Animate ANY Face From A SINGLE IMAGE (Thin Plate Spline Motion Model)
ฝัง
- เผยแพร่เมื่อ 11 ก.ย. 2024
- Head to squarespace.co... to save 10% off your first purchase of a website or domain using code CGMATTER
google collab colab.research...
patreon and stuff / cg_matter
my website www.cgmatter.com
#ai #walterwhite #breakingnews
I was very confused why I was getting notifications for a random yt channel I had never seen before. Then finally you included your face on a thumbnail.
Same dude!!!!!!
the fact that you narrate this hole video using the technique that you are teaching us is just amazing...lol
It's very holesome.
OK, I'm only a couple of minutes in and I can tell that the speech pattern, choice of words, pauses etc is very telling. "Either way", "I find it works best this way", "again, you can...", "we're good to go...", "in other words...", "before we begin"
"push this buh-tin"
I wasn't expecting it to fix something like choice of words, but I was expecting it to recognize his lisp, manage to interpret what he was trying to say like we can, and replace it appropriately.
One can conceivably compensate for their choice of words with A script, but an integral part of how someone pronounces syllables that not only cant be changed, but wouldn't be properly mapped if someone were to for example: use cgmatters voice to do this. You would have to compensate for his lisp when impersonating him to make the voice seem convincing.
@@nullvoid3545 I don't think you quite understand what "lisp" means.
@@douchymcdouche169 Your right, after some research I have discovered I was incorrectly using the term lisp as A shortened way of saying speech impediment when the word I should have been using for this specific one is rhotacism.
Thanks for calling me out, I learned something new today.👍
@@nullvoid3545 *wotacism
Phenomenal job, Mr.White!
yo can you do a meth tutorial next pleaseeeeeee
🤣
Its not woking for me, even when I don't even change the source files. I keep getting the error
[Errno 2] No such file or directory: 'checkpoints/vox.pth.tar'
Anyone know how to resolve this?
hey, I got same error, did you solve it?
Same :(
How to do the voice is coming up as the next vid I hope? Gonna use this for my mums birthday, doing it with her favorite artist, wishing her a nice birthday.
It's here!
th-cam.com/video/3yjB_P3UvfE/w-d-xo.html
@@helphelphelphelpmehelphelphelp thx mate
When I reach the part of "Try" it gives a error
How to do that? Where are programs? You give only 10% of info, how to do it live? In real-time?
"bryan cranston here, representing Squarespace..." hmm... already seeing how this tech could be controversial! neat!
Can you do it in blender?
Hi. I am also interested how your dubbing voice also turn into Mr. White's Voice. Is it also AI or just voice filter in any application?
I get this error, does someone knows how to fix it?
FileNotFoundError: [Errno 2] No such file or directory: 'checkpoints/vox.pth.tar'
It happens on step 6 (5:28)
me too! if you found a solution please let me know
@@sam_sands hey i got same error, did you solve it?
thank you walter matter very cool
Oh my god. I'm alive thank you :) I tried the webui thin plate but had no way of upscaling the resolution. This works great!
You're goddamn right!
Do it with an anime face.
This was working fine for a while but now no longer works. When running the "from demo import load_checkpoints" module, I get the folowing error message: FileNotFoundError: [Errno 2] No such file or directory: 'checkpoints/vox.pth.tar' even though the file seems to be in the checkpoint directory
hey, i got same error. could you solve it?
before we begin tho (chunk of meat in superposition appears to the right of heisenburger)
Top notch love the use of Bryan Cranston from breaking bad. So cool!
amazing wow ^^ how did you get walter white voice ?
Super! So you did this whole YT Video using this same method? And used Voice cloning too?
No matter how many times I try this step-by-step, it always just pulls back in Trump and Jackie Chan at the "try" step. I've deleted and re-uploaded source and driving five or six times. What am I missing?
The same thing happens to me, I don't know what to do.
Thanks so much Walter!
What does the video resolution depend on? My video is blurry and not as clear as yours
It's so strange. I followed the exact same steps in the video. At the start, I put in roughly 4K images and 4K videos, and after clicking on the "try" column, the output image still comes out as Jackie Chan (I also changed the names of 'driving' and 'source' files in the folder). I thought maybe the image and video size were too large, so I adjusted them to 480X720 resolution. However, even after doing this, once I hit generate, it still comes out as Jackie Chan. I checked the code, all the paths are correct, but it still generates Jackie Chan...
Got u, do exactly the same the author did, u might by passed things. I got the same result like yours, then I checked it, did step by step, then got the result I wanted
i tried with diffrent images and videos and it didn't work pass the step with the two in a video.
In the final result, the mouth is barely opening at all.
Yet I've shot a pretty clear video in which you can see all facial details.
Any idea why?
any way to do this locally on m1 and do a batch process of this ?
Does this work for a non human face? For example an image of a Pixar style cartoon bear face 🐻? or is it only able to detect “ human “ faces?
I also have this question....
I was able to generate a video but how do I then download it? I refreshed the screen and it vanished.
Thank you very much, may I ask what software you are using to record live chat water white videos?
omg, thank you, just thank you for teaching all this incredible knowledge
Did this break for anyone else today? Was using it yesterday fine but now I get errors at:
from demo import...
And can't move beyond it
Hey how would I go about changing the resolution of the outputted video? I'm trying to get a higher quality looking video, similar in resolution to the original photo.
Hey, I wasnt able to change the resolution. How do you do that? When i change it i get error in the last code block
How did you do the voice too?
Will this work, if I have a full body shot, my pose will remain static, but the face animate.
hello i need help i keep getting the runtime error: Sizes of tensors must match except in dimension 1. Expected size 32 but got size 33 for tensor number 1 in the list.
Thanks for sharing this, Walter!
Hi, thanks a lot for your video. Do you know if it's possible to change the frame rate at 25fps and the image size?
Does not work. Colab will not run the checkpoint code shown at the 5:28 mark in the TH-cam video. Tried about 25 times.
I don't understand what I'm doing wrong... I follow all the steps exactly like the video, but when I get to the "TRY" step, I get the video with the TRUMP image and the CHAN video, as if I haven't changed them for my own assets.
same. I wish I knew what to do
After you run the "Try" step, click on the refresh folder icon (on left side under files) then run all again from the top...This worked for me! (now I"m stuck at the "from demo" step...I think I have to change to local runtime under top right ram/disk?)
I’ve been messing around with this technique but I gave up due to what I thought was the 256x256 pixel limitation for output. Have you had success with higher resolution output? If so I’d love to give it another go with higher resolution assets.❤😊
I export the video to png sequence and upscale those using GFPGAN, then re-render the video.
@@ChrisEldridgeKino hey, can you explain me how to do that? If I export the video to png sequence then I have like 1000 pics
@@belenparedes2085 it will be a lot of images. Just look up how to use gfpgan or even a tool like upscayl to process all of the images in a directory. Then convert those 1000 images back into a video with ffmpeg
Is there a way to download the Git and run it locally via Pinokio?
the mp4 and picture i imported not changing when i click on try its still showing the old one
Amazing! How did you get your voice to sound like Walter White?
There are free versions, but you can use 11Labs to input your voice, input a model voice and it will convert your recording into the model voice.
@@astralislux305 Thank you! Can you please recommend some free versions? I don't really want to pay for 11 lab
@@astralislux305is there any free version?!
after getting your avatar video how do you make it to talk and mimic all your expressions in real time?
"Cgmatter? Uh aah
I'm the Walter"
this is a huge step up form first order model
Really cool video and I'm interested to try it out, but I'm a bit nervous about using this model for safety and privacy reasons. What do you guys think?
I don’t know code so I’m spooked to run it too
Wow that's crazy stuff, i'm interested in the full body
I would like to do something similar to what exists in the iphone.
Where should I go? is there anything already built for it?
Thank you for the content
Oh wow!
I can think of all kinds of fun ways to use this!! 😈
This looks interesting, I send a photo and a video and what about the voice issue, does the video need to contain the right speech?
for some reasons I can't get one of the steps to run...It always says "demo" not found. Can you help?
if you
haven't run all of the green ticks in the correct order will it matter? And if so is there a way to refresh the process? Thanks
what about the full body version??
It would be nice if you mentioned that this method won't work if you don't have the NVIDIA driver. Cheers!
I am thinking to apply this technic to a photo of 3d model then i dont need to do facial animation
is it possible to do that using a faceless image , for example spider man or dead pool ?
Is really handsome Bryan Carton.
Nicely explained. I would like u to make a video on full body to......
I learned this so much faster knowing Walter White is my teacher. Thanks!
Thank you. I'm getting weird artifacts around the mouth similar to another app. Not sure how to fix it.
Wouldn’t DeepFaceLab be a much simpler software to use than this one?
Yeaaaah! science beeeech!
So you can do live streams with this?
That's amazing! But how did you fake Walter White's voice?
Can you please make full body one please 🙏. Thank you so so much
how can i run it locally on pc
crashes every time due to low ram
this is insane man
how can you create such a 8 minutes with WHite's mask??
me 1 minute is longest. Google collabo crash
Looking for an answer to this question as well !!
Can a 3D FBX Model of the Animated Final Video be generated too ?
What's going on in the output when you open your mouth?
but how did you make it as your avatar? and using a voice changer
I get this error in the two last "from demo import": No such file or directory: 'checkpoints/vox.pth.tar'
give a link if you have a more detailed analysis, even if it is paid, thank you!
It's cool but 256x256 pixel limitation + looks like I can't do more than 5seconds, I don't know why. Too bad :)
It's dmn awesome.
Thanks for the video, works like a charm :)
Any idea of why I can't upload the video?
are you solved it yet? I've got the same problem
I have an error message for the fifth step "image file is truncated", can u help me please
my picture was way too big lol, I shrunk it and it's fine now !!
Spline available on mobile
this gg colab link is no longer existed
Thank you for tutorial! How do I get a higher resolution video?
Please do realse the one with full body . Thanks
I tried it yesterday an it worked just fine, but today it is giving errors!
The really trippy part about is is when the eyes look down for a second it’s crazy😂
Can I also make voice?
Thank you so much for sharing this video😊😊
Hi and Thanks.. Can we run this offline?
Can we use this for video call
How exactly do I change the resolution?
Pls I need app on Android that I can use it from phone
I can't get pass the try block :(
Use a video no less that 10 seconds. That's how.
what are you using for the voice?
how to make Walter White narranting you video?
Can anyone tell me where I can find the moving face video to download?
It's weird when your mouth is open wide though. Sort of Alien Xenomorph effect. 😄