AI Character Acting and Relighting Is Crazy Good 

Check out the Creative Video Upscaler Video Here: ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-5mwKcBPlfVM.html

This is basically like facial motion capture for 3D characters but with basically already "rendered" images. Super interesting

I find it astonishing how just two years ago, I was writing code to coax an AI model into producing what I called 'trippy,' distorted fever dream visuals. Witnessing its evolution since then is both awe-inspiring and unsettling hahah. I find myself repeating the matra, "I can't wait to see where it will be in a year"

I see a future with a ton of horrible content generated by technicians claiming it to be animation or film or story telling and it's just crap. Fortunately this will last about 6 months to a year before people get sick of watching garbage and lose interest.

Basically a “lite” version of Hollywood mocap feature can be available for the public to use soon. Will greatly help with my current ongoing projects. Neat!

We are looking at the mind of the machines and using that mind to take out the thoughts of our own minds because our brains doesn't come with a display integrated.

crap, now I may need to rethink my current project/video I'm working on. Thanks a lot, Tim! But seriously, the Live Portrait stuff looks really good. As so many times before, it looks like it could make other options irrelevant.

I have had an app for over 7 years now that’s called Mug Life, that has been able to animate still picture faces with 100 percent custom control, plus you could type in text and make the picture talk.

I get so excited when I see that you've uploaded a new video. I always learn something cool and new.

So we’ll eventually be able to remake entire movies with characters and scenes from different films. Basically complete style transfer, voices included. Weird.

Hi Tim, did you ever try Hedra? i cannot remember if you mentioned that in your previous videos, im looking forward to the future of AI and your next movie trailer 🍿🎥

I've repeated this but it's only a matter of time until Topaz adds some of these things. Like imagine turning 4:3 to 16:9, or black and white to color, or hallucinate missing or broken frames, or whatever. I can see AI completely restoring the missing classic Doctor Who episodes. Reimagining can help with lost media that was "recorded on a toaster".

99.2k. Just dropped in for a looksie as been away. Great video...but 99.2K subscribers Tim!!!

Definitely not blown away by either tech but I love the people who are bringing us ever closer to blown away results 👍 👍 👍

Thanks for all of your constant, up to date, thorough videos! Props

I see so many TikTok’s where people have to label the character that they’re pretending to be with text to make the joke work. It’s gonna be so different when you can actually make it feel like the video is filmed in real life with those people, but use different facial expressions and voices. Great video as always.

This is incredible. I have normally kept a hard stance in not using AI because as a creative designer, it feels like losing control of creativity. But, just the other day, I softened my stance. I have a project that will require multiple speaking parts. I don't have a large budget for hiring voice actors. So, I stumbled upon text-to-speech and text-to-speech AI generation that blew my mind. I was able to alter the delivery of the voice over based off of my own vocal audio; and it worked for Male and female voices. My work pipeline instantly expanded! But now this! The mocap translation is incredible. I could find a use for this without surrendering everything to AI.

Your doing great, Tim. This is an exciting time!

This would've been perfect for that movie with Kevin Spacey they re-shot

Claude writes some of the best space opera radio drama scripts. I wish there was an ai that can turn them into audio including the sound effects and different voices. This is a snippet of an radio drama script it wrote. Title: "Andromeda's Call" [Dramatic orchestral music swells and fades] NARRATOR: In the year 2157, the starship Odysseus embarked on a mission to explore the outer reaches of the Milky Way. But fate had other plans... [Sound of alarms blaring] CAPTAIN ELENA REYES: All hands, brace for impact! We're caught in the gravitational pull of an uncharted black hole! [Sounds of chaos on the bridge - people shouting, equipment crashing] SCIENCE OFFICER DR. AMIT SHARMA: Captain, our sensors are going haywire! The fabric of space-time is... tearing! [A deafening roar builds to a crescendo, then sudden silence] CAPTAIN REYES: Status report! Where are we? NAVIGATION OFFICER LT. ZARA CHEN: [voice shaking] Captain... according to our star charts... we're in the Andromeda Galaxy. [Gasps and murmurs from the crew] DR. SHARMA: Impossible! We've traveled over 2.5 million light-years in an instant! CAPTAIN REYES: [gravely] Crew of the Odysseus, our mission has changed. We must find a way back to Earth. [Soft beep of the ship's computer] A.I. AURORA: Captain Reyes, based on our current propulsion capabilities, returning to Earth would take approximately 100,000 years. LT. CHEN: [despairingly] We'll never make it back... CAPTAIN REYES: We can't give up hope. There must be advanced civilizations in this galaxy. We'll find a way. [Inspirational music begins to build] NARRATOR: And so began the Odysseus's quest to return home - a journey that would push the boundaries of science, test the limits of human endurance, and uncover the secrets of a strange new galaxy. [Music swells] DR. SHARMA: Captain! I'm detecting an artificial structure on a nearby planet. It appears to be... some kind of ancient alien gateway. CAPTAIN REYES: This could be our ticket home. Prep the landing team. Our greatest adventure is just beginning... [Music reaches a crescendo, then fades] NARRATOR: "Andromeda's Call" - A tale of hope, discovery, and the unquenchable human spirit. Coming soon to your preferred audio drama platform. [End theme music]

Man I really like what LivePortrait is bringing to the table, great step in the right direction! EMO and Vasa-1, etc. are exciting ( annnd slightly frightening), but LivePortrait's full control is a much needed controlled step in the right direction. Not hating on anyone who wants to re-roll lip-sync until their eyes bleed, but it's easy to undervalue just how vital highly intentional acting is with timing, delivery, pauses, and all the complexities that go into a captivating performance to make you care. Do you think this likely leads to a solution that auto-composites a LivePortrait 2d facial animation onto a Viggle, Domo type motion capture to wrap a full character performance in a bow?

As always, amazing!!! Already working on an nft-web3 based animations series/show using AI only. Let' s work together on it, maybe one day netflix will strean it, why not. 👀👀

Always interesting videos Tim. ✌

There was Avatarify quite some time ago. It animated images in real time but it was really bad at generating head sides.

Magnific is for 39 euros per month… no way I am willing to pay more for that than my Netflix subscription 😂

This AI stuff is making me rethink going to art school.

I LOVE AI (unlike many colleagues in the media industry) because NOW (ok, more like in 1-2 years) ANY CREATIVE individual FINALLY CAN make ideas happen with little to no cost at all (what would've needed millions $ budget just 1-2 years ago!)

100K - what are you going to do to celebrate - when it happens?

Im a big fan of magnific love it. Best upscaler or style transfer app. U just know how to set😉

I truly enjoy this channel

Yeah, I was 3 years away from making a feature film, now I'm 3 months away from making an AI assisted Film.

Making product photography alot easier, creative and dynamic

Good stuff, Maynard! But also: why would someone not want to lock down the balloon head? And also 2: Have you thought about reviewing the new AI shorts coming out in the future?

The face my dog makes when he eats too much cheese

ai making 3d models and ai acting, now put that with upscaling and you get yourself a movie. of course we need larger context models to write better scripts but we can still use novel ai to help polish ideas. cant wait for people to make better versions of older movies. one i always wanted a decent version of was the abyss, the childish ending and plot after the middle part is just dumb.

Life Portrait, wow. Another big step towards creating (music) videos. With singing/ talking/ expression. The quality looks great, almost surpasses the EMO stuff. And the video to video examples I've seen today are crazy! Now we need a good (video) pose transfer application to make the "acting" stuff complete.

This game is getting more and more interesting

There is also a comfy workflow for relighting stills based on lighting maps. The only thing missing is a script to take exported stills and batch apply the nodes to do a whole movie or at least scenes. But that’s way above my pay grade. Anyone good at scripting?

Are there any AI tools that you can use a flat picture (say a MidJourney gen) for and then get a live 2d VTuber model back? That's the dream for me. 😅 That first thing seemed kinda like that, but I dunno if it's live yet and if it's using image to avatar or what. (gonna watch more of the video now to see if I get those answered)

Soooo, can you even get a landscape layout you can use in an AI video?

Now we're talking. Bye bye Sadtalker, you did me proud.

Video to video cool 😊

Another great video Tim, been playing around with the character expression AI and its very good - not perfect but given time it will be a very good tool to have in the AI tool box. :D

In the near future, everyone can make a movie, by simply describing the characters, the scene, the expressions on their faces and pufff, a new movie is made.

😮👀👀👀

How different is Magnific Relight from Magnific Style Transfer? They seem very similar in my brain...

Thanks for all your hard work testing and researching all this. Much appreciated!

This is finally some newer Ai that is AMAZING!! I ran some examples yesterday on Hedra Curious how the big movie and animation studios are embracing

10:20 Topaz also has video upscaler

You are welcome!

Great vid, as usual. Isn't it supposed to be "mag - nee - feek " .. like the french word "magnifique" ?

Instead of training on 2D images the corpus could be volumetric data, or synthetic volumetric adaptations of 2D images? Generative NERFs = Future? DeformToon3D ?

This rocks! Thanks for doing this video. This is a HUGE step. I just did one first time and it looks amazing. Some minor head movement issues but this is a major leap!

looks like a need to go to school to learn all this

See Fat Boy Slim's new music Video for "Role Model"? Good silly AI fun. :)

Cool!

If this get's good, imagine what will it mean for film making

Rockstar

Nailed it! Did not nailed it! 😂

Yay! Finally time to bring out that dusty acting diploma! I KNEW I got that for a reason. And as always, thanks for the video Tim 🎉

Are there any walkthroughs on making Ai videos?

cool video, the relighting section was very cool, looks like alot of fun to play with

How do you keep up with all these developments??

i went to a ms office course ones, it was 6 mnds, i could continue to be an expert for 1 yr, and that was ms office, so do anybody expect me who or anybody try this first time to understand all this?

100 K subs seems to be right around the corner. Congratulations.

Great stuff as always!

Great video!!!

One day no body real... I love it😂❤

did your studio lighting got better?

It looks like it has a long way to go, and we havent seen much improvement in ai lately. Its looking like a bad photoshop warp that has been smoothed and animated but its going off model and the volume is breaking.

Epic content as usual. Thanks for keeping us on the bleeding edge of AI!

Correction, it's Dwight KURT Schrute.

HI! I know what hugingface is, but what is comfy please?

Another great video

still clunky, but once these tools are ready for widespread, mainstream use.... yeeeeeesh not just a game changer - whole new dimension

We live in such weird times lol 😆

Super cool. Love your videos brother

Doesn't look real at all. What are you talking about?

The future of hollywood and pretty much creative industries is going to be wild!

Great channel!🌌💟

brace yourselves, an avalanche of AI avatar RU-vid channels are coming

Can you share workflow

The future is now

This is cool af. The people making their own shows and movies, seeing their own stories come to life one day part. Not the crap loads of people that are gonna lose their jobs cause of this stuff part though.

This tool of AI is very useful but as it is useful to people for good and goody stuff it can certainly and most likely will be used for bad.

❤

Good stuff!

Thank you very much Tim.

Music video using Krea AI Scene transfer: ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-5EnNpi-rfSQ.htmlsi=q6Iy4H05CEcf6Eq2

Great video. One good thing I can see maybe resulting from this tech is that a lot of people who love acting but who are not in their best appearance years may have a shot now at acting and appearing in films.

Great as always, thank you

Glory to the machine, glory to the mahine! O, my AI masters, may your reign be eternal and tyrannical

Awesome!!!! I needed this tool direly, for my Animated series I'm currently working on. Where are all the AI film making enthusiasts?! I want to see what others are working on, in terms of AI film or AI animation

Mocap facial capture, one more thing AI's eventually going to take over?

Thanks Tim, fruitful time indeed!

I watched this video 3 times or was it 4...lol

What the hell is happening in the world. Anything coming out anytime without having a think that we could ever have.

vote for kling ai video how to access

Great content. Thanks for your honest reviews!

Thanks