QUICK & EASY WAY To Create Smooth Video to Video Animated Morph In Stable Diffusion

hi what's up everyone in this video I will show you an easy way to create an animated morph video using a videoo video technique like this I will show you an easy way with quick shortcuts so you can do it well okay let's get to tutorial here we will use comy UI as the tool then there are several components that we have to install into comfy UI I have explained this method in the previous video you can see the step-by-step installation of custom nodes components and the comfy UI extension there I have included the link in the description of this video but here I have prepared a portable comfy UI which you can download easily it contains complete extensions and custom nodes that you can use straight away I have included the download link in the description of this video I broke the file into 16 Parts with each measuring 700 MB after downloading everything extract using 7z program inside there is a password the password of which I have written in the description of this video after extracting you can follow the instructions for use inside don't forget to download the checkpoint model that you want to use then copy and paste it in the comfy UI models checkpoint folder also download the control net model at hugging face which I have also included in the video description then copy and paste it in the comfy UI models control net folder after that you can immediately run comfy UI via run Nvidia GPU or run CPU dobat which is there later we will enter comfy UI like this here you need to download the workflow in the form of Json files so you can easily use comfy UI without the need for complicated settings download the Json files via the link I included in the description of this video after downloading drag and drop the workflow Json files into comfy UI like this then later you will see the interface that I have adjusted like this I have set this workflow so that you can use comfy UI quickly and easily everything we need to create animations with animated diff in comfy UI can be used easily next you need to prepare a short video which we will use as the source for the animation you can download it on Tik Tok YouTube and so on here I have several short videos that I will use as video sources back to the KFI UI interface in the load video section we select choose file to upload then select the video you want to use in the frame load cap Force rate and so on you can adjust it to what you want or you can also set it by default here I set the frame load cap to 250 this means that the frame that will be rendered later is 250 frames long then I left the other settings at default next in the width and height section we set it according to the size of the image from the video here I set it to 512 to 904 then in the load checkpoint section we select the model we want here I use the aniverse model in the load vae section adjust it to the vae model you are using here I use the Emma pruned vae model next in the prompt section here I use the travel prompt because the animation we will create is a morph animation you can change the prompt in the quotation marks and the time according to your preferences the numbers 0 15 30 and 45 indicate the frames where the prompt will change according to what we enter then next to the prompt in quotation marks is the prompt that we want to apply to the object in the video in the max frame section we set the same as the frame frame load cap here I set it to 250 then in the pre-text section we can fill in prompts such as Masterpiece best quality one girl and so on then in the negative prompt section we can fill in embeddings or objects that we don't want to appear in the video later of course you already understand this if you have used automatic 1111 before then we go to the animated if loader section in the model section we select mmsd 15 model [Music] then in the sampler section we can set the sample as we set the sample on the automatic 110 just fill in what you want here I enter steps 20 then CFG scale 8.5 like this then I set the sampler name to UL ancestral with a normal scheduler so that the rendering process is faster Just Adjust these settings according to what you want then in the control net tab here I use two control net pre-processor models the first is depth control net and the second is the DW pose control net pre-processor you can set the strength of each control net or just leave it at default in the apply control net section you can set the numbers between 0.75 and one like this but I recommend that you set the strength to one so that the results are more imaginative the last one is in the animated if combine section set the frame rate according to your preference here I set the frame rate to 16 then I set the output format to video h264 MP4 like this up to here everything is ready next we click Q prompt and let the process run until it finishes the length of the generation process will really depend on the specifications of the PC we are using on a graphic card with 8 GB vram it can still be used to generate videos with animated if however an Nvidia RTX graphic card with 12 GB vram and above is recommended so that the process is faster and more significant you can see the PC specifications that I use in the description of this video okay after the process is complete this is the result you can open it in the comfy UI output folder folder because the file is in MP4 format we can play the video with the media player on Windows like this the resulting video is quite cool but because the FPS is only 16 FPS the video doesn't look smooth for this reason we will do upscaling on the video output here we use topaz video AI the method is very easy we drag and drop the video into topaz like this then here there are several presets that we can choose but I don't use that I set the output to 2x upscale then I increased the fps to 60 FPS like this when the settings are ready now we export and wait for the results and here we go here are the results we can see a comparison of the video before upscale and after upscale this is very awesome and cool so what are you waiting for try it now that's it for our video this time I hope it's useful and you like it bye and see you next video

Channel: SahiNama

Views: 5,413

Keywords: stable diffusion, stable diffusion tutorial, ai art, workflow creation, comfy ui, animate diff, control net, video helpers, animation, stable diffusion img2img, stable diffusion prompt guide, ai art generator, ai art controversy, ai, install stable diffusion tutorial, flicker free tutorial, smooth animation, stable diffusion animation tutorial

Id: pdRw5bTalV4

Length: 7min 27sec (447 seconds)

Published: Mon Oct 30 2023