Make anyone say anything with only 1 photo
Вставка
- Опубліковано 4 жов 2024
- Hedra AI review. With just 1 photo, make AI deepfakes, free & unlimited usage.
#hedra #ainews #deepfake #avatar #aitools #ai #agi #singularity
Thanks to uPix for sponsoring this video: Generate AI selfies in just 1 click. Turn yourself into a superhero, anime character, and more!
upix.app/
Hedra AI:
www.hedra.com/
Newsletter: aisearch.subst...
Find AI tools & jobs: ai-search.io/
Donate: ko-fi.com/aise...
Here's my equipment, in case you're wondering:
GPU: RTX 4080 amzn.to/3OCOJ8e
Mouse/Keyboard: ALOGIC Echelon bit.ly/alogic-...
Mic: Shure SM7B amzn.to/3DErjt1
Audio interface: Scarlett Solo amzn.to/3qELMeu
CPU: i9 11900K amzn.to/3KmYs0b - Наука та технологія
I tried putting my father's picture in there but it wouldn't take it because it says he's a celebrity LOL
omg he's having secrets from you
Hahaha
this kind of defense of ones image, only works when you are famous enough. That is horrible.
Memes are about to be WILD in the coming days
Right!😂
Amazing, reminds me of announcement by two big companies recently but they didn't allow Public access. Thanks for the information.
Yeah, very similar
haha😂😂
Indian girl: I am Morning Dove
Dave Chapelle: I am black feet 🤣
😂😂😂😂😂
It's still a bit too static, but nice upgrade 😊
yep, still some flaws, but a league better than existing ones! hope they increase the resolution too
what especially makes it obvious is that the voices really do not match the faces
@@theAIsearch First why you lie ?
ITS NOT FREE
ITS NOT FREE
WARNING. YOU MUST REGISTER WITH YOU AMAZON AND GOOGLE ACCOUNT AND THEN YOU CAN NOT DELETE YOUR ACCOUNT SO IT'S NOT FREE AND YOU PAY MONEY
DONT GO IN THIS MONEY TRAP
Moving too much in my op, if we could control the exaggerated movent...However its the best auto lip sync tool I know of out there. Don't worry they are learning from our critique
I'm waiting for the day when I can input pages of manga and previous anime episodes with voice actors and soundtrack and sound effects and have it spit out new episodes of that anime with the original voice actors all based on the new chapters of manga that I scan in. We may be just a couple of years from that point
I gave it a try and LOVE it. Thanks 4 sharing !!😊
BEST THING OUT IS VIGGLE HANDS DOWN (FOR TALKING AND FULL CHARACTER FULL BODY MOVEMENT)
Again I have to say that VIGGLE not only looks much more realistic than this but also the whole body can move - you take a photo and a video and click generate and boom it's awesome
It's not perfect but it's pretty good. I think I can work this into my workflow. They actually sing pretty well also. Thanks for the update, awesome tool!
Good luck!
The fact that the image/ visual is square makes it totally unusable for now. This was also the case with the unavailable competitors. (The EMO stuff is still better, but not open to public yet.) To make good videos, you need at least a movie ratio like 16:9.
Yeah, it would be impossible to alter that in After Effects or Premiere.
Lost opportunity: The company should have called itself Hydra because of the “many heads”
@@jamad-y7m You can copyright Hidra which sounds like Hydra
@@firefox8713hey thats my name!
It also works pretty good for manga drawing. But for laughing or sad emotion, it's still not there yet.
interesting. thanks for sharing
I like your channel because you don’t have a giant ego putting your face everywhere, even the thumbnails. Your channel is all about showcasing the product. I’m freackimg tired of other UA-camrs putting their ugly faces everywhere feeling themselves celebrities of AI…..Good job!
Thanks!
😂 Mona Lisa looking like Putin WTF?
🤯
I cant unsee that now 😬
I want this for video sequences, not still images.
I found it weird signing up for this program that people in Texas, Illinois, or Washington State are not allowed to use Hedra, why is this?
Likely due to legislation.
🤬
i'm also curious why
does it allow europeans?
@@PowerRedBullTypology I'm in Africa, it wanted me to log in with Google, so I bailed.
To ''AI Search' youtuber: Your voice is very good. Are you using an AI voice(and if so, which one?), or is that your real voice?
This tech could be great for movies that need to heavily de-age an actor or re-animate a deceased person. Unfortunately it does seem like it will eventually put some vfx teams out of work that accomplish the same thing with 3d models. This tech looks like it is probably 1/100 or 1/1000th the cost of an entire vfx team dedicated to shots involving de-aging or even re-animated deceased actors. Pretty crazy.
Why are Residents of IL, TX, WA currently not eligible to use Hedra products and services?
Illinois, Texas, and Washington are the only states that have regulations on the books about biometric identity collection, the use of that information, and the consequences for noncompliance.
14 other states have introduced legislation this year.
I would love to have a free alternative to hygen video translator.
Interesting tool. I am looking forward to the day that I can use an Avatar of myself in Zoom, with real time live audio, being able to replace my webcam capture and audio while doing face to face meetings. (simply replacing my webcam image , working in VR using IMMERCED without appearing on my webcam with HMD on my face. appearing normal to my audience) Maybe the technology already exist?
I'm not sure if Hedra is based on Hallo, but I've been experimenting with Hallo recently. The result is kind of mixed bag. It has potential, but there's a lot of issue that needs to be fix. Whenever I put green chromakey in the background, it immediately deteriorate and becomes unusable. The mouth movement gets weird sometime and it really depends on what photo you use. While technically unlimited length, the longer it renders, the more it gets messed up. It's promising, I'm hoping to see an improvement soon, but it's a great start.
very interesting! thanks for sharing
С собакой было забавно. Автор как ты успеваешь находить все новые сервисы? Их каждый день сотни появляются
I mostly find on twitter & some private facebook groups
This is amazing, I personally wouldn’t need it for anything however I’m impressed with the technology.
I’m looking forward to the locally installed off-line model that’ll take one picture and animate it based on a suggested action.
making a tutorial now. it's pretty good!
please teach us to make 'anime characters with lakers jersey' or any other sport jersey. appreciate your work keep it up.
Hmm... I did this with D-ID and Eleven labs more than a year back. This one has more head movement, tho.
In a few years time this will scare the heck out of anyone. 😠😠😠
* months
12:05 - Thanks! Looking forward to the tutorial on Hallo!
Coming soon!
When I first saw VASA 1 I said to myself the greatest day in the history of A.I will be when this is released to the public. .but it looks like Hedra comes a close second!
Thanks for showing us the app. Timeline for Hedra to reach the level of Emo Portrait or Vassa? 🤔 6-9-12 months? Never? That depends on its CEO and how much believe in the project.
The true future for those talking heads ANNs will be to give life to Chatbots like GPT4-O and Agents like GPT5, Personal Assistants, NPCs videogames and Movie Dialogs.
i predict within 6 months
@theAIsearch they have to train a network only for Animals, Comic, Anime and Pixar 3D inputs, or the creepy human hiperrealistic mouths will persist forever hahaha.
Wow! When do you think we will have this thing foss local and streaming to audio in realtime?! Probably early july haha
It wouldn't even surprise me at this point. By the end of the year, cartoon characters will work, too.
You ever thought about if you took a video and ran each frame through this and picked each frame position it would work on video. Well maybe. I guess you may have to run it through a morph type of engine to smooth out conversion. But there’s a better way this would be perhaps kinda resource hungry.
The ads on Instagram already use this.. it's freakin annoying
Nice, too bad it doesn' work for me.
Can't upload any picture, doesn't matter what.
Can't also animate using their images.
that's strange. try incognito or another browser
Great vid man. One quick question, not sure if was this channel? But one AI tool had the ability to upload a sentence or paragraph of text and speech, but change the words slightly. It sounded EXTREMELY GOOD, but quite niche because you could only slightly change. Have you come across this?
i think elevenlabs can do this, but were you asking about open source?
@@theAIsearch Yeah there was a tool that has insane results. Elevenlabs is great, but that's just cloning which sounds slightly off. I'll come back and let you know if I find it :)
I'm in Illinois. Apparently, citizens of here and a few other states are not allowed to use Hedra... Disappointment!
I wonder why
Thats a good thing. Dont fall for this rain rot
The eyes though, it would be great if you could adjust the number of blinks per minute or something, and what's up with those half blinks? As if they all have something in their eyes lol!
I'm just waiting for someone to finally be able to realistically recreate the lost classic Doctor Who episodes that were wiped. I've seen people attempt to add the audio to bts photos and tele-snaps using AI but it was months and months ago. I can't find anything that has been attempted with the recent tech released.
Hallo looks great, too! Looking forward to the installation tutorial you’re planning
This exactly was already possible on mobile phones 3 years ago , and now 3 years later it is a big deal suddenly
Where and ahow?
@@AIcatify You have to jump into a time machine and travel 3 years back and see for yourself. Every Mobil app store on android had several options to do these same functions.
At 7:04 the voice says to check or search Aros in the voice library for other voices. How do you do this to find other voices to use. ( Just getting started with my education to AI ) Thank you.
In all of these models so far, the teeth change shape and size when they are talking. To me it is both extremely obvious and super creepy. Especially when coupled with these voices that are also slightly the wrong side of the uncanny valley.
A year ago this was not possible, of course there are issues, it's very easy to criticise, can you develop anything remotely close to this ? NO. You clearly do not understand the complexity of all the movements. Take the teeth you harp on about. The AI has to basically guess, because it has nothing to go on, so when the mouth moves it has to predict where and how the teeth look, not easy.
Be thankful you can even access this and I'll tell you the only reason you can, to improve and train the AI. The tech, be it animation, music is being fine tuned, that's requires vast amounts of data, we basically provide that data by using it. Once they have the data and some huge coporate body buys it, you're not getting anywhere close to it. Do you seriously think movie studios will allow ordinary people to make decent films and take away their revenue. They will wait til the tech is mature and buy it, same for music etc enjoy it while you can and quite money because in 5 years then you'll really have a reason to moan.
@@marksutherland774 Since when is criticism not allowed? Get off your high horse.
It is possible to find something impressive and at the exact same time find fault with it.
Fun fact: Mentioning something one single time is not "harping on" about it.
Won't go into the rest of anything you said because you assume like 1000 things about me that are first: not true, and second: even if they were you would have no way to know that they were anyway. For what it is worth, I work in AI tech, but that is actually irrelevant to this conversation and also none of your damn business.
Or do you find being ridiculously rude and disrespectful to someone you know literally nothing about for making a benign comment appropriate behavior for an adult to have? I certainly don't. Have a good long think about what has happened in your life to make you think it was ever appropriate to write a comment like that to someone.
Don't bother responding. I already muted you.
@@timseguine2 oh dear, did someone hurt your feelings. Please tell me what AI company you work for, so I can avoid it like the plague, isn't SD3 is it ? the one that has failed miserably.
@@marksutherland774
Yeah, the vitriol in your answer was uncalled for. He doesn't need to understand how it works at all, he was criticizing the result, and just pointing out the flaws.
@@NoidoDev He said he works in AI, up to anyone if they believe that or not of course. So if that's true, he'll be fully aware of how it works and the complexity of animating a still object with it. Essentially the AI has to be like a cartoonist and draw out every frame individually, try it, see how not easy it is. Vitriol, no, telling it how it is, yes. He's entitled to his opinion, as I am mine. Indeed there was more "vitriol" in his first post and reply than any of mine. Nevermind very soon there will be something far more serious to concern yourselves with .............
Hedra often works quite nice, but the more something it's supposed to look like a doll, the weirder it looks. It should adjust more for age, weight and non-human characteristics - I mean through a prompt.
Didn't like their terms and conditions, but thanks for the update!
I was not ready for the voice at 2:00.
What is this thing with certain USA states are not allowed to use it? Strange dude.
Bro if you got a sponsor just don't cut the video any time and say "thanks to our sponsor " ,pls make it smooth
I've tried to upload audio (a part of a song 15 seconds) but it doesn't work. Do you know if songs don't work? (It's my own song)
I didn't try a song but voice audio from another source.
Songs should be fine to work. Beware if the part contains too much music only, and the face would be ugly.
"I herby represent that no imge, video or other media I upload will contain images of any individuals other than myself."
So you just broke their terms by using images that were not yourself.
Absolutely... NOT using that.
All of these video models very quickly forget the unique features of the input and create something very average instead, which is a bit sad
There really are ghosts in the machine...
Hedra is pretty cool. I have to rethink my time with tools like IClone and Omniverse, as I think in the near future we won't need digital content creation tools. I had to check a box to say I am not in Texas. I also had to fabricate (sounds better than lie) that the image is me. I realize they don't want to be sued, but an AI image doesn't care if you use its likeness.
I already jumped last year, seen the writing on the wall for Reallusion unless they incorporate AI
@@mariomills I have lots of things I don't like about Reallusion. Half the new features cost as much as the software. With IClone 7 I spent a bunch of time writing python scripts, then in IClone 8 they all break, without any help converting. Their devs are Chinese, so while brilliant, lost of things are lost in translation.
But is not free.
I can’t use gen3….. where is it????
I love Mizu from blue eye samurai. Make her real. I want to see her.
Audio visual memes? Hell yeah!
The eyelids get often messed up when blinking and it's quite noticeable. Btw it won't work well with v-shaped Asian faces e.g: the anime characters or any other v-shaped Asian face. Tried this with my girlfriend's face and now I'm getting nightmares. I guess the AI was trained mostly on Western style faces, paintings and images.
I mean, can generate voices in other languages?
i try hedra, that one sucks,is it anything better?
Here's a hack so you can use a celebrity look alike..Just save the original still image and NOT use the celebrity name. call it bozo or whatever the uplaod to lip sync site..seems to work for me
RunwayML
Only available on iphone... wtf
What can possibly go wrong?
absolutely nothing 😃
I understand that this is a rhetorical question, however, this technology is far from new, there was already a case when the deepfake was so real that it had to be deleted from everywhere, so as not to mislead people, I think in the future there will be a profession where people will check video/audio for the presence of AI
Wow, this is going to be crazy!!😂
nothing, i cant use a single image because everyone is a celebrity
@@ritpop right we're all popular lol 😅
Wow. This looks really promising.
BRO THIS IS AMAZING!!
I've been waiting for something like this for a long time!!!!!!!!!!!🗣🔥🔥
YEAAAAAAAAAHHH🔥🗣🗣🗣🗣🗣🔥🔥🔥🔥🔥‼️‼️‼️🥶🥶🥶🥶🥶
😃😃😃
can i improve quality?
I tried it, uploaded an image of a charmander and it said failed safety check: "underage".
Wtf?! Maybe can all these cloud AIs stop to censor anything?
are u sure charmander is over 18?
anyways, i'll make a tutorial on an open source alternative soon. stay tuned!
@@theAIsearch charmander is not a even a human, how can it be considered as underage? Who will suffer from it?
I'm so thankful
Lazy
@@CosmicHarmony58 bless you darling, thank you for being here on this planet. I love your username, its so mesmerizing. may you live in peace, may you live happily.
Nothing good stays free. .
@@budekins542 except kindness
@@budekins542 Then I guess it be free forever.
how do i put this into twitch permanently for a vtuber
Anyone else having issues with the voice? I tried to use the British voice for a 5th time but it rendered the American voice 3 times. Moreover, the American voice copy's pitch changed each download.
So, ya, still in Beta.
awesome i can create speaking character for my channel.
Imagine not needing to sign in and give up data, but then you would be asked to pay in money, not in personal information, but of course that is wrong, you will give up financial information. I will just look and keep going, and say hi to a real person on my walk.
You only need an email address that you can generate on the fly.
@@Ken.- Nothing is truly free, I had to pay for my face.
13:17 that is terrifying af
lol
AI moderation on content is illegal censorship. It shouldn't matter what age perceived on the face is in the photo.
If it's not a government doing the censoring, then it's perfectly legal.
Sadly none of those does realtime by capturing a performance of a person in front of a webcam or DepthCam
I've tried with my picture and changes all my face almost too animate it, also the lips and teeth are completely fucked. Very different results from the ones you show here
Generated video quality is very low, how can I improve it *for free
Capcut Video Upscaler.
it has voices in other languages?
Can it look like it's singing?
When this becomes 'The Face of AI', everything will change. It will be amazing.
as animators we learned that people blink on average every 15 seconds. ai disagrees obviously.
does this have api documentation
I'm inn one of the states that does not qualify to use. Why?
As an autistic: "this robot has more facial expression than me" (opens whiskey bottle)
Can't wait to see your tuto on HALLO because it seems it's available on pinokio
coming soon!
Any other things like this I can try? would love to anime my YT avatar
yes, im making another tutorial soon
until this is stand alone app/program we can download and use off line freely, it is nothing.
You should try prompting "Chinese woman in her early 20s with brown hair". And change Chinese to Korean, Japanese, Vietnamese, etc. They all look exactly the same. It's pretty amusing. But in all fairness, you get the same result if you tried "American woman" vs "British woman".
their img gen model is probably very limited and censored. best to use SD and upload your image
I'll be back when they can sync music lyrics with some expression.
Can AI remake an existing instrumental and make it sound better and play it in a different style or is that not a thing yet?
yes kind of. see this: ua-cam.com/video/X7DnlE-c2eI/v-deo.html
@@theAIsearch Thanks a lot. I'd like to use music tools like this offline with local installation, just like RVC and Stable Diffusion. Are there any?
Please put D-ID put of business. Their API is highway robbery
Very entertaining video. Amazing technology that will only improve at a great pace.
thanks!
I'm from Texas so I can't use it.
ha ha
Proxy server.
i wonder why
@@theAIsearch
Me too. It could be that if people might be able to create pr0n, or use a image of a public figure, or that it says something which is illegal, the penalties would be too high for them.
It's pretty useless, actually, because none of them look like actually talking people who talk like talking people. None of the faces transport any feelings and head movement doesn't actually match most of the time. Eyebrow movement isn't there at all, which turns everyone into zombies. For what it's supposed to be doing it's just as crap as all the others.
When I tried, the teeth looked like they came from The Big Book of British Teeth.
It’s bad quality. It doesn’t look like the demo videos shown here.
The heads don't move as much in my videos. And it doesn't always work well, but I'm only using the free version.
Soon what do we need human fer
Is it working with others languages ?
yeah, it doesn't really care what the actual language is. I guess results may vary depending on language but for me it worked not just with English.
Now make this realtime with chat gpt 4o level of recognizing emotion and gives back human-like responses!
great idea!
Imagine this technology in year 2030
or next year
@theAIsearch That's the thing. I wish this would take 10 years. It feels like things are going to happen tomorrow. I think the majority agrees that it's going too fast.
Yea but you can still tell it’s not real
@@PaganSkye wait for tomorrow :D
by 2030 we will probably have access to an ai world