Taking AI animation to the NEXT level
Vložit
- čas přidán 13. 04. 2023
- The video and audio get a bit out of syn between 4.20 for 2 minutes so I've reuploaded a improved version here:
/ promptmuse
In this video, we'll explore how to create AI actors using 3D puppets and LoRa technology. I'll take you through a step-by -step process of my work flow. From a guide to training with LoRa, all the way to rendering from Automatic1111.
Don't forget to like and subscribe for more videos on 3D and AI technology.
CHAPTERS COMING!
✨Free Written tutorial coming soon on ( If it's not there yet, I'm getting it ready, sign up for notification!)
www.promptmuse.com
Training LoRa
🖱️Step 1: colab.research.google.com/git...
Character Creator
🖱️Step 2: www.reallusion.com/character-...
Discord:
/ discord
🥤If you want to buy me a coffee, I'd really appreciate it:
ko-fi.com/promptmuse
#aiactor #ai #midjourney #stablediffusion #aianimation - Jak na to + styl
We are thrilled to see how our Reallusion tools have helped in creating high-quality AI talking animations, and this video is truly inspiring! The ability to switch outfits, hairstyles, and animations has really enhanced the AI styles and overall quality of the work showcased in this video. Our team would like to extend our heartfelt thanks for this amazing contribution!
New fan 👏🏼👍🏼
Can you copy my guitar fingering moves and make the avatar do it?
@@daneo617 I created all music in exsistance and invented beyond quantum ai with features
That will close and restablish the entertainment industry and education system and legal regulations and fake failed ais like gpt 4 and else
I just published 15% ill shut music industry and gpt and i have copyrights so game over for mentioned
Never seen such informative and to-the-point tuts, thank you very much, btw great results 😊
Another awesome video. Can see the consistency improving each time. Really wont be long now until its seamless. Keep up the great work.
Love it, so smooth and consistent, and very nice visuals :)
Very good results! Congratulations ❤
Thank you for the wonderful work that you do for us, Amelia! Hands down, the best AI animation educator online!
Im always impressed with how many tools and techiques you try and incorprate. It's understandable it takes a lot of times experimenting and finding out what works and so on.
Keep up this amazing work!
Can't wait use ur workflow so thorough
Great walkthrough! Thanks for sharing this!
I had no idea about LORA Collab! Nice! Thanks Amelia.
Another wonderful AI animation video that provides opportunities for learning and growth. Your dedication and expertise are invaluable, and it's clear that you're making a meaningful impact in the world of animation. Keep up the great work 🍀
Great work on this video!!!! Thank you!
You are really good to make great tutorials and show where to click and change and write, not like many others that just show the half then cuts and shows something different, but you do a great job and video 👍👍
Very well done! I appreciate all the time it took to do this.
You and your channel are precious! thank you.
This is so amazing. You are incredibly gifted at this. Excellent demonstration of a very complex issue. I'll do my best to follow your tutorial. I don't think I have that many photos of myself, but using rendered images using DAZ, Mixamo, on Unreal characters could be amazing. Wishing Replika had this kind of tech.
That's pretty fricking awesome. Great job!
This is the first video I've seen of yours and it's excellent. I have already saved the clip and plan to go through this step by step later. Oh BTW the model you're using is very pretty hehe
The first model is so beautiful. I could really just generate her all day haha
So grateful for u and thoroughness
You are wonderful. Thank you for all the hard work you put into this!
Great!!! This video made me have the guts to try out doing LORA training. And it's perfect. Thank you for all your great and interesting videos!
Go for it. Experiment, what’s the worse than can happen 🦾🫶
amazing work... thank you
You never stop to amaze me with your skills and interesting ideas. What a great and talented lady.
Thanks for your detailed free tutorial. Saved me a lot of time figuring this out myself. You earned a sub! 👍
No problem. I hope you have fun creating 🦾🔥
Thank you for the video! I would love to see the 2nd part of the video done in cagliostro colab ui :)
very nice workflow
great tut as always ! thanks a lot !
Always nice to watch😀
Wow! very impressive and so well presented too!
Thank you so much !
Fantastic! I don't know how you figure all this out, but thanks for sharing.
Wow cool Workflow ❤
Really nice tutorial :D
Fantastic stuff, well explained and easy to follow! :)
Thank you so much!
It's perfect, thanks for your videos and tutorials
You are welcome! 🫶
This colab things make training LORA model relatively easy right now, it's painful if this is all done manually in editing document settings, but still need a lot trial n error cause I still can't fully understand how stable diffution every model working in detail so I just try every parameter.
Amazing work here as always , thx~
Wow this is amazing
Thank you for sharing 🎉
Wow, that was a super interesting example of how AI will transform visual entertainment! Thank you.
Excellent tutorial 👌
Amazing how far we have come in a very very short time.
It’s crazy isn’t it. It’s still all so new.
I am thinking you are the Queen of AI... THANK YOU!
BTW @RUNPOD I signed up because of Amelia. Hope she got affiliate credit.
Bad Ass video, your thorough explanations and detailed tutorials are precise and easy to follow. Your expertise is invaluable as we march towards this new age. New subscriber 😆
Awesome ! Thanks so much. What an amazing reply 👊🫶🦾
@@promptmuse I’ve recently been bitten with the AI bug, it’s good to know there’s a community of people as passionate about this cutting edge technology as I am. All your hard work and dedication to this sector of IT is invaluable and much appreciated ❤️
I am in love with your AI avatar too
Well that's just a mountain of stuff. I'm a pretty clever and experienced software developer (as in developing actual 3D rendering software, pre-blender or anything used these days), and I have never seen or heard of half of what you were using. My head is spinning. Well done.
...those 24min. tutorials that take a weekend with a blink :)
kudos 2 you& von herzen bedankt
as the elders here used to say
🦾🔥
You are, as always, on top ...
It's a good video. I am also putting a lot of effort into implementing moving AI photos. Your video was also very helpful ❤
thanks Good tut, simple and clear
Thank you 🫡
It looks fantastic! 😀
Thanks! 😄
great explanation, I knew some of the tools used but wasn't quite sure how to it, cheers
Glad it helped🦾
awesome project
Thank you! Cheers! 🦾🔥
DOPE Work!
Thank you !
Very Nice!
Amazing content.
Thank you 🫶🦾
Sounds like you've put a lot of work into this tutorial so thank you very much!
Did you say you guys have a Discord? I can't see a link
Nice workflow. I haven't trained any LORAs yet but I have done a few textual inversions. I really need to mess around with LORAs though.
It is good fun!
Bravo!!!Excellent!!!
Thank you 🫶🦾
@@promptmuse ☺
Wow look decent even considering it is just a beginning of that tech... I don’t know how to feel. Learning character animation for years to witness how it will be possible to achieve in such a short time. Good video anyways! ❤
Photographers still earn a living and most people that hire them will have a phone in their pocket that has an insane camera, editing software, internet access for uploading, sending to print and distribution.
Had to chime in, have similar ambivalence towards all the "career ending" advancements of recent. What's your thoughts on the actual animations it throws out, tick all the 12 principles etc?
@@-AT-WALKER The thing is, most photographers are and will still be needed because they usually take photos of real events, persons or places, to replace them you would need an actual robot and that would be very expensive, maybe stock photographers are the ones who should worry the most. On the other hand, animators set their stories in fictional places or stylized real places, when the AI learns to do their job in much less time and without wage costs studios won't hesitate to replace them.
At least for now AI isn't able to crank out a full Pixar movie by itself. It's a tool that you still need a human to put in the work to make it do what you want. It takes a different type of artistic skill set that requires no drawing ability. It sucks for the artists who will no doubt be replaced in the future and they will inevitably need to adapt to survive in the industry.
Girl, you are the best!
No, you are the best 🫶
Iclone 8 and CC4 cost $900 USD and you need to own an apple Iphone X (or higher) to use motion live which is even more $$$.
You can create your 3D animated Avatar with Daz studio for FREE
plus a $20 Dollar face transfer plugin to make a Daz genesis character look like you or whomever.
A pretty model 😊
Realised the audio and vid goes out of sync… annoyingly CZcams provides no fix for this so uploading the correct video to my website: Here is the fixed audio and full video www.patreon.com/PromptMuse
Thank you, I know it’s really hard work and your work is really helpful.
I was also thinking at some point you could host a film festival somewhere that shows user created content on your channel based on your guides.
It would be interesting to see what other people are making and how they’re putting this technology to use in animation and filmmaking.
@@CultofThings Fantastic idea 🔥
@@promptmuse maybe put out a call for people who have made something with some of these techniques. Note to them that the length isnt important, like its fine if its only a 3 minute clip. THen request they write up a quick explanation of their general process, as in just 1 or 2 paragraphs taking onto of which software and techniques were applied.
That way, if there is a library of videos people have made, anyone can view them, looking at the descriptions, and get a solid understanding of how each set of tools end up with a certain look, how the end results vary, the time used, common hurdles... and in general what would be most exciting to me would be to be able to explore different workflows just by watching the end results of each.
I would hope such a library would be viewed as a collection of experiments using different workflows, not as a entertainment thing like say netflix. Im not expecting amazing stories or anything, Im just looking for how switching out one tool or technique for another results in a different look.
You don’t have to use Birme to crop it. You can use stable diffusion to downsize the images.
hi there i'm all new to this and definitely following you to learn more... i have a question, is it possible to import any of the ai 3d models you generated into a 3d program??
brilliantly helpful. cheeers
Glad it helped 🫶
You no longer have to crop your images for training, you don't even have to have a 1:1 ratio. Just check the bucketing option and you can choose your resolution. You can use 768x768 with the SD1.5 model also and have hi-res results without hi-res fix
Interesting. Where can we read more on this?
Can't wait for part 2 of Create a 3D World using AI Images I wish to see how you place the character in the 3D world you created
Gym has o😅😅😅😅😅 1:52 m😊 2:21 years 😊 2: 😮ono 48 😊😊 uh😅😅😅ooooiooo😅😅 3:19 😊😅 3:41
Your videos are really great, thanks for all those contents! regarding consistent caracters, I tried to create some for a comic book, but they are historical figures with precise costumes and it does not work, I think the technology is not ready yet for very precise jobs, what do you think?
Danke!
Hello PromptMuse. Thank you so much for your work!
You have amazed me with the way you teach and also thanks for showing Collab this time.
Any similar alternatives for iclone for android users. I'm saving for an iPhone, so I was just curious hehe.
Can't wait to get back to AI again :)
You can use meta humans in unreal. There is a android ARKit for facial movements. This is all free 🫡
Alternatively there is blender with Faceit / rigify 👍 It takes time to learn, but you can get amazing results
I love your videos, even though I have zero time to do any of this 😅
Very interesting and well made. Glad i found your channel.
Have you tried adding fantasy elements to realistic images (elven ears, demon horns ork teeth, weird skin colors and so on)? Have had problems with them my self
Will do, but you can use this technique to train your 3d character into a LoRa. Then there should be no issue at all 👍
Hey pretty -I am stuck at the 10 files saved from the training instances... so, do I install stable diffusion web ui and also download model and vae, found right below that? There's a ton of other cell options going even further down. Sorry for being so noob! I love your work - get it, gial!
Thanks
I ❤ ur content 👍🏼
Hi - Thank you for taking the time & effort to make these very helpful videos. I noticed that iClone only supports Windows. What alternative tools can you recommend for Mac? Thanks in advance.
Amazing video!! Wondering if I can use this for 2d
You’re a saint
There is something incredibly beautiful in You.
Great Video. Do You have an video on how to train image to get multiple position. To us as an thumbnail.
Hey! Something cool that I think you might be interested in! ❤
There is this new AI thing called "nerf", basically it makes a 3D scan from a video!
Wouldn't it be cool if we tried to make a video using AI generated images, then turn it into a 3D scan? 🤯
It should work for characters, objects and environments!
Hey! I actually have a unfinished tutorial video on my desktop. I never published it as was trying to come up with a creative way to show it haha. But this would be so cool! If I do this I’ll give you credit on the vid !!
Could you provide an example of idle moments, including hand gestures and simple movements of the head, and demonstrate them with audio for lip syncing?
When you click it for ai animation but end up knowing all about LORA training ❤️
same here
Great video! A couple questions. At the img2img step it looks like you used the token mkmks instead of mksks that was used in training, which might have affected the quality of the output. Also, does ControlNet still work if you don't put anything into the ControlNet image input area? If there is no input there does it take it directly from the img2img input?
Hey, yes control net still works without anything in the input area. I only use the input area for specific models. Yes it takes from the input frame that you uploaded and follows that as a guide. Ah yeah haha, I’d probably of got a better likeness with the instance
Thank you so much 🙏🏻 Do you know how can I use this amazing technology on my Mac 🥺 ?
I m not english native and I just love your accent.
At the beginning, when you say "back" and "another".
Very charismatic
Cool
Ai isn’t as busty as the creator but it works. Good job 👏
Drops just into uncanny valley 😮
Awww. how do you do it on runpod?
ну ты и красота конечно !!!
You are fire
Can you do a tutorial for the training in a local machine?
Amazing! How do you get to learn this, do you read the manual? Kidding aside, how did you animate the model? Thank you.
I use Characters creator but alternative options are blender/ metahumans etc I spend hours playing around haha
where do you go to get stable diffusion prompt, i'm lost on that step
It would be great a tutorial on how to add a 3D background!
Hey, I’ve got one for Blend on my channel 👍
@@promptmuse yes, I watched "Create a 3D World using AI Images" and it's awesome, but can't find part 2, I mean getting the iClone animated character into the blender scene. That would complete the tutorial.
I'm a bit confused on the Lora process with A1111.
Does it save as a CKPT or only as safetensors?
What are the benefits of using it versus anything else?
Is there a more in depth guide on it?
Greetings from the Pope to the most beautiful AI teacher in the world :)
I like your videos and you are very pretty... what program should I use so that the one I create appears in the video and moves like me and speaks with my voice
Hi, great video, but something happened around 4:25, and after many things are out of sync, a couple audio edits got mixed up in there as well. (I heard siri pipe in at 7:08 ) It pops back into sync at 7:45 or so.
and honestly, thank you so much for making your tutorials typed up, i just prefer that so much, but i understand it gets wayyy less hits and monetization than youtube. thank you!