This Might be The Next AI Animation Trend | IPIV’s Morph img2vid AnimateDiff Tutorial

[Music] in today's tutorial I'm going to guide you step by step on how to create stunning morphing animations like these using comfyi we'll go through the entire process and I will share with you the specific settings that I use to take my animations to the next level and as a bonus I will show you how to significantly boost your generation speed if you're new to com UI and not sure of where to begin you can pause this video and watch my dedicated tutorial on how to install comy UI once you've done that launch comy UI you will initially see the default workflow but we're going to use a different one that you can download from this page the link can be found below and I would like to thank the Creator ipiv for his exceptional work on this workflow click here to download the workflow and to import it simply drag and drop the Json file into your comi interface now I understand this might seem overwhelming if you're a first-time user but the process becomes much easier once you understand the purpose of the nodes and the settings which I will go over soon the creator has been kind enough to include instructions inside these Yellow Boxes which I found very helpful if you're importing this workflow for the first time you might notice a bunch of red boxes this is completely normal it simply means that the workflow uses certain nodes that aren't yet on your system the easiest solution is to open the manager go to install missing custom nodes install the missing nodes one by one restart confy Y and the red boxes should now be replaced with the newly installed nodes another crucial thing to remember is that every new workflow you import will likely require the use of some AI models that aren't on your computer yet to make this easier I've added a list of all models how to get them and where to save them in the description box so make sure to pause and complete that step first once you've downloaded all the necessary files don't forget to click on refresh now I'm going to show you several settings that can significantly improve your output first let's change the motion scale value the higher it is the more movement you will have in your video I recommend setting this between 1 and two and for this example I will set it to 1.6 next we have the strength of our animate deflora increasing this enhances consistency between frames and reduces variations and flicker this could be good or bad depending on your personal preference I usually set mine between 0.4 and 1 and for this example I will go 0.8 over here you can load a checkpoint of your choice this will directly affect the visual style of your output so feel free to experiment with different models and for this video I use the Juggernaut model by default the video resolution is set to 288x 512 I highly recommend not changing this unless you want a horizontal video in that case you can swap the width and height I'm going to stick to vertical iCal video and although this is a very low resolution we can upscale it later to get a crisper and much sharper video the low resolution helps us process the video at a much higher speed and I have even more tips on how to get really fast Generations so make sure you follow through till the end you know when it comes to crafting AI animations like these there is a secret ingredient that often gets overlooked sound design I'll be honest my own sound design skills needed a bit of a tuneup that's when I found this class by Duan bovich on skillshare watching this I learned how to find the right sound effects and layer them together to tell a more compelling story what's great is that this class caters to everyone especially beginners and it's just one tiny fraction of what skillshare has to offer skillshare is a community of Learners and creators with top-notch classes covering a variety of topics such as animation photography and more the classes are tailored to different experience levels so if you're already familiar with the basic of sound but want I dive deeper into say audio editing in premere pro skillshare has got you covered now for the first 500 viewers who sign up through the link below skillshare is offering you free access to the entire class library for a full month now moving on to the next node I suggest setting this to randomize and I'm going to make changes to this node at a later stage now instead of using prompts to describe the animation we are going to use images as our input and that's exactly why we have four four different load image nodes right here simply click here to upload an image I've used mid journey to generate my images if you want to do the same it's crucial that you know how to write the right prompt and I have the right video on that after you finish importing the remaining images you can proceed to the IP adapter nodes in this section you can choose from different presets to control how closely the video should resemble your images a lower strength means the output video will look entirely different for this example I'm going to set this to high strength I've attempted to use photos of real people but unfortunately I haven't been successful at maintaining the same facial features I plan to experiment more with this and if it works I will make a tutorial on it so make sure to subscribe below that you will find your IP adapter Advanced nodes that you can use to control transformation strength on individual images and I'm going to set it to one for this example the weight type is set to Linear by default which works perfectly fine but I've been getting more Dynamic results using the EAS in and out setting now let's go down to the control net nodes the first thing we'll look at is the load video node as you can see in this box there's a direct link to a video this is a black and white animation that will guide the motion structure there are a bunch of other options that you can go with which I'll leave down in the description box to load a different animation simply right click on it choose copy video address and paste the link over here I prefer the default animation so I'm going to revert back to that on the right side you will find a setting for control net strength by increasing this you can make the motion follow the structure of the control net animation or you can reduce it for a more subtle effect I don't like this to be too high neither too low so let's go with 0.35 moving on to the right we have the first G sampler node when it comes to generating this node will basically be doing the heavy lifting steps controls the overall quality of your output the higher the more details you're going to have but it will also res result in slower processing I like to set mine around 20 make sure you do not change the CFG value and leave it set to one both sampler and scheduler work handin hand the default options are completely fine but I found that uler and normal tend to give slightly brighter and sharper results if you have time to experiment with different combinations here I highly recommend you do so the second K sampler node you see here is meant for the upscaling stage I'm going to rename it real quick and there are a few things to change here but before that let's go over to the video combined node and guys if you want even better quality go ahead and change the CRF to a lower value but keep in mind that this will increase the size of your video file I personally prefer to have better quality at the cost of that so I'm going to go ahead and set my CRF to five in all the video combined notes here you can choose an upscaling ratio if you want to upscale to 720p set the scale by to 2.5 if you want to upscale to 1080p set this to 3.75 which is exactly what I'm going to do next let's focus on the upscaling cas sampler set the steps to 20 and match the sampler name and scheduler to those of the first Cas sampler you can also increase the dno value which will drastically improve the upscaling quality but it will also alter the visuals after upscaling if you want no visual changes you can set this to zero but be aware that the upscaling quality will be very four I found that 0.5 is a good compromise you can go higher but I recommend not exceeding 0.8 and as I mentioned earlier I will set the upscaling resolution to 1080x 1920 you might notice that in the first video combined node the video frame rate is set to a low 12 this is intentionally done to reduce processing time but thanks to the interpolation node we can double the frame rate for a smoother video the color correction node is a great addition if you want to tweak the final look of your video however I prefer to handle this in Premier Pro so I will revert all the values back to their default settings the last video combined node will output the final video the frame rate is set to 24 which is perfectly fine but I personally prefer to work with 25 FPS those were the most important settings that you need to change I'll provide a link to this and other workflows in the description box for you to try at this stage you could actually start generating video but given the number of different nodes especially those related to upscaling and interpolation the process may take some time before you achieve a satisfying output however the good news is that you can preview the animation and only proceed to upscaling once you're satisfied with the result this can be done by temporarily disabling certain nodes including these four nodes right here to disable them simply press crl M and do the same thing for these nodes over here as well this really useful trick was shared by another YouTuber named Abe and I'm really grateful for him for sharing this tip once you've done that you can hit Q prompt to start generating given the low resolution at which we're generating this process should run fairly quickly even on machines with lower vram you can then preview the animation under the first video combined node if the output doesn't meet your expectations feel free to go back and adjust settings change images or experiment with different settings and then generate a new preview remember just like any other AI tool that I've covered on my channel the settings you choose will greatly depend on the input there's no magic formula or a single setting that will give you the perfect output the nodes work in combination and you may need to adjust several settings before you get any good results if you're passionate and have the time I highly recommend experimenting once you're satisfied with the result reenable the upscaling nodes go back and change the C type to fixed and hit Q prompt the duration of this whole process will depend on your hardware and settings but once it's over you will get an upscaled version of your preview that should look a 100 times better to access the video files open comi and find the output folder you will find subfolders organized by date each containing all the videos that you've generated and your final outputs can be found in the interpolated folder if you create something using this method don't forget to tag me on Instagram keep an eye on the pinned comment under this video for additional tips and updates if you're interested in learning more creative uses of com VII check out the other videos on my channel stay creative and I'll see you in the next video peace

Channel: MDMZ

Views: 29,109

Keywords: ipivs morph, ipivs morph - img2vid animatediff lcm, morphing transition effect, morphing animation, morph face tutorial, comfyui animatediff, animatediff tutorial, ipadapter tutorial, prompt travel comfyui, prompt travel animatediff, prompt travel animation, how to use animatediff, comfyui tutorial, ai animation tutorial, generate ai animation, generate ai video, ai video tutorial, stable diffusion tutorial, ai animation, ai video generator, animatediff comfyui

Id: AUKQfJZX-78

Length: 11min 18sec (678 seconds)

Published: Fri May 10 2024