Inside Scoop On That Sora Commercial
Вставка
- Опубліковано 7 сер 2024
- Let's dive deep into the first commercial ad spot created using Open AI's Sora. Opinions are mixed, and I decided to put on my detective hat to give you a full analysis, plus info you won't find anywhere else.
I've also got a look at the latest from Kling, the new AI Video Model, who have a new Image To Video feature, and it's looking pretty good. Rounding out with a detailed workflow to make your AI Video really shine!
What’s Inside?
Ad Spot Analysis: We take a detailed look at the Toys R Us commercial created with Sora, breaking down the visual effects, CGI elements, and more. 🧩🎥
Exclusive Insights: Hear about my conversation with Native Foreign, the creative house behind the ad, and learn some behind-the-scenes secrets.
Tech Deep Dive: Check out the latest updates on Kling's image-to-video feature and a new quality bump workflow. 🔧💡
LINKS
Krea.AI www.krea.ai/apps/image/enhancer
Topaz Video Upscaler: topazlabs.com/ref/2518/
00:00 - Introduction to Sora's First Ad Spot
0:59 - Sora Toys R Us Spot
02:27 - Toys R Us Commercial Breakdown
4:04 - CGI and VFX Analysis
4:34- Nitpicking some details
5:47 - Insider Info on the producion
6:51- Where is Sora's Image To Video?
7:26 - Why keep talking about Sora?
8:05 - Kling's new Image to Video Update
8:30 - I pass the Will Smith Test
9:10 - Kling Cinematic Action Scene
9:55- Workflow Walkthrough
12:00 - Kling Reached out - Наука та технологія
8:59 Eating pasta like that without twirling the pasta inside a spoon holding it with your fork is about one year of prison or community services here in Italy. 😂😂
Haha. Meanwhile the hamburger example is EXACTLY how the US rolls!!
Thank you so much for keeping us updated! These are exciting times. :)
Good to see you back in your proper studio. Kling access sounds great! Looking forward to the update.
It’s so good to be back!!
I do like the detective shots. Looking good.
Great breakdown and super appreciate the workflow suggestions for upscaling and colour correction, the difference after that processing is substantial!
Thanks!! People often ask if Topaz is worth it, and I’ll say- kinda? I mean, it really does take things that extra 5%, which can often be make or break with AI video:
Krea’s creative upscaler is really promising! A few more control sliders and I think it’ll really be there! Right now it can be a bit heavy handed and hit or miss. But, when it hits, man does it hit!
Great analysis. One of the best AI channels around at the moment. Deserves more subscriptions.
Thank you for making this video. I think that commercial looks kind of cool. It's a start! I've been posting tons of videos that I've created with my own artwork and Midjourney and mostly Haiper. All with Suno music. I'd like to start doing some experimenting with Krea, but it doesn't work too well on the phone or tablet. Sadly, my computer died recently and I don't have a new one yet. I'm still trying to create as much as I can using my phone and tablet until I can get another one. I can't wait to get my hands on something like Kling! 😁
the irony of not using the J sound of Jeffrey for Giff and immediately use the J sound of Giraffe 😂
Haha, valid! Point remains, it's GIF!!!
Oooh, was hoping someone would deep-dive on this. Nice!
Thanks man!! As soon as it dropped yesterday, I put on my sleuthing hat!
Toys R Us ... Where an A.I. can be an A.I.
: )
Haha! Now I want to go generate that in Udio!!
Appreciate the breakdown of that com Tim!
The commercial was excellent and inspirational, but thanks for your precise analysis of how it was probably made and designed. Keep up the good work. The benchmark has moved, and this is in the early days.
lol, studio A for the win! Thanks Tim, great video as usual.
How was working on a project today and I literally couldn't find any stock footage for what I was looking for can't wait for sora
Cheers Tim, that spaghetti test was impressive!
Not bad, right? If only AI me had some table manners!!
@@TheoreticallyMedia is that even a real thing now?
Nice substitute audio on the Toys Я Us ad spot, Tim. Also, I never got the whole J sound thing with gif, either. GIF stands for Graphics Interchange Format, and Graphics has a hard G sound last time I checked, so thanks for continuing to set the record straight on that. 👍🏼 And wow, shots fired at Temu (deservedly so). 😎
Haha, I was going back and forth between Wish dot com and Temu. Ultimately, I find Temu more obnoxious!
And YES! Gif forever!!!
cool vid wasn't even aware of this
The Sora thing just happened yesterday, so you didn’t miss it by much!
yeah thanks for keeping these coming! Love your videos and to the point style. FWIW one of the videos I watched from a Chinese guy who used Kling said it's actually pronounced K-Ling. But nobody is saying it correctly. So he's either trolling or knows what he's talking about XD
That is very impressive!
Do you know anything about Hedra AI? (if you covered it in this or another video, apologies).
I am always curious about new tools, but I like to know more before jumping in to using or supporting them.
So, I missed the window on covering Hedra's launch. That was my bad, considering I had a great call with them right before they dropped. I even tested it in advance for a video I was planning to make, and then...well, just a bunch of regular life stuff happened, and by the time I had a chance, it felt like most of you already knew about it.
On the plus side, I have seen what they're building toward, and it's pretty impressive.
Good people, small team. Worth support in my opinion! I'll actually be covering them soon on the channel!
Great video Tim as always, looking forward to seeing the Kling video. The Toys R Us advert wasn't bad but as you say a fair bit of 3d assets used - but that's where I see AI tools helping and making thinks quicker and easier (in time). AI is just another tool after all and combined with a 3d workflow then I think you can get some pretty good results.
On a side note - anyone know when Runways Gen 3 is being released? I'm looking forward to trying it out :D
Oh man, literally feel like a kid hunting around the house for presents with Gen-3. No word yet! But looking every day!
@TheoreticallyMedia Hope it not gina be a Sora 2 and not materialise? Maybe theybpreppeing the servers so it nit crash or take hours like Luma Dream Machine.
Your tweaks in Topaz really improved it a lot! Maybe with one more AI video generation it'll almost be acceptable. By the way, which ai did you use for your strikingly handsome detective character video?
I've learned so much from this man regarding AI applications.
definitely my fav ai chanel
i definetly agree Sora is the Bench mark as of now to llok upon for other VG. But i still belive we wont get access top Sora easily in any coming days. so my work flow will be major way focused on tools availabel eaisly like Luma, Gen-3 and Kling. BTW Tabi here. hehee
For sure looking forward to you checking out the Kling access. It's great that they are aware of us Westerners wanting to play around with Kling. It's good for us to have an alternative and good for them as all of our generations will surely give them valuable data to improve the models down the line also.
You look great as an Ai detective sleuthing through a black and white film noir setting. Excellent work! Edit: Gif, NOT Jif!!!
So cool.
Honestly that toys r us was awesome
I think the key to the Sora commercial is showing it is a tool. In this case, several tools were likely used: Sora, After Effects, Cinema 4D perhaps. To me this is not a bad thing, and not surprising at all. Just one more step, using another new tool. A powerful one, yes, but a "does everything with no effort" app? No. Not for some time at least.
Oh. 100%. I look at it as a raw material generator. You’ve still got to prep, sculpt, and cut to get anything worthwhile out of it!
Being a voice actor and creative this is a threat to careers like mine. Now I’m thinking, do I jump on board before it’s too late.
To be clear, I don’t think it’s a threat. Big disruption? Yes, for sure. But it’ll eventually level out.
The way I see it, it’s like 1983 and Tron (the original) comes out, and you’re thinking: maybe I should get into this computer graphics thing…
I think at the very least, you should jump in, just to know what the current tech can and can’t do. It’s hamstrung, for sure- but that’s actually great for you, as you’ll know how to perform and make things it can’t.
I’ll say, currently AI voice is not great at multi emotional reads. You can’t give it a monologue and have it bring energy up and down. Basically, it can’t act.
But yeah, for sure jump in and experiment. Just because it’s interesting and fun. Yesterday I caught a Radiohead song while driving, got an idea in my head. Came home and knocked this out in about an hour: x.com/theomediaai/status/1807168440379023522?s=46&t=Gezvp5mbBNzrK3tbLcCFzA
Omg. The character concistency… 😳 Give me access to SORA! Im curious how it works tho, are they just continuing from the previous generation and thats enough for SORA as reference or does SORA create a model within the model? Or is this video only partly SORA generated?
GO TIM!!!!
8:25 hey! That’s me! 😁
HA! It is!! Thank you again so much for that! My wife walked by as I was cutting the video and was like "Accurate" haha
Forget Sora, it's all about Studio One! But seriously, great info Tim! btw Dream Machine's Image to Video is not so great imho, the enhance feature on anime/painting images takes way too many liberties and can't be dialed down. Also, Topaz is worth it!!!
Wonder if you can generate xi eating honey in kling?
Haha. Something tells me you’d be getting in an unmarked van late that night if you did.
That tweet you showed says 700 generations *with 55 chosen clips*. That's not a 700 to 1 shot ratio. It's 700 to 55, which is *13* to 1.
You are much better at math than I am! (To be fair; I think Paul T said 700:1 in an interview, and that’s the number that has been stuck in my head)
I believe that in the next 12 months will be available models that will generate 30 seconds of video. And they will become more affordable. What I've realized over the last 2 years is that every 3 months something comes out that pushes neural networks forward.
So everyone have a great generation.
Yup! I think of it in game console generations. We have now solidly entered the PS2 era.
When we all hit Sora level, I’ll call that the PS3 era.
I promise you that your timeline is over-estimated by at least 6 months.
Krea enhancer can't really handle mouth/lipsync movements, so it's a big problem if you have characters talking or singing.
Topaz vs Magnific?
Kling vs Luma Dream Machine (e.g. is it worth trying to get access to Kling)?
Gonna take the hit and try to get access to Kling today. I’ll go through all the steps and let you know how confusing it all is!
I believe Kling is free for now, so there is that!)
Personally I think the dream like quality of AI suits it really well.
In the cinema, too, from scene to scene to the hero, different glasses, different umbrellas or paintings on the walls have changed
Yeah, constancy is a real problem for Text To Video. Undoubtably Sora would be a real powerhouse with Image to Video, but as of yet-- no one can play with it.
@2:33 I'm not sure that Geoffrey is a "100% CGI character". I think it's more likely to be a form of control net and/or a rendered 3D image used as input for a form of img2vid.
Sparkles I would agree are AE or similar, added in post, and the colour grading etc as you say is likely to have happened post gen, because why wouldn't ya? :)
Yay welcome back studio A lol
It is SOOOOO good to be back. Made the most of Studio B, but man-- I really felt so boxed in there! It was a massive pain, but the whole project was getting the ceiling insulated-- Studio A is basically a giant barn with a 20ft vaulted ceiling. It's super cozy in the winter/fall-- but, STUPID hot in the summer.
But, not anymore!!
So-- yeah, the Studio B era was a rough go, but at least you won't see me pouring buckets of sweat in the next few videos!
Awesome
please drop a hint at the kling sign up workaround :)
He said at the end of the video that he will test the workaround and create a video or post it inside the community feed.
I received a message saying that I can use the new voice model, but I need to subscribe to the Plus plan!! 😭
Oh man, I'm ON the Plus plan and still haven't gotten that message! I'm convinced I'm on an OAI blacklist because-- well, half the time I'm praising them, and the other half I'm firing shots at them.
We have a complicated relationship!
(I'm kidding about the blacklist. I'm 100% sure Sam Altman has no idea who I am)
We also don't say "jraphics" (in jraphics interchange format)
Dude, what model is that bass?
It's one of the Ibanez SR short scales. As primarily a guitar player, I LOVE the short scale! You can really rip on it!
your theory about those being assets added in could be false. just let people know it's what you think. i think there is a big chance you could be wrong because SORA was able to generate videos with high consistency and even animation 3d cartoon like sample. its very obvious that sora understands such contexts and can easily imagine these flying toys into the video. SORA generated a dog with physics that is indistinguishable from real world physics, and i dont think it would be hard for SORA to generate toys.
and also transitions that look like video post processing effects.
the future of ai video will kill most editing tools, because manual editing effects in the end is a visual thing and it can be imagined into the video by these ai tools if you describe it to them
Well, I'll say-- although not mentioned in the video, I went over this with a pro-VFX buddy of mine. And, maybe I don't mention it on the channel enough, but I do have over 20 years working with WB/Disney/Fox etc on productions (non-AI, obviously)-- so, I've got a pretty good eye for spotting assets.
In the toys example, I mean-- really take a look at that toy shelf. Most of them are kind of globby messes, with the exception of two or three of them, which look highly detailed. If I had the time, I'm 90% certain that you could find those exact models on Turbosquid.
Again, not bagging on any of this. As I mentioned in the video, they had weeks to put this whole thing together-- from concept to finish. That's a big task.
I actually live next to a Toys R us. Like a still standing store (Canada)
Also thanks for the vid
Oh man-- no one tell them! They might have not gotten the memo!!
Also-- please tell me there is a Tim Horton's nearby...and if so, I'm on my way!
@TheoreticallyMedia oh those are like Mcdonalds. You can find a timmies literally anywhere you look.
That's odd.. There must be several different dubs of the audio for the commercial going around, the one in your video sounds different than ones I've seen.
"per usual the original audio is muted"
Yup! Next time I'll put a graphic on the screen. A lot of people seem to be missing it...
i like this channel
Ahhh Studio A...like coming home LOL
There’s something repulsive about the advert, in that it literally has the opposite effect of attraction. It’s like a strange stoned horror movie made of stock footage.
I feel the same about generated text, its uncomfortable to read before you find out its ai generated and after you find out i get a metallic taste in my mouth and a little depression about everything around being plastic shit, including people 😢 they dont think anything, they just quote stuff they saw on their phone and try to fit into conversation. After 1,5 years of learning python and how to build stuff around LLMs - i hate ai and feel sorry for people going crazy about it
It has been generated to a brief. AI can't exist without human creative input. I think it's great!
Change toy stores “forever” is a bit rich coming from a company that’s almost entirely defunct
Lazarus is always the song for me.
I liked studio B :(
Haha, I was making the best of it, but that room was driving me crazy. If I turned the camera around, you’d see a pretty massive disaster of kids toys!
Honestly it's the sound design of the Toys R Us commercial that I dislike the most. As a first step into AI-video created commercials, not bad. I think my expectations will be higher in the future.
Oh, just to clarify, the audio I put together in Udio super quickly. I have to mute the original audio because of licensing. It’s a big no-go on UA-cam.
Trust me; their sound design is MUCH better.
Folks in Australia would say "Jiffery"
You seem very sure about what was Sora and what was assets.
I always leave room in case I’m wrong, but I did go over the video with a VFX pro pal, and he pretty much confirmed my suspicions.
There’s no way to 100% know, unless we see the actual raw output- but, y’know, both of us have done enough commercial work to spot sparkly AE particles floating around.
Which is to say, we’ve both had Disney as a client. Ha.
9:00
😆 🤣 😂
100% Sora generated or not, the final outcome is very good. No one is going to stop the video to find the discrepancies. Well done Sora ... with extra help or not.
Nah, it’s kinda creepy.
Yeah, these people are delusional. The commercial looks and sounds like a cyber nightmare.
Okay, I would never speak from Studio B again 😅
First rule of Studio A is, never talk about Studio B!!
One day, we’ll have Studio C, but that one will be planned out in advance and look super nice!
I thought Toys R us closed all of its USA stores a decade ago...
Godzilla would beg to differ 😜
I would have thought the ad was creepier than I did if I hadn't known beforehand that it was AI. But the V.O. is also a big giveaway. And "she" was definitely creepy.
OK, my own conclusion is neither Sora nor that post-production house are production ready 😂🤷♂
1000%. And I’m good with that. I think this tech can grow and evolve to be its own thing. No need to chase cinema- just let it evolve into its own art form.
…I’m barking at a tornado though.
There's a kind of paradox implicit in the commercial use of AI tools that create 'realistic' images and video in that their value as tools depends on their outputs NOT being seen as AI. For that commercial to work I need to see that kid as a real person- if I start seeing him as an AI fake person somthing goes wrong because my mental alignment with the message in the ad gets 'derailed' by my knowledge that the kid is not real, even though he looks real. Indeed it's the fact that he looks like a real person but is not a real person that makes the whole thing fail.
It may turn out that the real limitation on the use of these tools in serious production will not be technical but psychological, it simply may not be possible to achieve real connection with an audience once that audience comes to the understanding that the things they are seeing are completely fake- especially fake people, because the only reason to use people in commercials is to create empathy and identification- but if I know that those people do not in fact exist will I still feel an empathic connection- will I in any way 'indentify' with those fake people?
It seems to me that before the industry of persuasion and aspiration fully embraces fake content they might want to find out if they will be able to take their audience with them on this journey into fully fabricated worlds. There is a difference between a photograph and a psudeograph- superficially the latter may pass as the former- but the question is; can an authentic message ever be delivered by a medium that is one thing pretending to be another?
Considering stage and screen, I very much doubt that suspension-of-disbelief will prove to be a hard limit once we exit uncanny-valley.
@@cosmiclounge I don't think it's an uncanny valley problem as such- even a totally convincing synthetic human might present the same problem, perhaps in a more acute form.
The problem arises from a more subtle consequnce of fake humans, which is the ambiguity of their ontological status. Are they 'things' or are they 'people' ? On first sight this seems a stupid question I know- clearly they are things and not people- but in practice this clarity may not be so easy to maintain.
For example- most people who engage with Chat GPT will find themselves framing their requests using the kind of language they might employ with humans- so instead of an outright command to 'tell me the capital of Albania' they might say 'Please tell me the capital of Albania' and so on. There really is no need to talk to an AI as if it were a person, but the impulse to do so is surprisingly strong- and to be fair Open AI have encouraged this by having their machine talk like a person in return.
So if even a text based interface can exert a degree of moral pressure to treat it as a person and not a thing, how much more potent will this pressure become if the AI is a photorealisitc human avatar that is virtually indistinguishable from a real person? I suggest that for most people it would be very difficult to be overty rude or dismissive of a realistic fake human without feeling a bit guilty or upset, because we are so powerfully conditioned to treat humans as more than things- and anyone who does view humans as mere objects are deemed to be seriously mentally disturbed.
Why is this a problem? Well because it creates a psychological state called a 'double bind'. Faced with a highly realistic fake human, either as an on screen avatar or even eventually perhaps as a physical robot- most people will find it very difficult not to respond to these things as if they were people- especially in public settings where other humans can observe their behavior. On the other hand they are in fact things- machines.
This creates a contradiction- If I respond to the AI as if it were a real person I look foolish and perhaps a little deranged because this is an inappropriate response. But- if I treat the AI as a thing I may appear grossly ill mannered and rude because the thing in question looks so much like a person- and so this too is an inappropriate response. In short there is no way for me to respond correctly- the very presence of this entity in my world has become problematic because it's human enough to exert a subtle moral pressure on me to treat it as a person- but not human enough to actually justify being treated in this way.
All this sounds trivial but it's really not because it touches on some very profound and deeply ingrained social conditioning and as a result may stir up surprisingly powerful emotional responses. Put simply the presence of a realistic human thing in my world puts me in a no win scenario where I must appear as either foolish or rude- I must either treat it as person or as a thing- but I am not happy with either of these choices because in both cases I look bad in front of others.
Result; I become angry- I become resentful. The fact that I may not even have a clear idea as to the source of this anger makes me even more angry. All I do know is that I am not comfortable interacting with fake humans- and the more real they appear to be the more uncomfortable I feel. As a consequnce I forsee that the presence of fake humans in media may turn out to be more problematic that is currently envisaged. People may simply not tolorate their world being populated with fake humans and turn their anger upon those who create them.
Lol glad you said something about his name. It unfortunately seems like the name of a villain
👋
Heya Louis!! We're back in Studio A!!!!
marketing '') everyones talking about it
Hahha-- and we can't get away from it!
The people that green lit this need to see a psychologist.
The music in the original ad is so creepy. I found the whole ad to be somewhat disturbing.
The commercial is actually amazing quality!!! 😲 👍 👍 👍
Gotta admit, don't hate it. There are some issues, that I kinda nitpick on-- but overall, not bad.
@@TheoreticallyMedia Agreed. It's easy to nitpick, and as a perfectionist myself, I nitpick my own stuff the most.
But yes, very impressive results overall in spite of the flaws
Indeed. We also have to remember that it's for kids. They'll think it's amazing.
I think it is hard to judge, since it took them long time to do (so lots of cherry picks) and a lot of extra editing.
Nah, it’s still pretty creepy looking.
If this was made with AI, does that mean there were no actors?
Not bad 😊😊😊
please tell us the cheat cheat for kling hahah
Just imagine what we'll be able to do in a year's time!
No dude AI is taking jobs bro
I was really disappointed by the overall quality. By the time it's released we'll have a open source that almost matches it
The Toys R Us spot is fine for the intended audience.
But... Was it easier cheaper quicker than current methods?
@@raymond_luxury_yacht That is the question.
@@itsjustme373 No, that is the point!
Ai brainrot commercial
Wow - that was rubbish. These AI have such powerful abilities, it makes it seem like they typed in "bad early 90's CGI" . It also shows a stunning lack of imagination, where you can create ANYTHING you can imagine, and the person says "Ok, I want to see a kid walking into a toy store". It's like how people use ChatGTP as a search engine, they can't comprehend how to use it because they're stuck in a previous mindset that limits their outlook.
"pretty sure"? "likely"? I was hoping you actually had proof of any of your claims. :( why not give some reasons behind why you doubt various parts are not sora generated?? (EDIT: 2 of your claims seem confirmed from your source, but not the majority of what you said)
Sora is hard to judge because all we have to see is cherry picks anyway. Luma is best among those that are easily available, at least if we just talk about the underlying generator, runway is probably better overall if you consider everything even with just gen 2. Runway gen 3 or sora can maybe beat it, but we will see when or if they ever get realesed.
The fact they had to digitally enhance the output with their own content isn't a point in SORA's favor.
No, it's not just you
Man, was that commercial creepy looking. I guess it doesn’t matter how advance your AI is, it’s gonna look creepy no matter what.
For sure there’s an uncanny valley aspect here. I’m actually fully ok with that. I’d rather see AI video branch into its own art form, as opposed to chasing cinema.
2:50, well, you are wrong. The creator has clearly said over and over, it is pronounced JIF.
ua-cam.com/video/ofm4XCdZzI4/v-deo.html
Haha, oh I know, Steve Wilhite-- I was going to make a joke that he'll probably have the pronunciation on his tombstone, but then I thought to check, and sadly learned he passed away in 2022. No matter how you pronounce it, the .gif changed the internet forever. Rest in Peace, Steve!
@@TheoreticallyMedia What a contribution!! And one with controversy, what more can you ask for :)
It’s made a lot of progress since it released, but still has a ways to go until it looks natural… Honestly, I hope it never gets there
I’m actually in agreement with you there. I’ve often said on the channel that AI video can be its own thing. It doesn’t have to chase Hollywood level cinema. It can branch off and become a new form of storytelling.
…but, that’s also me yelling at a tornado.
This cant be a legit ToysRus ad, why would they use an AI voice too???
Oh, to clarify, the soundtrack was generated by me. They’ll often used licensed music, which is a no no on UA-cam. So I swapped it out.
@@TheoreticallyMedia ahh...
"Our jobs are safe"
Maximum cope. You might not like it, you might disagree with it heavily, I even believe that you'd be extremely valid in doing so, but you are deluding yourself if you think your job is safe just because the first ever AI commercial hits the uncanny valley hard. Maybe people already don't remember, but merely a year ago video generation AI could only produce 5 second clips of the most cursed acid trip like nonsense imaginable. And a year later we have AI generations that are good enough that a company feels comfortable enough making an ad out of it?
As the saying goes, this is the worst it's ever gonna be, and it's only gonna get better from here. I'm not even sure if it'll be possible to tell real video from AI generations in 5 years without an AI detection tool of your own or inside knowledge. 5 years is extremely generous too. 1-2 years is probably all the industry needs.
I'm thinking 1 to 2 years before we see generations that are at the point of being useful to full-pro productions. Like, 4 to 8k outputs that work really well for backgrounds/establishing shots etc...3 to 5 years before we're seeing full performances from virtual actors on an "AI set"
But ultimately, I think all of this is just another genre. Almost like Electronic Music begat EDM, there are still people who love their Classic Rock.
Traditional cinema will always exist, and have resurgences. It might even get better because of it. If you think of the last few years of Hollywood blockbusters-- I mean, a number of them are so bland, they practically already feel like they were created by AI.
@@TheoreticallyMedia Traditionally-made products will always have their niche but the majority of people won’t care much if the AI alternative is cheaper and more accessible imo. even if they do most companies are gonna downsize their human workforce in favor of AI. either way we’re probably going to see a major job crisis sooner than later
First AI Sora Generated Commercial: Lifelike child, i.e. Indistinguishable from real kid, no contract fees, no insurance, zero liability, no cast schedule, does exactly what is required, no daycare, zero travel expenses, no parental engagements, and patented image in perpetuity.
Sora is make believe. If we can't use it, it don't exist as far as I'm concerned. I'm sick of their teasing nonsense.
Oh, it exists. Much in the same way an IMAX camera exists. Most mortals will never use one though, due to the cost.
@TheoreticallyMedia But is it fair to put it in the same realm as all of the other text to videos generators if the vast majority of us haven't had to opportunity to test it and discover it's flaws like we have Runway and all of the other text to video generators? It's like we are giving Openai credit for tech that we've never had the ability to test out like we have everything else.
The timing and movement are extremely wooden and unnatural
the kid looks creepy . one of those dudes from polar express.
So strange to have Sora "released" for three months now, even creating an ad, but still not being released publicly. Are we seeing a glimpse of the future, where only certain select individuals and groups have access the most powerful AI tools?
No.
Sorry. Watched the whole vid, but my mind never got past the Jeffrey, JIF, GIF comment.
Massive soulless company who get their plastic toys made in poor countries now get to pay less for their marketing , yipee! Horrid ad too but I wouldn't expect much from toys r us to be fair.
They should stop with the teasing cause nobody will give a f when its released. Its not 2001