, honda, funny, muscles, abs, sixpack, 6pack, diet, food, polymath, books, reading, god

Id: om2n1G74yd4

Length: 12min 46sec (766 seconds)

Published: Sun Oct 08 2023