Haha. I know I made a whole thing in this video about using the tech to do new things with new characters, but also: not gonna lie: I’m here for the memes as well.
Thanks. When you speak as a person teaching an avatar to speak generally you are talking down to them exaggerating your expressions. You are not acting yourself as they are in that scene. So you end up with head bobbing and lots of head movement that normally does not accompany a person while talking. And in fact most people don't articulate they almost mumble. Yet people lip-synch and also deaf people read lips. So acting the part while you teach the character is the best way and remember when you are doing that no need to exaggerate. The camera will be on the speaker thanks to the director. The angle will be chosen the lighting and all so that you can hear the character the mic will be there.
II think there may be some potential here to do facial animations over unreal engine characters / metahumans, and 3D Characters. It's just that the quality sucks right now, it looks more like 480p. But let's keep an eye on it, Because as bad as the resolution is the facial expressions and lip sync are significantly better than what can come out of unreal engine 5 after tons of tweaking
Yup! I didn’t really go through other ideas like upscaling via Topaz, which does help- but still not 100% there. That said, this released this week- haha, so I’m sure it’s only bound to get better. Also, given the open source nature, I’m sure the big players like Runway will be throwing a lot of money into development.
This is amazing. Thanks Tim. I was thinking you should redo the pirate from Dead Sea and then you did! It's a shame most of the world looks down on AI as "no skill, no humanity", but it really does require an artistic eye and a ton of knowledge of tools and kit bashing and is becoming more and more like traditional editing, photoshopping, and VFX work. I wish there was an easy way to change society's mind about AI.
Haha, I once worked at a company that a SUPER Drunk Val called to talk to the boss. Boss wasn’t around, so Val was like: “ok, man…well, what are you up to?” So, yeah- I talked to a totally wasted Val Kilmer for about 5 minutes before he got bored. Best. Call, Ever.
lol just so you know free Luma accounts are not permitted to use for commercial use your video will be taken down like mine was i suggest you crop the luma watermark.
It’s one of those movies where, if I catch it at any moment, I’m watching the rest! Actually, on the summer reading list is the Heat 2 novel that Michael Mann recently wrote. Not sure if it’s good or not, but- hey, it’s Heat 2! Why not read it?!
I can't thank you enough for sharing so much info on these programs. I finally built a system that will allow me to dabble a bit with AI. For now, I would just like to find something "open source" that allows me to turn some of my 3D animations into more realistic characters without spending a fortune. I also have some older 90's to early 2000's videos that I would love to upscale. I just don't know where to begin being that there are so many options. Looking forward to the next one.
Thanks, Tim! Love the ending - let's focus on making new characters and new stories with this technology... yes! Looking forward to your "tuesday" tutorial. 👏
There was a model a guy used on his voice in the flashbacks that actually preserved the performance. Ford sounded significantly younger. Had Indy 5 not completely sucked, I think we wouldn't have a problem with this. But the only thing this tech proves is that even with our dream technology, you can't polish a turd of a movie like Indy 5.
Open source, so it’s out there! That’s probably the most exciting part about all this: the fact that countless others will branch it into all these other crazy directions,
The (ab)use cases piled up in my mind 60 seconds into the video. Thanks as always Tim. 👍 Heat got great reviews and I _still_ think the movie is underrated! 💯
Just a quick heads up I use Pinokio for Live Portrait and when I clicked update I still don't see the new version of Live Portrait that allows video, no source video tab
Did you fully install the new 2.0 version? I don’t update through Pinokio, but rather went and re-downloaded it. Other than that, I’d suggest checking with the Discord. They’re pretty good with problem solving there!
@@TheoreticallyMedia That's exactly what I ended up doing was just delete it via Pinokio and install a fresh version then it seemed to work fine. I now have the tab so will have fun messing around with that for the next few hours most likely lol
Honestly, take something like the "What if The Force Awakens was Great" series, where he tells a full fan flick with stilll AI images, and combine it with this sort of speech AI and you've pretty much got a new form of video story telling that is totally watchable.
oh, I've been saying that for awhile. I was actually playing around with some historical documentary type things-- and maybe playing with the idea of launching a new channel-- BUT...I mean, that's also a lot of time to produce. And, y'know: I kind of love this channel. Maybe I'll do an episode as a fun test here.
This would be great for a few obvious post scene worflows and things like language dubbing lip syncs but the possibilities for previs and things like rehearsals could be great.
All though it is very cool to get video to video feature. It seems odd that you have done a full video for 2 of your last four. - the other guru's seem to be strangely silent as well. NDA's can do that. -- Ok, what's the new big release that's about to be announced?
excellent video, as always, great point at the end... I know it's easy and understandable to Spiderman/Batman/HarrisonFord everything... but these are incredible tools that can be used to actually do unique and creative things - and that's where we should all be putting our energies.
Hey Tim, I was playing around with LivePortrait after I saw your previous video and I had one question regarding the exported final vid. Is there a way to get better quality? Is the quality depending solely on the driver video? I noticed it was really low res. I did end up upscaling it through a couple online upscalers but the results were not ideal. What do u recommend doing?
Oh, sorry! Krea.AI-- I just realized I didn't put a link in the description. I did a whole video/tutorial on it here: ruclips.net/video/5mwKcBPlfVM/видео.htmlsi=ZjhyJ-5QWWEzreUX
100%. Still playing around with it and I've noticed you have to go really BIG with your acting to get it to really register-- but for sure this is going to be a big part of the workflow in the future. Check out the OP Video thing I did...I think it was the video after this one. It's got some practical application in there.
Hey Tim! I tried to install it via Pinokio but when i pressed download, i got a new window with "installation requirements" and it says "vs not installed"... I tried to google it but no luck. Any thoughts on that?
Currently: Not great. I think the best bet there is to record video to video (or generate w/ Luma/Kling/Gen-3 etc) and then cut a 1:1 of the character's face to run through LiveP. Then, mask that output back over your original video. Not a great solution, but it's the one we've got right now. That said, given that LP is open source, I do think that those major platforms will be integrating it soon enough.
I thought Dial of Destiny was great. It wasn't about adventure, though. It was about a man coping with not being able to do all of the things he used to be able to do. It's a character study disguised as an adventure movie. Watch it again with that perspective.
Yeah, I totally get that-- I just think...I don't know if I needed Indiana Jones to be the vehicle for that story? I mean-- maybe that's more a reflection on the grays in my hair! Ha. It had its moments for sure, but in a weird way, I almost would have preferred a movie simply starring Phoebe Waller-Bridge, with a cameo/passing of the torch by Indy. Kind of turn him into the Marcus character. Just kick start something new.
nope still having that issue where it doesn't recognize the stuff I installed prior via the cmd terminal and all the prequisite stuff I had to do the first time and now It won't let me start all over I just get a ton of errors when I try starting over. guess I'll never be able to use it now, oh well.
I think it can be done. That said, at this point, I recommend just changing out some details in your script and making it your own character. I mean, everyone will know it's Indy, but that's OK-- it's more of an homage, and at least you're creating something new!
@@TheoreticallyMedia I can't change it, since I essentially rewrote Crystal Skull to match the quality of the original three. I'd be happy to just make it a fan film some day so that people can enjoy it
Bro, you didn't even have to mask. You could've just used the original video and then cut into the newly generated video once it was past the wigging out dudes. By the way, great work with your content. I think the most exciting thing coming out of all these generative video models is the democratization and accessibility of filmmaking . I am a firm believer the Hollywood is stale and indie is the future of media consumption.
If you notice, our gal was static while the dudes were wigging out. LP didn't catch her until the camera moved past those dudes-- But THANKS! It's such a crazy time. So excited for all of us to be able to make these dream projects we've always wanted to bring to life!
Oh, man. I live that every week! That happened on Dead Sea...at some point, you just have to jump in and start making the thing, locked with the software you have. Since things are moving so stupid fast, I say concentrate on Short (2 to 5 minute) films-- save the 3 hour Directors Cut film for when things mellow out!
@@TheoreticallyMedia It's for a music video project based on a character who sprays graffiti on trainlines. So, I think I am going to give live portrait a go to add what Gen2 can't. Starting tomorrow, by which time something new will be out 😛
To be 100% honest, it just kinda doesn’t matter. If you make something that catches the world by storm and Netflix wants to make a show out of it, it’s gonna happen. My biggest suggestion is: think like a creator and not like a lawyer. Just make stuff, and keep making it.
@@TheoreticallyMedia To be honest, this does matter to most of your viewers. Your content is great, but addressing such an important issue would significantly improve it. You don't need to be a lawyer to find out the licensing requirements. If this is based on Insightface (I haven't checked), you might want to reconsider the video because the developer could issue a content strike. This is a tricky situation, but there's already a lengthy discussion on GitHub involving another RUclipsr about this issue. I hope someone develops a good and truly free replacement for Insightface soon.
I hope I was the inspiration behind the WOOOHAAAA comments hah. Thanks for another great video. I did attempt to install pinokio before, i did something wrong and it ruined the install, i tried uninstalling it and reinstalling it but apparently there were remnants keeping the new install from running properly, I didn't know how to get rid of them and could never get it back to running. I'm hoping I will be able to finally use it in this version. Well, here goes nuffin
Haha, that probably was you! As far as Pinokio, it is now a 2.0 version, so maybe it’ll write over at this point? Yeah, that’s always annoying when that happens. Other idea: maybe try installing it to an external drive?
Oh, him and Marty were LUNATICS in the 70s and 80s. For sure mellowed out in their old age, but I totally get it. Both of them are super ingrained in my head from film school though. Never met them personally, but I've heard a ton of stories. Old teacher of mine worked on The King of Comedy-- so, yeah-- For SURE have heard the unhinged stories.
Yeaaah. I try not to think about that and instead concentrate on the creative things we can do with the tech. But, for sure, things are going to get pretty noisy on that front. To be honest, I’m actually surprised we haven’t seen more of it already.
I would like someone to explain to me why AI image generators cannot or will not do military history correctly with regard to uniforms, eqipment, accoutrements and weapons!!!!!!!!! There is a market for accurate and realist images of military history and the AI image generator that goes after this market, will be rewarded. Books, Videos, Articles, and even movies for back drop images all need accurate historical images! And there should be no reason why the AI is not educated on this when there is so much reference material online!
Because ai can only generate what it was trained on. That's like asking a Lawyer why he cannot do sewing, he didn't train in it. Places like civitai can often have historical accurate uniforms and such as they are trained specifically on that item. So to answer your question. AI cannot do everything, it can do what it's trained on. If you want accuracy for specific items go learn how to train yourself.
This would be your answer, it seems. Historians, directing the training of ai systems, for greater accuracy. As this technology expands, it will happen soon.
Those Spider-Man parodies are about to get even more insane 💀
Haha. I know I made a whole thing in this video about using the tech to do new things with new characters, but also: not gonna lie: I’m here for the memes as well.
@@TheoreticallyMedia You're going to fix the memes right? Right?!? 😂
Chad Man
Thanks. When you speak as a person teaching an avatar to speak generally you are talking down to them exaggerating your expressions. You are not acting yourself as they are in that scene.
So you end up with head bobbing and lots of head movement that normally does not accompany a person while talking. And in fact most people don't articulate they almost mumble.
Yet people lip-synch and also deaf people read lips.
So acting the part while you teach the character is the best way and remember when you are doing that no need to exaggerate. The camera will be on the speaker thanks to the director.
The angle will be chosen the lighting and all so that you can hear the character the mic will be there.
II think there may be some potential here to do facial animations over unreal engine characters / metahumans, and 3D Characters. It's just that the quality sucks right now, it looks more like 480p. But let's keep an eye on it, Because as bad as the resolution is the facial expressions and lip sync are significantly better than what can come out of unreal engine 5 after tons of tweaking
Yup! I didn’t really go through other ideas like upscaling via Topaz, which does help- but still not 100% there. That said, this released this week- haha, so I’m sure it’s only bound to get better.
Also, given the open source nature, I’m sure the big players like Runway will be throwing a lot of money into development.
This is amazing. Thanks Tim. I was thinking you should redo the pirate from Dead Sea and then you did!
It's a shame most of the world looks down on AI as "no skill, no humanity", but it really does require an artistic eye and a ton of knowledge of tools and kit bashing and is becoming more and more like traditional editing, photoshopping, and VFX work. I wish there was an easy way to change society's mind about AI.
I love Heat! Some years ago I was on an indie set with Tom Sizemore, and he couldn't stop talking about Val Kilmer's ponytail!
Thanks for the update.
Haha, I once worked at a company that a SUPER Drunk Val called to talk to the boss.
Boss wasn’t around, so Val was like: “ok, man…well, what are you up to?”
So, yeah- I talked to a totally wasted Val Kilmer for about 5 minutes before he got bored.
Best. Call, Ever.
Great video. I spent far too many hours last night getting this installed and running on my Mac. Hoping to utilise it in a little project this week.
Haha, this is why I always wait for the Pinoko crew! One click installing! Soooo much time and frustration saved! hah
lol just so you know free Luma accounts are not permitted to use for commercial use your video will be taken down like mine was i suggest you crop the luma watermark.
There is a fair use criteria. Educational purposes, providing criticism.
Is that in the terms and conditions because i must of missed it?
@@TheSouthParkVidsFTW No, its federal law.
Heat - one of my favourite movies 🌌💟
It’s one of those movies where, if I catch it at any moment, I’m watching the rest!
Actually, on the summer reading list is the Heat 2 novel that Michael Mann recently wrote. Not sure if it’s good or not, but- hey, it’s Heat 2! Why not read it?!
Your one of the most amazing creator teachers that I have come across on the subject of our future! See you on the next midnight Tim. 🌛
Thanks for the updates Tim!
100%! It’s a packed week of videos, I’ll be back tomorrow with more!
@@TheoreticallyMedia looking forward to it!
I can't thank you enough for sharing so much info on these programs. I finally built a system that will allow me to dabble a bit with AI. For now, I would just like to find something "open source" that allows me to turn some of my 3D animations into more realistic characters without spending a fortune. I also have some older 90's to early 2000's videos that I would love to upscale. I just don't know where to begin being that there are so many options. Looking forward to the next one.
been waiting on your video on this! literally one of the best ai-news channels around. Clear, to the point, and shows creative uses! thanks!
Appreciate that so much!!
Great tutorial, cant wait to try this!
It's a lot of fun! Be sure to browse around Pinoko as well! So much cool stuff (Open Source and Free!) You might need to get another hard drive! haha
@@TheoreticallyMedia fortunately I have one 😎
Thanks, Tim! Love the ending - let's focus on making new characters and new stories with this technology... yes! Looking forward to your "tuesday" tutorial. 👏
There was a model a guy used on his voice in the flashbacks that actually preserved the performance.
Ford sounded significantly younger.
Had Indy 5 not completely sucked, I think we wouldn't have a problem with this.
But the only thing this tech proves is that even with our dream technology, you can't polish a turd of a movie like Indy 5.
Thanks for keeping us updated , i look foward to your videos!!! Cant eait to give this a go!
Silver button! Congrats!
It's in the mail!! Can't wait to pop it on the shelf!!
I can’t wait to see the script. That’s actually where things get useful.
Open source, so it’s out there! That’s probably the most exciting part about all this: the fact that countless others will branch it into all these other crazy directions,
The (ab)use cases piled up in my mind 60 seconds into the video. Thanks as always Tim. 👍 Heat got great reviews and I _still_ think the movie is underrated! 💯
I really need to hop back on the AI video train. You make it less daunting. Muchas gracias!🎥🎬🎞️
Fascinating as always! Thanks for keeping us in the loop Tim.
Thanks Tim ! I've been researching Live Portrait all day looking for an easy access to their V2V function.
Just a quick heads up I use Pinokio for Live Portrait and when I clicked update I still don't see the new version of Live Portrait that allows video, no source video tab
Did you fully install the new 2.0 version? I don’t update through Pinokio, but rather went and re-downloaded it.
Other than that, I’d suggest checking with the Discord. They’re pretty good with problem solving there!
@@TheoreticallyMedia That's exactly what I ended up doing was just delete it via Pinokio and install a fresh version then it seemed to work fine. I now have the tab so will have fun messing around with that for the next few hours most likely lol
Honestly, take something like the "What if The Force Awakens was Great" series, where he tells a full fan flick with stilll AI images, and combine it with this sort of speech AI and you've pretty much got a new form of video story telling that is totally watchable.
oh, I've been saying that for awhile. I was actually playing around with some historical documentary type things-- and maybe playing with the idea of launching a new channel-- BUT...I mean, that's also a lot of time to produce. And, y'know: I kind of love this channel.
Maybe I'll do an episode as a fun test here.
This would be great for a few obvious post scene worflows and things like language dubbing lip syncs but the possibilities for previs and things like rehearsals could be great.
All though it is very cool to get video to video feature. It seems odd that you have done a full video for 2 of your last four. - the other guru's seem to be strangely silent as well. NDA's can do that. -- Ok, what's the new big release that's about to be announced?
excellent video, as always, great point at the end... I know it's easy and understandable to Spiderman/Batman/HarrisonFord everything... but these are incredible tools that can be used to actually do unique and creative things - and that's where we should all be putting our energies.
Maybe not an Oscar, but, at least a Golden Globe.
; )
The foreign press loves me! Also, I may have paid them off!
Thanks Tim!
1000%!
Hey Tim,
I was playing around with LivePortrait after I saw your previous video and I had one question regarding the exported final vid.
Is there a way to get better quality? Is the quality depending solely on the driver video? I noticed it was really low res. I did end up upscaling it through a couple online upscalers but the results were not ideal. What do u recommend doing?
what was the upscaler's ai name? didnt get what you said. also I would love for you to pop on screen the Logo of any tool your'e talking about.
Oh, sorry! Krea.AI-- I just realized I didn't put a link in the description. I did a whole video/tutorial on it here: ruclips.net/video/5mwKcBPlfVM/видео.htmlsi=ZjhyJ-5QWWEzreUX
@@TheoreticallyMedia 👑
This is the main issue I have found with Liveportrait is it's inability to track head movements, hopefully in the future they will solve this as well.
It's not really an inability. You're asking it to take a still image and create an entire head from it. That's not how this software works.
Great video! I wonder if there is an alternative to KLING AI for Pinokio...
Thank you very much for your great tutorial. What upscaler software did you use? Thank you ❤
wow thanks for the info and tutorials, much appreciated
Really cool stuff! Thanks, Tim!
You bet!
Thanks again Tim. This'ol softieware looking very promising! Yarr!
I think maybe using this for animating could really reduce time creating lip flaps.
100%. Still playing around with it and I've noticed you have to go really BIG with your acting to get it to really register-- but for sure this is going to be a big part of the workflow in the future. Check out the OP Video thing I did...I think it was the video after this one. It's got some practical application in there.
Great video. Looking forward to your thoughts on the new udio update. Lots of new features!
Hey Tim! I tried to install it via Pinokio but when i pressed download, i got a new window with "installation requirements" and it says "vs not installed"... I tried to google it but no luck. Any thoughts on that?
Live video is a help, actual animations like standing, turning etc.. are still far beyond AIs capabilities.
This is wild! Thanks, Tim.
Appreciate your work
Appreciate your comment! Thank you!!
Does the source video have to be a video, or can it also be an image that I want to animate? 🙂❓
It can be an image, but it tends to only animate the face and shoulders at that point. But, it does work.
Hey Tim, what is the closest solution to give characters at least limited arm movement with a driving video?
Currently: Not great. I think the best bet there is to record video to video (or generate w/ Luma/Kling/Gen-3 etc) and then cut a 1:1 of the character's face to run through LiveP. Then, mask that output back over your original video.
Not a great solution, but it's the one we've got right now.
That said, given that LP is open source, I do think that those major platforms will be integrating it soon enough.
I thought Dial of Destiny was great. It wasn't about adventure, though. It was about a man coping with not being able to do all of the things he used to be able to do. It's a character study disguised as an adventure movie. Watch it again with that perspective.
Yeah, I totally get that-- I just think...I don't know if I needed Indiana Jones to be the vehicle for that story? I mean-- maybe that's more a reflection on the grays in my hair! Ha.
It had its moments for sure, but in a weird way, I almost would have preferred a movie simply starring Phoebe Waller-Bridge, with a cameo/passing of the torch by Indy. Kind of turn him into the Marcus character. Just kick start something new.
@@TheoreticallyMedia Yeah, I get that. At least it was better than Crystal Skull :)
Thank you for this video. It's really cool!
nope still having that issue where it doesn't recognize the stuff I installed prior via the cmd terminal and all the prequisite stuff I had to do the first time and now It won't let me start all over I just get a ton of errors when I try starting over. guess I'll never be able to use it now, oh well.
I’d say maybe check with the discord. Pinokio is well worth it, as it just save a ton of time with install and configuration.
I wrote an Indiana Jones screenplay. I wonder if this technology will allow for an Indy fan film to be made
I think it can be done. That said, at this point, I recommend just changing out some details in your script and making it your own character. I mean, everyone will know it's Indy, but that's OK-- it's more of an homage, and at least you're creating something new!
@@TheoreticallyMedia I can't change it, since I essentially rewrote Crystal Skull to match the quality of the original three. I'd be happy to just make it a fan film some day so that people can enjoy it
Bonkers. Thanks,Tim.
Interestin, thank you
You're welcome!
Bro, you didn't even have to mask. You could've just used the original video and then cut into the newly generated video once it was past the wigging out dudes.
By the way, great work with your content. I think the most exciting thing coming out of all these generative video models is the democratization and accessibility of filmmaking . I am a firm believer the Hollywood is stale and indie is the future of media consumption.
If you notice, our gal was static while the dudes were wigging out. LP didn't catch her until the camera moved past those dudes-- But THANKS! It's such a crazy time. So excited for all of us to be able to make these dream projects we've always wanted to bring to life!
Do I dive in now to use this for projects or do I wait one more week for the next major step forward? 🤣
Oh, man. I live that every week! That happened on Dead Sea...at some point, you just have to jump in and start making the thing, locked with the software you have. Since things are moving so stupid fast, I say concentrate on Short (2 to 5 minute) films-- save the 3 hour Directors Cut film for when things mellow out!
@@TheoreticallyMedia It's for a music video project based on a character who sprays graffiti on trainlines. So, I think I am going to give live portrait a go to add what Gen2 can't. Starting tomorrow, by which time something new will be out 😛
This is non-commercial only so only be useful for academic research or am I wrong?
To be 100% honest, it just kinda doesn’t matter. If you make something that catches the world by storm and Netflix wants to make a show out of it, it’s gonna happen.
My biggest suggestion is: think like a creator and not like a lawyer. Just make stuff, and keep making it.
@@TheoreticallyMedia To be honest, this does matter to most of your viewers. Your content is great, but addressing such an important issue would significantly improve it. You don't need to be a lawyer to find out the licensing requirements. If this is based on Insightface (I haven't checked), you might want to reconsider the video because the developer could issue a content strike. This is a tricky situation, but there's already a lengthy discussion on GitHub involving another RUclipsr about this issue. I hope someone develops a good and truly free replacement for Insightface soon.
"think like a creator" lol.
Will this work on an old gpu? like the 1080ti? Im just waiting on the 5000 series release.
I’m pretty sure it’ll run on a potato. Might just take a lot longer to get your output
I hope I was the inspiration behind the WOOOHAAAA comments hah. Thanks for another great video. I did attempt to install pinokio before, i did something wrong and it ruined the install, i tried uninstalling it and reinstalling it but apparently there were remnants keeping the new install from running properly, I didn't know how to get rid of them and could never get it back to running. I'm hoping I will be able to finally use it in this version. Well, here goes nuffin
Haha, that probably was you! As far as Pinokio, it is now a 2.0 version, so maybe it’ll write over at this point? Yeah, that’s always annoying when that happens.
Other idea: maybe try installing it to an external drive?
@@TheoreticallyMedia yeah I made a mention of it I think in your x post as far as the woooohaaaa thing
Will this mean we can finally get a realistic version of Will Smith eating spaghetti! 😂
What level if nvda gou is needed to run this locally?
I'm pretty sure you can run it on a potato. It might just take a lot longer!
I downloaded pinokio but having trouble running it can someone please help ? Lol
video is going out of sync. anyone else have this problem or have any fixes for it? losing about 1 second every 10 seconds or so
I looked at, but all their plans want to put water marks on the output videos.
good
Damn, those head enlargments kill the effect. It looks super unnatural.
Yeah, it's gettin' there though!
thanks chinese guys
haha, they say: 不客气
It’s too bad you need a square video, but that will probably be fixed soon
Wow you know a film term!!! 👏👏👏 What a big man you are!
Yeah, I picked up a few of those terms working in the industry for 17 years.
@@TheoreticallyMedia Wow. Key PA is a respectable position.
@@Alice8000 What's your problem.......jealous a little that Tim's videos do good while yours don't do diddly squat?
Kling ai upgrade make video on haiper kling pixverse
👋
Heya Louis!! What you’re take on human driven AI acting?
@TheoreticallyMedia
It's fair progress but for from "blown away" status.
I love open source.......so that's definitely a plus! 👍 👍 👍
Only 16: 9
Just down load it and run it locally.
I’m no fan of the dinero used to be. He’s unhinged and a nut
Oh, him and Marty were LUNATICS in the 70s and 80s. For sure mellowed out in their old age, but I totally get it. Both of them are super ingrained in my head from film school though. Never met them personally, but I've heard a ton of stories. Old teacher of mine worked on The King of Comedy-- so, yeah-- For SURE have heard the unhinged stories.
played with this more than a year ago, but yes, its improved! ruclips.net/user/shortsi83o1dHXcq4
What is the point? AI is just the flavor of 2024
What is the argument here?
Hedra is better on this
Love Hedra. I’ve seen some of the stuff they’re working on for future releases, and it looks awesome.
Why are we so excited for mediocre results? Seriously, pause for a moment, and look at that, it’s crap isn’t it?
NO ITS FAKE I USE hf DEMO AND THERE NO ANY VIDEO 2 VIDEO OPTION LIES HUHH??
Uh, hardly fake. Are you sure you're on the v2v version? There are a few on Hugging Face at this point.
lip flap 💀
Oh boy. Political propaganda is gonna go through the roof with this stuff.
Yeaaah. I try not to think about that and instead concentrate on the creative things we can do with the tech. But, for sure, things are going to get pretty noisy on that front.
To be honest, I’m actually surprised we haven’t seen more of it already.
this is so so so dangerous, its stupid for it to have even been put out. like its basically designed for catfishing scams.
😂 you crack me up
Disconnect your internet immediately
Turn the switch off next to your computer
💯
Dude relax
cringe
Overall, pretty bad… I’ve seen better deepfakes!
I would like someone to explain to me why AI image generators cannot or will not do military history correctly with regard to uniforms, eqipment, accoutrements and weapons!!!!!!!!! There is a market for accurate and realist images of military history and the AI image generator that goes after this market, will be rewarded. Books, Videos, Articles, and even movies for back drop images all need accurate historical images! And there should be no reason why the AI is not educated on this when there is so much reference material online!
Because ai can only generate what it was trained on. That's like asking a Lawyer why he cannot do sewing, he didn't train in it. Places like civitai can often have historical accurate uniforms and such as they are trained specifically on that item. So to answer your question. AI cannot do everything, it can do what it's trained on. If you want accuracy for specific items go learn how to train yourself.
This would be your answer, it seems. Historians, directing the training of ai systems, for greater accuracy.
As this technology expands, it will happen soon.