This FREE AI makes anyone say anything (with only 1 photo)

Рет қаралды 44,852

Ай бұрын

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.substack.com/
Find AI tools & jobs: ai-search.io/
Donate: ko-fi.com/aisearch
Here's my equipment, in case you're wondering:
GPU: RTX 4080 amzn.to/3OCOJ8e
Mouse/Keyboard: ALOGIC Echelon bit.ly/alogic-echelon
Mic: Shure SM7B amzn.to/3DErjt1
Audio interface: Scarlett Solo amzn.to/3qELMeu
CPU: i9 11900K amzn.to/3KmYs0b

Пікірлер: 279

@Potent1al Ай бұрын

Memes are about to be WILD in the coming days

@Author_SoftwareDesigner 23 күн бұрын

Right!😂

@northwestrepair Ай бұрын

Indian girl: I am Morning Dove Dave Chapelle: I am black feet 🤣

@MrzodiacBE Ай бұрын

😂😂😂😂😂

@Valsi876 Ай бұрын

I tried putting my father's picture in there but it wouldn't take it because it says he's a celebrity LOL

@laughingoutloud3713 Ай бұрын

omg he's having secrets from you

@lukmanhakim9440 Ай бұрын

Hahaha

@Michael_H_Nielsen 12 күн бұрын

this kind of defense of ones image, only works when you are famous enough. That is horrible.

@criticalme1878 Ай бұрын

At <a href="#" class="seekto" data-time="424">7:04</a> 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.

@thebrownboy1453 Ай бұрын

Amazing, reminds me of announcement by two big companies recently but they didn't allow Public access. Thanks for the information.

@theAIsearch Ай бұрын

Yeah, very similar

@yoursubconscious Ай бұрын

haha😂😂

@gridvid Ай бұрын

It's still a bit too static, but nice upgrade 😊

@theAIsearch Ай бұрын

yep, still some flaws, but a league better than existing ones! hope they increase the resolution too

@PowerRedBullTypology Ай бұрын

what especially makes it obvious is that the voices really do not match the faces

@mystictrue2400 Ай бұрын

@@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

@user-ti4tf4kn3o Ай бұрын

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

@PoppeaMusic Ай бұрын

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)

@NoidoDev Ай бұрын

I didn't try a song but voice audio from another source.

@boyboy168 17 күн бұрын

Songs should be fine to work. Beware if the part contains too much music only, and the face would be ugly.

@Dave-Galt Ай бұрын

Why are Residents of IL, TX, WA currently not eligible to use Hedra products and services?

@Ken.- Ай бұрын

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.

@AdamIverson Ай бұрын

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.

@theAIsearch Ай бұрын

very interesting! thanks for sharing

@jamad-y7m Ай бұрын

Celebrities not allowed. you are allowed to make fake videos of plebs but heaven forbid you do something to someone who has money to hire lawyers

@undefined6512 Ай бұрын

We don't want Alec Baldwin saying something that would start World War 3.

@NoidoDev Ай бұрын

I'm surprised people would even care about that more than about using it for very good looking artificial women.

@basspig Ай бұрын

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

@leanderjackiegrogan Ай бұрын

I'm inn one of the states that does not qualify to use. Why?

@fire17102 Ай бұрын

Wow! When do you think we will have this thing foss local and streaming to audio in realtime?! Probably early july haha

@Lugmillord Ай бұрын

It wouldn't even surprise me at this point. By the end of the year, cartoon characters will work, too.

@SlavgaardsManga Ай бұрын

Any other things like this I can try? would love to anime my YT avatar

@theAIsearch Ай бұрын

yes, im making another tutorial soon

@zachary3603 Ай бұрын

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?

@theAIsearch Ай бұрын

i think elevenlabs can do this, but were you asking about open source?

@zachary3603 Ай бұрын

@@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 :)

@spiritsofearth Ай бұрын

how do i put this into twitch permanently for a vtuber

@inmyfreetyme 19 күн бұрын

I gave it a try and LOVE it. Thanks 4 sharing !!😊

@playthisnote Ай бұрын

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.

@personmuc943 Ай бұрын

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?

@theAIsearch Ай бұрын

yes kind of. see this: kzfaq.info/get/bejne/jp10oc92ksmWlX0.html

@personmuc943 29 күн бұрын

@@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?

@Rishon-fy9er Ай бұрын

please teach us to make 'anime characters with lakers jersey' or any other sport jersey. appreciate your work keep it up.

@lorenakademar5267 Ай бұрын

I mean, can generate voices in other languages?

@bpvideosyd Ай бұрын

I want this for video sequences, not still images.

@heartshinemusic Ай бұрын

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 <a href="#" class="seekto" data-time="969">16:9</a>.

@cstar666 Ай бұрын

Yeah, it would be impossible to alter that in After Effects or Premiere.

@lorenakademar5267 Ай бұрын

it has voices in other languages?

@themoviesite Ай бұрын

What's the website for Hallo? Google playing hard to get.

@theAIsearch Ай бұрын

github.com/fudan-generative-vision/hallo

@themoviesite Ай бұрын

@@theAIsearch thanks.

@RICHARDCLARKNINJA 28 күн бұрын

BEST THING OUT IS VIGGLE HANDS DOWN (FOR TALKING AND FULL CHARACTER FULL BODY MOVEMENT)

@lou7139 Ай бұрын

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.

@gappsanon4869 Ай бұрын

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?

@mygamecomputer1691 Ай бұрын

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.

@theAIsearch Ай бұрын

making a tutorial now. it's pretty good!

@v.svishnu2380 Ай бұрын

does this have api documentation

@firefox8713 Ай бұрын

Lost opportunity: The company should have called itself Hydra because of the “many heads”

@jamad-y7m Ай бұрын

Or they could just call it Hedra because you can trademark that

@firefox8713 Ай бұрын

@@jamad-y7m You can copyright Hidra which sounds like Hydra

@haidara77 Ай бұрын

@@firefox8713hey thats my name!

@imqqmi Ай бұрын

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!

@Hsound Ай бұрын

Is it working with others languages ?

@leonpeeon Ай бұрын

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.

@SLSDigitalMusic Ай бұрын

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!

@theAIsearch Ай бұрын

Good luck!

@nholth Ай бұрын

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?

@Wilem35 Ай бұрын

Likely due to legislation.

@Apuat Ай бұрын

🤬

@theAIsearch Ай бұрын

i'm also curious why

@PowerRedBullTypology Ай бұрын

does it allow europeans?

@themoviesite Ай бұрын

@@PowerRedBullTypology I'm in Africa, it wanted me to log in with Google, so I bailed.

@recepk.6326 Ай бұрын

Hallo looks great, too! Looking forward to the installation tutorial you’re planning

@BenLombard57 Ай бұрын

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?

@julx97 Ай бұрын

I would love to have a free alternative to hygen video translator.

@ChristianIce Ай бұрын

Nice, too bad it doesn' work for me. Can't upload any picture, doesn't matter what. Can't also animate using their images.

@theAIsearch Ай бұрын

that's strange. try incognito or another browser

@Kali-dw5cq Ай бұрын

I can’t use gen3….. where is it????

@RachnaShrivastava-pt5vg 9 күн бұрын

can i improve quality?

@TwoThreeFour Ай бұрын

It also works pretty good for manga drawing. But for laughing or sad emotion, it's still not there yet.

@theAIsearch Ай бұрын

interesting. thanks for sharing

@twilightfilms9436 Ай бұрын

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 KZfaqrs putting their ugly faces everywhere feeling themselves celebrities of AI…..Good job!

@theAIsearch Ай бұрын

Thanks!

@SolvingTheWealthProblem Ай бұрын

RunwayML Only available on iphone... wtf

@MrErick1160 Ай бұрын

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

@gameon2000 Ай бұрын

😂 Mona Lisa looking like Putin WTF?

@theAIsearch Ай бұрын

🤯

@revenge425 Ай бұрын

I cant unsee that now 😬

@yoursubconscious Ай бұрын

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.

@DivineQuizQuest Ай бұрын

Can it look like it's singing?

@Crazy_Truth Ай бұрын

Generated video quality is very low, how can I improve it *for free

@swarlly 28 күн бұрын

Capcut Video Upscaler.

@budekins542 Ай бұрын

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!

@NoidoDev Ай бұрын

<a href="#" class="seekto" data-time="725">12:05</a> - Thanks! Looking forward to the tutorial on Hallo!

@theAIsearch Ай бұрын

Coming soon!

@DataJuggler Ай бұрын

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.

@mariomills Ай бұрын

I already jumped last year, seen the writing on the wall for Reallusion unless they incorporate AI

@DataJuggler Ай бұрын

@@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.

@vihariii4754 Ай бұрын

I'm so thankful

@CosmicHarmony58 Ай бұрын

Lazy

@vihariii4754 Ай бұрын

@@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.

@budekins542 Ай бұрын

Nothing good stays free. .

@vihariii4754 Ай бұрын

@@budekins542 except kindness

@Ken.- Ай бұрын

@@budekins542 Then I guess it be free forever.

@user-ei4ol4dw7i Ай бұрын

С собакой было забавно. Автор как ты успеваешь находить все новые сервисы? Их каждый день сотни появляются

@theAIsearch Ай бұрын

I mostly find on twitter & some private facebook groups

@JoeyBlogs007 Ай бұрын

In a few years time this will scare the heck out of anyone. 😠😠😠

@jrd3807 Ай бұрын

* months

@elmentor8919 28 күн бұрын

This exactly was already possible on mobile phones 3 years ago , and now 3 years later it is a big deal suddenly

@AIcatify 8 күн бұрын

Where and ahow?

@elmentor8919 8 күн бұрын

@@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.

@somedude5951 Ай бұрын

Impressive content. Subscribed.

@theAIsearch Ай бұрын

thank you!

@azhuransmx126 Ай бұрын

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.

@theAIsearch Ай бұрын

i predict within 6 months

@azhuransmx126 Ай бұрын

@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.

@ALWAYStheEIGHTIES Ай бұрын

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.

@ryanlarson8096 Ай бұрын

Hmm... I did this with D-ID and Eleven labs more than a year back. This one has more head movement, tho.

@tolgames1693 Ай бұрын

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?

@theAIsearch Ай бұрын

are u sure charmander is over 18? anyways, i'll make a tutorial on an open source alternative soon. stay tuned!

@tolgames1693 Ай бұрын

@@theAIsearch charmander is not a even a human, how can it be considered as underage? Who will suffer from it?

@RICHARDCLARKNINJA 27 күн бұрын

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

@SharpDesign Ай бұрын

I'm in Illinois. Apparently, citizens of here and a few other states are not allowed to use Hedra... Disappointment!

@theAIsearch Ай бұрын

I wonder why

@CosmicHarmony58 Ай бұрын

Thats a good thing. Dont fall for this rain rot

@edm3184 Ай бұрын

The ads on Instagram already use this.. it's freakin annoying

@jcy089 Ай бұрын

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.

@craigington73 Ай бұрын

Very entertaining video. Amazing technology that will only improve at a great pace.

@theAIsearch Ай бұрын

thanks!

@BringThemBack2024 Ай бұрын

I'll be back when they can sync music lyrics with some expression.

@petergedd9330 Ай бұрын

What is this thing with certain USA states are not allowed to use it? Strange dude.

@SenshiV Ай бұрын

Didn't like their terms and conditions, but thanks for the update!

@timseguine2 Ай бұрын

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.

@marksutherland774 Ай бұрын

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.

@timseguine2 Ай бұрын

@@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.

@marksutherland774 Ай бұрын

@@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.

@NoidoDev Ай бұрын

@@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.

@marksutherland774 29 күн бұрын

@@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 .............

@noop-chair Ай бұрын

Bro if you got a sponsor just don't cut the video any time and say "thanks to our sponsor " ,pls make it smooth

@TheNervousnation Ай бұрын

<a href="#" class="seekto" data-time="797">13:17</a> that is terrifying af

@theAIsearch Ай бұрын

lol

@NoidoDev Ай бұрын

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.

@apatsa_basiteni Ай бұрын

Audio visual memes? Hell yeah!

@canaldetestes4517 Ай бұрын

Hi, thanks for your video, this looks like Halo.

@theAIsearch Ай бұрын

thanks!

@Futurist_05 Ай бұрын

Great video tutorial, ❤❤❤

@theAIsearch Ай бұрын

Thanks!

@dreamzdziner8484 Ай бұрын

Wow. This looks really promising.

@spyglass1005 Ай бұрын

I love Mizu from blue eye samurai. Make her real. I want to see her.

@Anna-mv9ew Ай бұрын

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

@hatuey6326 Ай бұрын

Can't wait to see your tuto on HALLO because it seems it's available on pinokio

@theAIsearch Ай бұрын

coming soon!

@marcduchamp5512 Ай бұрын

Soon what do we need human fer

@lorenakademar5267 Ай бұрын

I'll try it!

@katokatokato4354 Ай бұрын

the import audio option doesnt work, the default audio are all americans bois, sucks tbh, no kawaii loliPop voice whatsoever

@theAIsearch Ай бұрын

yeah the tts isn't great. sounds too robotic. try to do voice to voice using RVC then upload the audio to hedra

@NoidoDev Ай бұрын

I made a video with a imported voice made with a cute Japanese voice, talking in English, and it worked nicely.

@anthonymannwexford Ай бұрын

When this becomes 'The Face of AI', everything will change. It will be amazing.

@context_eidolon_music Ай бұрын

Whoa, thank you!

@theAIsearch Ай бұрын

no problem

@HalkerVeil Ай бұрын

"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.

@SS801. Ай бұрын

Nice btw why do you not give hearts to comments

@freakingerror1037 Ай бұрын

Holy shit this channel hepls me alot

@theAIsearch Ай бұрын

😃

@tripperm10011972 Ай бұрын

When I tried, the teeth looked like they came from The Big Book of British Teeth.

@mickkollins Ай бұрын

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

@chris1.2 Ай бұрын

as animators we learned that people blink on average every 15 seconds. ai disagrees obviously.

@JerrySmales Ай бұрын

I love it!

@darrell20741 Ай бұрын

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.

@Ken.- Ай бұрын

You only need an email address that you can generate on the fly.

@darrell20741 Ай бұрын

@@Ken.- Nothing is truly free, I had to pay for my face.

@incyphe Ай бұрын

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".

@theAIsearch Ай бұрын

their img gen model is probably very limited and censored. best to use SD and upload your image