OpenAI’s Sora Is Here, But...

[@daxinnthenomad]

WHAT A TIME TO BE ALIVE! ( 4:48 QUEUE DANCING SUSHI EMOTE )😃

[@axzphere03]

Bro speaking in token per second

[@BrianMosleyUK]

Needs the cerebras chip upgrade 😂

[@Ariel-z3f]

It's as if there is a full stop after each word.

[@envynoir]

😂😂😂

[@c1p0p1c]

Straight out of the from the Uncanny valley. if this is not AI, I am even more confused.

[@LORW-xd1ir]

I am an old subscriber and i like his content but I really wish he take some narration lessons 😂

[@MykolaDolgalov]

This looks very interesting and exciting. However a very sour side effect of all this is that social media has become flooded with fake images and videos, gullable people like and comment, and it's very sad.

[@maxpower4370]

The problem with text-to-video is that 80% of what you generate is garbage. When you try to generate something slightly complex, it fails miserably.

[@eduardotoscano8916]

The amount of energy wasted on those things...

[@NzeRekoRec]

And it's well known fact in any visual media that getting in to 80-90% is the easy part. That last 10-20% usually takes up the majority of time. It might be that AI video models (with current approach) can't break through that 90% barrier. Image quality has improved massively but same underlying issues are very much still there.

[@aspzx]

Which might be ok if you could do 100 iterations to get the result you want. But the token cost is prohibitive even if you are paying £200 a month.

[@elwendyrofys5944]

Don't look at where we are now, but where we will be two papers down the line.

[@v1nigra3]

@@elwendyrofys5944you could tell them a hundred times and you will still see the same comment

[@NoenD_io]

OpenAi while having most of the ai not open source is ironic

[@jonathaningram8157]

At this point I have very few interest in closed source ai.

[@mawkernewek]

or you could just say its fraud

[@Culpride]

You mean the nonprofit organisation that has a "capped profit arm" wich rakes in millions using the training data they were only allowed to use as a non profit research? :O

[@NakedSageAstrology]

Keep an open 👁️

[@TerragonAI]

What is actually the best local rtx4090 video model? Animatediff 😅 ?

[@rendermanpro]

What a time to be alive...................... .........tell that to the artists and animators, and their family. ("learn plumbing")

[@da_cat]

The biggest problem is consistency, i can barely get these AI generators to build up certain details on an image they just generated. If i ask for a sheet of metal, it gives me one, if i tell it to add rust on the edges it does all sort of weird spooky shit except putting fucking rust marks on the edges

[@gmunny46]

This sounds like it would be a job for stable diffusion and inpainting/masking, which is much more tunable and can run on a range of cards, locally!

[@mattmmilli8287]

That’s the biggest issue going on for image generation. No one really figured out a way to give it detailed guardrails and requirements yet

[@da_cat]

@@mattmmilli8287 I don't think making a full movie is going to be a real feature any time soon. You can't make a blockbuster 1h movie if the AI changes characters and stuff by itself every clip. Imagine if the main character has moles moving around it's face for the entire movie lol, start of the movie its near the lip, middle of the movie on the cheek and at teh end of the movie somehow its on his forehead

[@mattmmilli8287]

@@da_cat I don’t think anyone planning to. These are useful for like quick few seconds of B roll in KZbin videos and stuff right now

[@da_cat]

@@mattmmilli8287 yea they are very useful even so

[@bokchoiman]

We will live in AI slop world before GTA VI

[@ircluzar]

when we all become able to be film directors, there will be no film directors.

[@sapphyrus]

I still feel like I entered an alternate timeline in 2022 with how AI is so rapidly advancing when all we had was nightmare fuel pictures 2,5 years ago. Looking forward to local models!

[@kotzka4626]

If a sci-fi worldbuilder tried to explain this much progress into just 2 years of their alternate history I'd call it bullshit but here we are.

[@Octamed]

"We will all be able to become film directors". Nightmare fuel.

[@arthurfleck42]

Goodbye reality. Nothing on screen can be deemed evidentially real anymore.

[@sub0fathom]

Yeah I wish we could just go back to the prehistoric era when everything you saw online could be trusted!

[@jonathaningram8157]

We still can make the difference most of the time. But in few years (months?) it's wouldn't be possible

[@cinematic_monkey]

@@jonathaningram8157 Most people can't. Most people thought that coca-cola christmas ad was real. We're doomed.

[@florentraffray1073]

I feel like I have a sensitivity to AI generated videos that makes me feel physically slightly nauseous. It's like my eyes pick up on the fact that there is no truly rock solid and consistent 3D space and then it makes me feel kind of sea sick.

[@unnecoppro]

When the guy who regularly gets excited for the smallest advances in AI and graphics is cautious about your progress you know you messed up.

[@marcp6341]

Watch the video frame by frame, you can control the frames with dot and comma. Then you see so many inconsistencies and the whole thing doesn't look as good as it's hyped up to be.

[@JuniorDjjrMixMods]

I took too long time to release... And still, not released on my country (Brazil). When it was announced I was extremely impressed, but so much time has passed, so many competitors, that it's not THAT impressive anymore.

[@zoeherriot]

It's funny that during the part about object permanence with the city street, a guy in the background just vanishes.

[@neetpride5919]

real life has been so crazy lately, we don't even need AI to make crazy videos anymore

[@3Runner95]

The way that AI messes up is too similar to our dreams

[@TheDiamondMan_3]

My thoughts exactly. Results from AI image/video generation do seem very similar to how our dreams look.

[@NakedSageAstrology]

Merry Griftmas. 👁️ 🤖 🌲

[@tjn0110]

2:41 I appreciate the "hold on to your papers" on-screen visual alert.

[@Juan-qv5nc]

Its good academic etiquette.

[@amarug]

I am on one hand so excited to play around with it, but on the other hand I am getting really worried about the fact that are we headed to a really bland future where there is no art of any kind anymore? I mean also movies, drawings, TV, music... everything is just created on the fly, tailored to the person, to optimally milk that last drop of dopamine we have? Will it be a future where there is no more wonder and excitement about these things? I will never forget how as a teen, I saw a making of of the Lord of the Rings, and that mix of miniature models, CGI using tools that when just thinking about the math behind it made me smile, like 3D tracking, rayrtacing, all these things were in such a magnificant balance. The epic music of Shore. Also add physics simulations, where we actually solve for the physics in simplified ways etc. to approximate some behaviors. I am aware that all these things, as now AI are ultimately just "new tools", buit as I said, I felt there was a point of amazing balance and despite CGI it needed extreme skill from engineering to artistry to create anything that was impressive. I feel like this newest evolution of "tools" really blended all these wonderful boundaries, so that content will explode to the point where nothing is interesting anymore. Now it's all going to be replaced with matrix multiplications that are so monstrous that no one really knows what each layer truly does anymore. I do have the "what a time to be alive", but I also get more and more depressed, since it seems kinda lame and boring if you push it far enough into the future. When I hear a great song, I want to feel the people who made it, I don't want to question whether an AI just composed it for them and they pretend - thats all just so so lame. I am an engineer myself, so the art world is a bit removed from me, but I got artist friends who have basically given up. One friend would still get contracted for concept art in video games until about a year ago. Then it all stopped and it was clear to her it would not come back. She was not on that "super epic fame" level yet to hold her own. But she was good and could make a living. She is working as a gardener now. Kinda sad... Anyone, tell me that I am wrong, tell me why the future WILL be exciting and full of wonders. I need it.

[@mumiemonstret]

So true. The first AI videos on Instagram were so funny and exciting, and I thought "Am ''i going to get used to this too?". It only took a couple of months, and now the excitement over both those videos and other, that may show skillful tricks but also could.be AI, is dwindling.

[@Ashketchum2025-i7b]

you are right here. It's a pure challenge but, I will say wait for moment hopefully the AI at the end will need human creativity and support. Let's see what will happen in future meanwhile, I'm too afraid the generic content will full up the whole internet :((. Something needs to be done.

[@CharlotteLopez-n3i]

Mind blown by Sora's capabilities! Can't wait to see what the future holds for AI-generated videos. Super Pup is my new favorite thing

[@chriswatts3697]

I just say : Hunyuan Video from tencent - local !

[@radekmojzis9829]

DALLE came out less than 4 years ago. What am i even watching.

[@AlexJohnson-g4n]

I'm blown away by Sora's video generation capabilities! The potential for local models to offer similar quality for free is game-changing. Can't wait to see what the future holds for AI-generated videos!

[@E.T.S.]

There are already pretty decent local apps, like CogStudio, Facepoke and PyramidFlow. They can be found in Pinokio (virtual computer for AI apps). Of course the quality is not as good as online services, and it can take 10-20 minutes to render 7 seconds, but still very impressive. Another way to make unique AI videos is in Krita with Acly's plugin. In Live mode the plugin turns what it sees on the canvas into an AI rendering about twice per second, guided by an (optional) text prompts. By clicking a record button every generated preview is made a frame on the animation time line, so you can basically sketch an animation, move layers, zoom in/out and so on. You can i.e. create a Picasso-style animation in a few minutes. Acly's plugin comes with Comfy, good models and controlnets (all optional, you can also use an existing installation). Search 'Acly' on GitHub.

[@jonathaningram8157]

I'm much more hyped by hunyuan

[@sillicon8227]

I was just reading their threads, sounds super interesting. Plus is a max of 5 seconds (it appears to be 5 seconds of 720p or 10 seconds of 480p) and pro is a max of 20 seconds Plus users ($20 monthly) will be able to generate 50 clips a month. Pro users ($200 monthly) will be able to generate 500 clips + unlimited on a slower speed.

[@leebumble]

The incoherence has a dreamlike quality to it.

[@denali6935]

Yeah it’s almost human

[@cbnewham5633]

It's poor compared to other offerings. And as for "pricey" - yeah, $200 a month is ridiculous just to remove limitations that other companies are offering as standard.

[@OpreanMircea]

"this is the worst version of it" I dunno, I still remember the Will Smith eating spaghetti one

[@kellymoses8566]

These AI movie generators are going to be used to create some TRULY horrific things.

[@Dr-Zed]

Still looks too uncanny to be useful

[@mattmmilli8287]

To me, it is ready for b roll on documentary KZbin type videos that are on the screen for a few seconds at most.

[@AalapShah12297]

5:20 Damn that seems quite optimistic

[@leiferikson2210]

We can't even run good image models yet

[@john_blues]

Hunyuan Video is open source and is VERY impressive. It just take a lot of VRAM to get quality results though. Hopefully they will continue to work on it to make it more efficient.

[@design.dmitri]

as a designer I'm extremely excited, especially when the industry starts producing the tools that actually help me with my design process, rather than giving finalized products that don't make complete sense even in the details. At least I hope AI won't be able to design a fully functional revolutionary product in the near future.

[@Science_Flash]

have you tried to finetune your Voice?

[@petersuvara]

MKBD did a great review of this. Major issues with things like object permanence, lack of audio and physics, make it unlikely it will be used for anything more than short fun novelty clips in the short term.

[@AIMotionpictures-e8c]

Sora is frozen. They still don't let you create accounts. 🤯

[@phyzix_phyzix]

The bookshelf in the remix changes

[@spectrobit5554]

I'm glad someone else shares my sentiment, Sora is impressive, but it isn't noticeably that much better than it was when it was first announced, and definitely not noticeably that much better than the competition.

[@falco830]

Life’s about to become a fever dream, hold onto your paper ballsacks cause shits about to get crazy.

[@wikwayer]

Thank you professor Joshua Kanye Farid

[@dansv3872]

It all breaks down to prompt understanding . I've used them all and if you feed them a high quality picture the result will be a high quality one , but the level of distortion and prompt understanding will make the difference . I'm not very happy with the way the actual solutions are respecting the prompt , sometimes it's mostly luck . So I'm waiting for Sora

[@dustinandrews89019]

AI seems to be taking a bite out of a lot of creative stuff but not so much mundane stuff. I am afraid AI is going to do all the fun stuff and leave me more time to do the laundry, instead of the other way around. I also am anxious about massive job loss in white collar work, without replacement jobs, at least in the short term. AI will do a lot of our jobs, but no one will have any money to buy the results of the labor.

[@boltvanderhuge8711]

all these generative models are going to continue to have issues until there's a proper spatial foundation model

[@takeraparterer]

openai glazing is crazy

[@h.l.69]

Will you discuss Apollo Research’s paper about OpenAI’s O1-model? It could explain why we are seeing plateu in performance.

[@Eclyptical]

1:21 I think that's a seahorse

[@scrutch666]

....and it sucks !! all that will emerge from this tech is simple blends for stock footage and as an great overlay render filter and of course abstract music videos

[@153SCORN]

True, Open A.I is taking longer and longer to do things and X-A.I and even META A.I have perfected things that even open A.I GPT still haven't gotten around to fixing. Maybe Open A.I is unaware of where their A.I is broken? Anyway it seems that in somethings META is better equipped.

[@Jeremy-Ai]

Thank you again for all your hard work sir. :) Jeremy

[@SaumonDuLundi]

What a time to be alive!

[@kipchickensout]

DallE is so bad at most things (like consistency, realism, accuracy), I hope this one is an exception edit: of COURSE it's yet again available in 155 countries - but NOT in europe, welp it'll be only a few more months for us :)

[@Jadestonk]

Have you ever heard about VPN? lol

[@LubosMudrak]

European Union is the second best VPN advertisement in the history of humanity

[@kipchickensout]

@@Jadestonk ofc, it doesn't make a difference, I assume because I'm _logged in_

[@dantemeriere5890]

It's been at least a couple thousand papers since the AI promises of being a second industrial revolution earlier this year, so what happened? Where is it? The artificial hype channels like this try to create just doesn't work anymore. You can see in the comments that a lot more people are aware of all the artificial hype. People are now used to AI and understand the term "artificial intelligence" is misleading. People have come to comprehend just how limited it is, and the fact it's no longer improving is the main problem. All we are getting are little gimmicks sprinkled over 5 year old technology that hasn't changed to convince investors to continue wasting money, even though we hear every year from these con artists that in a few months AI will do at least ONE thing properly. The painful truth is that AI is only good for very low effort porn. That's all it has achieved, that is all it is, and if this is all it can do, it's all it will ever be. Nothing but a new tool for low effort porn, created from billions of stolen copyrighted pictures and footage.

[@_John_P]

AI view of good boys: The rear end is just the front end with a hole in the chest.

[@teamredstudio7012]

I want to see what Tencent's new model can do locally after it's been quantised

[@GD15555]

Goodbye jobs.

[@ONDANOTA]

5:32 damn! lol! those memories.... time flies

[@dhruvbnaik]

I created that video at 5:30! Awesome it’s been used here love watching your videos.

[@NzeRekoRec]

You prompted it. Creating something is whole different thing.

[@dhruvbnaik]

@@NzeRekoRec I predicted this comment, yes prompted it. If it makes you feel better.

[@Tahazif_TheCool22]

​@@dhruvbnaikedit the original comment.

[@dhruvbnaik]

@@Tahazif_TheCool22 No prefer to leave it, will let NzeRekoRec retain credit for correcting me.

[@eSKAone-]

When men create a child (by nuttin into a woman).

[@jimday666]

Impressive and scary at the same time

[@LKDesign]

I know what the Coca Cola Company is using this for.

[@TheBlueFlame112]

20 years down the line as a morning ritual. I want to create a movie of my morning dream. Then post it on my story for dreams app, and hope i make the spot of the top 100 craziest dreams for the day in my country

[@kadencoates8774]

That's such a brilliant use for it!

[@papalegba4449]

By then your dreams will be live streaming :)

[@luc8254]

Imagine this 2 more papers down the line wow...

[@ThislsYusuf]

A good usecase for these models is prototyping VFX. AI is good at imagining whimsical effects that can then be recreated in artist digital creation tools. Otherwise using Sora (or the other models) on its own can't cut it.

[@Streeknine]

So, we know they've had this for a long time... I think OpenAi was just waiting for it's competitors to catch up... like advance voice chat. They know they have the best model..and they are improving it. But they hide their models until everyone else catches up and then release them and say.. "See, ours is better". I really don't like.. I know it's a promotional gimmick to wait and show their model after everyone else releases theirs but I guess that's just how it is with OpenAI. They won't show your cards until everyone else shows theirs.

[@BudgeChem]

Let us all brace ourselves for the high-pressure firehose of cute garbage AI-generated videos posted on FB and YT, and the hundreds of millions of bots and dumb (sorry) folks in the comments insisting to the point of threats that its all real and irrefutable proof of their religion.

[@andrewwatts1997]

I just tried Runway AI. I don't know what is going wrong here but, when prompted to make a character speak. The camera just moved around a little and nothing about the face moved ? So amazing...

[@cthzierp5830]

So it's an anime generator :)

[@6teeth]

When everyone can make fantastic mobies, nothing is fantastic anymore. Just a sea of fantastic things. When very few could make them, they really stood out, now, not so much.

[@taylo692]

What's the best software to add dialogue to videos since no sound?

[@AmazingArends]

I thought your poster frame was Joe Biden at the White House Christmas tree lighting ceremony! 😂

[@Gael_AG]

I’d rather spend 200$ in a mix of runaway , kling and luma

[@Benjamin-yq8yl]

It feels like science fiction

[@tyalikanky]

It requires some optimization to work locally in real-time.

[@isekaidev7803]

I don't think we are getting this on PC anytime soon. Even image gen takes a huge amount of RAM. I imagine the amount of VRAM required to run a video gen is huge. And we got Stable Diffusion for free because the creator was naive and didn't know how to monetize(Even competitors got it for free. What was he thinking? That's not how you run a business. If out of idealism home use was made free I'd understand because not every one has beefy GPUs and go through the trouble of installing SD, so you can still monetize with a cloud service. But everyone and their mother was running Stable Diffusion services online with no money going to the company). So , it's unlikely that we're going to see a free version because of the costs involved. As I said, we weren't even supposed a SD quality image gen if it weren't for bad business decisions.

[@MeNoOther]

I am still curious, which streaming site will integrate ai video in their app first? Disney+, Paramount+, Netflix, HBO?

[@onlyyoucanstopevil9024]

NETFLIX

[@Cineenvenordquist]

Ha, whichever likes shipping hardware I guess. Amazon cube can switch to being a Pineapple Pi with an optical compute unit in it... To make sure it can run it'll come with a human energy harvester like the Rick & Morty stepmaster thing...

[@_Ape_]

Every time i get excited about this things, i go over, try the thing, get impressed, then try something i TRULY want and it comes out horrendous and I realize we are still years away from true reliability. All these systems from all these companies are incredible. But they are all terrible. We are still in the foundation building stage, we are not even close to a point were we no longer need other methods. Chat GPT still cant count days properly.

[@tideypods4808]

What are some good competetors would you guys say

[@fabiocoltro]

This is the real stranger things episode.

[@marcelbricman]

impressive feat, however, similarly to chatgpt not making you an author, sora will not make you a director. it will mainly make you into someone who pays for ai slop.

[@claudioagmfilho]

🇧🇷🇧🇷🇧🇷🇧🇷👏🏻 Sora seems like nothing more than a gimmicky concept-I haven’t even bothered trying it based on the reviews!

[@CastleRene]

Oh no, I can't do AI. I shall not watch thy video, for it is evil!

[@greyjet]

Extra-ordin-arily

[@AladinDönis]

Bro really said dragon instead of seahorse

[@dustinandrews89019]

It's a "sea dragon" in Asia, so close enough?

[@SlayerDUDE1993]

if you think the author uses AI for voiceovers you haven't seen his old videos

[@Sergeeeek]

I have, sounds a lot better

[@Ichigo_Kuro_s]

It's not here, still can't use it in EU

[@joaodecarvalho7012]

Local models?

[@richcolour]

Every AI video has mistakes, so not very useful in the real world.

[@hypercube33]

Por best boi having two asses

[@artificialanimeuniverse5063]

Man, I cannot even use it.

[@sourcecaster]

Not available in EU yet though :)

[@iamapie13]

I used to visit this channel with excitement about the new tools in 3D rendering, physics simulations and all other tools that expanded what I could do with 3D art, now when I visit this channel all I see is the looming dread of my degree becoming less useful and more replaceable

[@cthzierp5830]

Stone in focus

[@Hakeraiden]

But will it make better Coca Cola Christmas ad 😅

[@thomasgoodwin2648]

"A stroke of the brush does not guarantee art from the bristles." - Kosh 🖖🤬👍

[@nescaufe1991]

It’s here

[@Kerbacho]

Probably the beginning of the end of human creativity

[@ilbailba]

Dear fellow colors...

[@NLPprompter]

i think... lower parameter?