I would like to see them recreate the same type of video. Interacting with somebody consuming spaghetti is a difficult video even now for AI
@phillipcoetzer8186 Says:
2:49 her feet are sliding
@CannaSama Says:
Just think of how different life will be for the childs being born right now
@ArkhamCityIsGOATproveMeWrong Says:
2:24 just love how she switches her legs around.
@past_the_event_horizon Says:
Man this is scary
@articobum3878 Says:
its starting to feel like blender is useless now
@articobum3878 Says:
there goes my job again
@ashtoncartner Says:
2:39 the girl's legs swap places and everyone occasionally does a lil' shuffle.
@alexm.a.d892 Says:
Wow this sounds like a terrible thing
@cyberjunk7258 Says:
they change a lot
@shieldgenerator7 Says:
plot twist: this whole video was AI generated
@martythebeaglenese9023 Says:
2:26 her left and right legs completely switch mid-step. Also, the feet are never stying in one spot while on the ground. Sora still has some way to go (thankfully)
@venuecam Says:
Opens Open AI: "solve world hunger"
@stephenrasberger6807 Says:
It will be funny when creators stop providing content for AI. The AI will eat itself.
@georgeuzzle Says:
I thought for sure you were going to say this whole video was AI generated at the end. I'm paranoid already.
@tubester358 Says:
For me the distinction is mainly around cool, high quality content versus production-ready content. There's always a lot more consideration from the professional artist, designer or editor than just "a cool drawing" or "smooth footage" when doing something upon request, considerations that their client might not thing of in terms of handing off the right deliverables (files/items) and additional guidelines about how to effectively use them/present them. AI can generate stuff but someone needs to be able to determine which of those generations are the best to use & how to tweak them.
AI tooling will mostly be useful for fun (personal content) and for creative professionals to use as part of their process, but the right tooling for those professionals will take the longest to develop well unfortunately, and in the meantime attention-focused social media algorithms will necessitate saturating that content market with lower quality stuff.
@Orangemushroom554 Says:
Lol now youtuber jobs are getting replaced by AI
@lestershenk924 Says:
I like it. 😊
@valentinhetzel4422 Says:
Her legs are smoothly switching sides 02:25
@mtjoy747 Says:
I imagine AI slowly gradually taking over everyone's jobs, so we can all enjoy this endless "retirement" more LOL - if some cricket players retire at 30, how about the rest of us doing the same?
@FeZaCx Says:
this is just crazy... speechless
@Mr.French.Fries. Says:
a trustable and unremovable watermark huh?
I think that's only possible if the watermark is spread throughout the video instead of just a corner.
and for that to not cause disruption in viewing experience,
maybe the watermark will hafta be a specific identifiable number and pattern of particularly distinguishable colored pixels spread throughout, say every square cm of AI generated videos,
or an even smaller area to make sure cropping a specific required portion becomes entirely useless since the smallest meaningful cropped portion will still contain that identifiable pattern that would act like an unremovable QR-like watermark that will be able to identify the software and prompt used and the generator of the image. Designing an AI that can read these watermarks will be a fairly easy task, the working will be similar to QR scanners but way more complex.
such watermarks will cause no viewing disruption since they will be on the scale of individual pixels.
ofcourse there will be prolems of somehow keeping these watermarks working across different resolution conversions but that's not really a big deal give that we've already come this far.
screen-recording will still contain these watermarks and so the only way to effectively get rid of the watermark would be to shoot the screen using another camera which is perfectly distinguishable from an actual video, so people will automatically not believe such videos given the obvious effort to get rid of the watermark.
what do ya think?
@DanielVerberne Says:
Serious question - does this tech concern CGI artists in any way? Is it a potential tool for anyone wanting to tell stories to knock up quick visualisations that could then be taken further, or is possible AI tool descendants of the ones we have available now might eventually be able to 'ingest a detailed screenplay-like prompt' and produce, for lack of a better word; a movie?
@DanielVerberne Says:
To what extent are these models producing far greater results because there's an exponentially greater sum total of material for them to ingest and 'learn' from? I'm assuming there's been far more actual improvement within the model itself than just the data available to it, but that could be part of the answer.
My understanding is that these models know nothing about humans or animals or planets or cheesesteaks, they know about language and the mapping language to images and patterns they've been given examples of, times many billion.
@Zoeincredible Says:
More job losses. We'll have to conform to the future in order to survive.
@misurebanks1538 Says:
Those "flaws" you pointed out was only pointed out because you know it's AI.....you would have never ever acknowledged the frame rates in the freakin reflections...😂😂
@Dave_the_Dave Says:
There is a downside for some roles for artists and producers and photographers and technicians. But at the same time, this technology massively increases accessibility for people who have a creative vision and limited resources to realize it. But you could say the same thing about other creative tools that have made artistic expression easier: Digital cameras instead of film. Cheap drones instead of helicopters. Red video cameras instead of huge film cameras. Cell shaded animation instead of hand-drawn animation. Imagine if young Stephen Spielberg had tools like this. The equivalent kids are out there right now and this type of technology lowers the barriers for people to make their feature length films, or new types of art projects we haven't conceived of.
I am on the side of making creative expression more accessible for all.
@unclemercy Says:
could tell it was fake when he was reading on the cloud because his hair was different and also anyone who has sat on a cloud knows that its more like a beanbag.
@josephkamo1324 Says:
Shower thought: AI doesn’t generate images or videos but takes prompts that align with a real action/event happening in another dimension🫨
@symonjamesmusic Says:
Thank you for such a great run through, explaining, exploration. Subbed and look forward to future adventures.
@SamTheDestroyerGD Says:
Fire in the hole
@K_N_Jsquad Says:
Imagine when video games get this far thats gonna be crazy
@musicfanBRA Says:
I wish you included a beautiful Sora video of 2 golden retrievers podcasting on top of a mountain. They look totally real, beautiful and lovable.
@b3n5chn31d3r Says:
For some reason, the fact that the Old Californian Western scene had a 2 leged horse made it more impressive. It makes it very clear it's not just stealing other clips... it's making every part of it. Unfreakinbelievable
@BillypilgrimII Says:
As for worrying about politicians etc. Nobody believes 'the news' as stated. Everyone's a cynic. But curiously less so, when it comes to whacky stuff like aliens, ghosts and religion. For example, the Mercury Theatre's 'War of the World's' broadcast. People get suckered in by the fantastic and conspiratorial - climate change is a 'hoax'. But are deeply cynical when experts give their considered opinion or when politicians...speak. So AI is more likely to have people worrying about martian invasions than a change of US policy vis a vis Israel.
@BillypilgrimII Says:
In the case of stock video of locations, the real world paid use cases are pretty much all tourism related and thus would have to be the real deal anyway. A big brand would always shoot its own stuff and celebs are needed to endorse anyway. The most obvious use cases are small budget ads for your local pizzeria etc.
@jan-martinulvag1953 Says:
Do you know what is AI? The spirit world. What makes it real or unreal? Love.
@meteorheartofficial Says:
I want to use this to make videos for my music. Could they make it so you can upload images of yourself and then AI can use your image, so instead of random people you could have a virtual you doing crazy things.
@joey4691 Says:
OMG... going to completion exponentially in the blink of an eye...
@joytekb Says:
How much smarter any three year old child is than two years old
There is better sound and prompt understanding coupled with emotion reading and is no more creating poop instead of not
@yahu5988 Says:
imagine what you could create with this
@hardheadjarhead Says:
When they get the voice tech down it’s going to crush voice actors.
@hardheadjarhead Says:
Anyone notice the candle had a double wick/flame?
@tgarz23 Says:
sad that the future of this incredible technology will be mostly Russian troll farms mass-producing fake videos of their political enemies worldwide
@rossmurray956 Says:
i love art theft
@Daekar3 Says:
The establishment is going to cling desperately to the ability to trust photo and video evidence. Horrible injustices will be perpetrated and your rights will be tossed onto the flaming altar of the status quo. It's not going to make a difference, though. The fight is already lost, and you will never be able to unquestionably trust a digital image, video, voice recording, or anything else again. There will be well-intended attempts to do things like cryptographically guarantee the provenance of an image or video, but they will ultimately be ineffective on a large scale.
Get used to questioning everything you see folks - you should have been doing it anyway, but now it's not optional.
@ingidraws Says:
You should have done something freaky with your hands at the end, just to mess with us ;)
@MicahWright666 Says:
Don't trust the hype. AI generated videos will NEVER render sensible imagery because it's limited in even more ways than AI image generators are. The physics will never be real. Even the videos you're praising here are full of gibberish, and they're essentially just feeding Sora the same prompt that they labeled their stock footage with, and it's just regurgitating the same thing back. Read some articles about Ed Zitron about it. It's just a plagiarism machine.
@CafePoderoso Says:
11:00 I think its sad... I feel sorry for people that work with drones and stuff, because people think that they always have a job, and I know a few people that work a lot to pay their bills with small projects. Each small project help them to endure the year till they get bigger contracts, which dont last life time... And if people are considering to use that, o well... I just feel sorry for them
@JohnSmith-ef2rn Says:
I've often said that AI images look like how dreams appear to me, or how things appear in my mind when I imagine things - the broad structure is there, but the details are fuzzy. How close are these LLMs and neural networks from sentience? I'm not saying they are, but some of the output seems very close to.... well... what a thinking being would create.
LATEST COMMENTS