Sora has a New Trick! Plus a Dive into the Latest Features & Music Video

Today we're diving into OpenAI's latest Sora news, including an interesting new inpainting feature, although it might not quite work as we expect. Plus, I wanted to round back on some earlier announcements that I feel were kind of missed in the initial launch.
Additionally, I've got insights into a recent music video for the song "The Hardest Part" directed by Paul Trillo. This video leans heavily into Sora's surreal, dreamy aesthetic as it follows a couple from high school into adulthood using an infinite zoom technique.
I'll analyze this fantastic Sora-generated music video to highlight both the amazingness of the AI's capabilities as well as its current challenges like character consistency issues. While not perfect, it achieves some truly mind-bending shots that would be extremely difficult practically.
LINKS:
The Hardest Part: • Washed Out - The Harde...
You Might Like:
Endless Storytelling with AI NPCs & Simulated Realities: • Endless Storytelling w...
CHAPTERS
0:00 Intro
0:25 Sora's Video Inpainting Capabilities
2:11 Sora's Image-to-Video
2:56 Video Extension
3:34 Why Haven't We Seen Image to Video In Sora?
4:10 Adobe Integration
5:12 The "Hardest Part" Music Video Case Study
6:26 Analyzing the Music Video
7:17 Sora and Midjourney v4
8:35 Don't Wait For Sora

[@karlwest437]

I think it's pretty mind-blowing that Sora can generate realistic moving, walking people, anyone who's tried sculpting and animating a human character in a 3D app will tell you how immensely difficult that is

[@TheoreticallyMedia]

I think we're going to see a lot more interesting work coming up in the gen-3d world pretty soon. Tools like Moo-tion (which I'm looking at again fairly soon) are getting pretty good at that hybrid AiVideo/Gen-3D thing...Personally, I think that's where we're headed!

[@timothymcglynn1935]

Nice

[@Guest_1138]

It's not that Sora can't create great things, but it seems you can't control what it makes & you can't get anything done timely, it can't fix mistakes, it can't alter its own work. So it must be near impossible to make a cohesive film with the exact style you want, using Sora. At least a 3D app gives complete control over every aspect. So your time is not wasted. So, yes, AI seems like it will be a miracle, just not yet.

[@beastmode4617]

Hollywood is done.

[@reachthezora1912]

@@Guest_1138 I mean... this shit is not even out yet, they just announced it last year so I don't know why people expect it to be used in Hollywood like next month lmao

[@chariots8x230]

Character consistency would really take Sora to the next level as a storytelling tool.

[@TheoreticallyMedia]

if we get that, plus the Image to Video? I mean-- that makes it pretty unstoppable. Just that Shiba Inu alone-- imagine all the MJ Sci-Fi characters we've seen in Gen-2? But Sora-fied? Wild....

[@vadenallen4873]

Your content is greatly appreciated! Thanks for all of your hard work! Cheers 🍻

[@TheoreticallyMedia]

Thank you so much for that!! Really appreciate it!

[@coolkoreamag]

Absolutely awesome, thank you for sharing and explaining it so well :)

[@curiousrefuge]

Great coverage!

[@timbaker1428]

I think that for ai to get to the point where professional quality, feature length, lifelike films can be consistently made, the prompts will be less about creating a scene from scratch and more about creating detailed assets. For instance, you use the prompts to generate realistic, fully-rigged, 3D models of your actors. They become locked in as an asset that you can drag and drop into a scene, since it’s a saved asset, it should keep perfectly consistent throughout. This means you’d need to create every version of said actor, including clothing changes, just to keep them consistent. Then your sets and locations. Your props. All of these would be 3D models, generated by prompts and locked in for consistency. Once you have all of your assets generated and you’re content with them, that’s your media library…you can start prompting your scenes and shots, calling upon the assets from your library. Would this require an incredible amount of servers and processors to generate? Sure, but in think ai companies know this, so this level of video production may not be readily available for the enthusiasts, it’ll be reserved for studios and those who can afford to use the technology at this level. Studios would still save a fortune building content this way. As infrastructure improves, technology costs will come down and the indie filmmaker and hobbyist could join in on the fun.

[@ixiTimmyixi]

I enjoy your Runway version of the Infinite zoom!

[@TheoreticallyMedia]

oh, credit where it is due: That was Blaine Brown!! His watermark is in the video!

[@HistoryIsAbsurd]

Always with that top notch unbiased information! Thanks as good sir

[@TheoreticallyMedia]

Super appreciated!! I always try to avoid hyping Sora too hard-- I mean, the obvious saltiness that it's locked up-- but, I still maintain it isn't going to be that great when we finally DO get it. As Paul said: 700 shots and he used 56? I mean...yeah.

[@HistoryIsAbsurd]

@@TheoreticallyMedia Yeah like this information in this video was exactly what people were worried about . Although that video was bad ass as hell. Going from that night scene to the street through the car was just beautiful!

[@robertdouble559]

576p is Standard Def PAL resoultion. The bane of my existence for many years!

[@TheoreticallyMedia]

Is that what it is? I didn't have to deal much w/ PAL back in the day-- but that makes sense! Stupid PAL and their weird frame rates as well! (haha, says NTSC...) Thanks though, I'll file that nugget away for the next SMPTE trivia night! HA!

[@PerFTNilsson]

@@TheoreticallyMedia Jeez ... "NTSC? Never Twice the Same Colour" *old European saying* 😄

[@uttula]

My mind immediately went to ”that’s just the standard oldskool DVD resolution” … but yeah, since I live in a country we used to broadcast in PAL, of course that would be ”standard” to me :-p

[@Thebrainymonkey]

I was going to say the same thing. 1024x576 16:9 square pixels, 768x576 4:3 square pixels, 720x576 4:3 D1 pixels. It's all burnt into my brain just like people's defunct phone numbers from 40 years ago.

[@robertdouble559]

@@Thebrainymonkey right there with ya!

[@histofactos]

What is that background. Grandpa reading room ?? 😂 Love your videos 🤩

[@TheoreticallyMedia]

Haha. Funny enough, the exact opposite! I took over the kids “play” room while Studio A is getting revamped. Technically, it was supposed to be their study room…but, kids. That said, given the wall, I think in a previous life, that room was 100% a grandpa reading room!

[@histofactos]

@@TheoreticallyMedia 😁

[@dylan_curious]

Wow, SORA In-painting is going to be next level!

[@TheoreticallyMedia]

It'll be huge (when) we finally get it! I'm doing my best here to manage expectations on Sora, since I don't think it's going to be the AI Video to Rule Them All, as it is being touted. But, it IS going to be pretty amazing...

[@ChrisCapel]

Just curious, what kinds of films/media did you make before all this AI stuff came along? Is there anywhere we can look at your non-AI-related work?

[@TheoreticallyMedia]

My main bread and butter was making "making of" features on DVDs and Blu-Rays. Honestly, I did so many of them that there's a solid chance (if you owned a moderate collection) you have something I did. Some of the heavy hitters: I did some work on the Aliens Boxed sets (not the main filmmaker there though), Star Wars OG Trilogy, and The Clone Wars. But, there was also stuff like Juno (if you ever saw that), The Biggie Smalls film, My Name is Earl (legit, that was a fun show!)-- god, I can't even remember anymore. Just tons of projects. Had a LOT of fun, but obviously this was all corporate factory work at the end of the day. Some level of creative influence, but ultimately it was client work. Aside from that, there are some short films floating around, a bunch of spec scripts that went nowhere, a comic that actually got optioned (but not picked up)-- So, yeah-- haha, I've kicked around for a bit!

[@aisamsonreal]

Thanks for the update! Is the first 2-sec a voice clone?

[@TheoreticallyMedia]

Hey man! No, that was just bad ADR! haha-- haven't quite found the sweet spot on this mic! Funny story: We've been on the same couch (which sounds kind of weird...)-- I did a podcast w/ Andrei Jikh a few weeks ago. I was laughing thinking: Samson sat in this very spot!

[@aisamsonreal]

​@@TheoreticallyMedia Ha! That's awesome. Would love to catch up in person sometime. Keep rocking the amazing content!

[@johndev2212]

I'm loving the homey vibe of Studio B, only thing missing is a big mug of coffee. 🙂

[@TheoreticallyMedia]

You can see one just under the screen on the right! haha- I'll bring it out full and center next time around!

[@TheBlueRage]

Can't wait to purchase this. It would be interesting to see what Maya, Blender, Unreal, and Unity do with this technology.

[@ZeroDebtInvestments]

What's the AI video generator called at the end of the video? Blade Round?

[@TheoreticallyMedia]

oh, sorry-- Runway Gen-2. But Blade Round sounds super cool and now I want to make that a company name!

[@Blizaine]

Thanks for the shoutout at the end! 🙏😊

[@TheoreticallyMedia]

Oh, 2000%! Great job with that video!

[@PawFromTheBroons]

Now that monitor was a serious throwback! ❤️

[@TheoreticallyMedia]

Haha, still rockin' it! It weighs a ton, but it's still kickin'! I'll probably upgrade this summer, but I'm gonna miss the beast!

[@PawFromTheBroons]

@@TheoreticallyMedia Compared to a Studio Display, it is indeed a behemoth.

[@AllanGildea]

Holy smokes, Tim. The walking animations!

[@TheoreticallyMedia]

I mean, it's crazy. I think it does best with the waist up shots, but some of those full body walk cycles look pretty darn good. A few are wonky, but-- for the most part, the best we've seen yet!

[@ECHSBACHS]

The Infinite Zoom Videos are Amazing.

[@TheoreticallyMedia]

I kind of love them too!

[@thinkaskandlearn9685]

Would you recommend any online film making course that is worth it?

[@TheoreticallyMedia]

Drop me an email at [e-mail]

[@sleeplesstortoise]

i already mastered character consistency in stable diffusion XL (only for 2D image though), now all i need for sora is to have good img to video generation

[@TheoreticallyMedia]

I think you’ll see some stuff coming from the 3d world that might better suit you for movement before we get our hands on Sora. Like I said toward the end of the video: lots of tools right now, go explore them! Don’t wait!

[@sleeplesstortoise]

@@TheoreticallyMedia the thing is i only need for 2D animation, like anime. i already made comic video using stable diffusion in my second channel. would be cool if i can make the characters move. so far img to video i've found can only generate only up to few seconds and they're not good enough. i'd check your other videos to see what other tools are there. Great channel btw

[@sealfsalve]

We're such a fan of your reporting and insights! Can't wait to experiment with Sora when it's publicly available. If/when you've got a moment, we'd love it if you could check out our latest music video (Gilded Cage) and let us know what you think - our current pipeline is pretty much all AI, save editing in Premiere.

[@sealfsalve]

kzbin.info/www/bejne/iaK0o5KdiriDia8

[@TheoreticallyMedia]

I’ll check it out!

[@ebonypascal1]

Thanks for the update on Sora! Promising. I'm wondering if this will become a worldwide used tool, or if people's interest will drop because of the "hallucination" aspect (like the weird cars, the guy running inside the mirror, etc.). Like, will this primarily be a great addition to music videos creation (commonly validating new/unleashed visual styles) or will this pass the test of real storytelling, involving details like precise framing, ultimate use of light and shadow, camera move, detailed scene composition, etc. My fear is that we will never see the in between stage and go from where we're at now to a movie fully generated by AI with very little human control. When my personal dream is a fully controllable tool to free movie creation for all. Kind of what happened with samplers in the music industry in the 80s. Maybe these AI video generators shouldn't be that smart after all, but more like perfect at executing human's vision. We shall see

[@TheoreticallyMedia]

Indeed! I think a lot of the warp-y stuff will eventually fade away, much like they have with the image generators. But I'll also say, I'm going to kind of miss it when it is gone. I love the weirdness! One day, I really want to do a film-- maybe in 2050, where we have a flashback scene to 2023, and the entire scene is done in this style. I mean...it's a pretty "if you know, you know" gag-- but I think it'd be a pretty good one!

[@grerovambrozoyuz9426]

Soras video output looks like in my dream. People in the bus while walking in the bus that never ends.

[@tomannejules]

Hello. Have you tried Krea yet? Pretty cool. I'm an animator and Krea allows key framing.

[@noobicorn_gamer]

Waiting on your ElevenLab's latest music feature teaser review now lol

[@wiicrisis]

Probably watched 50 of your really great videos. Way about tim(e) [*pun intended] to hit that subscribe button 😳

[@TheoreticallyMedia]

Oh excellent to hear! More the better as we make the march to 100k!! So close!!

[@astralislux305]

Nice painting

[@TheoreticallyMedia]

Thanks! I was getting yelled at a lot in the comments for the bare 70s faux wood wall. Haha, needed to hang something up there!

[@raaghavgr1990]

Does anyone have any idea about when we can expect Sora to be released and what will be the length of the videos that we can create once Sora releases?

[@Simplicity4711]

They don't plan to release it, since it can easily be misused in a dangerous way as I understand.

[@circelink]

Keep AI weird! ;) Great stuff Tim, as always.

[@TheoreticallyMedia]

We really need that on bumper stickers!! Thanks so much!!

[@thinkaskandlearn9685]

The real problem is "how SORA will retain characters identity in a film"

[@TheoreticallyMedia]

Well, if the Image to Video aspect is released, that'll be the solve. Unless, OAI has a Consistent Character feature for Sora v2, whenever that happens!

[@thinkaskandlearn9685]

@@TheoreticallyMedia very good point. Thanks. Do you recommend any online film making course?

[@gabrielsandstedt]

just generate the same character in text to image tools and then use sora for image to video - solved

[@thinkaskandlearn9685]

@@gabrielsandstedt are you sure sora will produce the same character image in every scene?

[@urielcalderon9639]

talk to AI about how to use legalise terms to disclaim a loophole around that

[@ohwhatworld5851]

Is that the Apple XDR display? It looks so chunky.

[@TheoreticallyMedia]

old school cinema display! Still kicking and working like a charm! It is not only chunky, but SUPER heavy!

[@lionelson7098]

Some time ago, I've been thinking that the work/products of companies like Viggle and Mootion are reasonably more important than those of Sora

[@TheoreticallyMedia]

Same. I did that whole video on Viggle/Kaiber last month. And I'm going to be looking at Mootion again pretty soon-- BUT, I think all of this ultimately lands when we see the hybridization of Style Transfers, V2V and a generator like Sora. That's going to be the ultimate toolkit-- and probably not too far away?

[@JayJay3D]

OK, I love what I have seen of Sora.....but now the dust has settled (and I'm definitely NOT hating here) I'm not blown away by it. I have a feeling part of the reason its not released yet is its not ready and what we have seen may be cherry picked results after numerus video prompt generations. I personally feel making a semi quality pro short film using 'just' a tool like SORA is still someway off (18 months or so). It may well be a hardware issue (NVIDIA may well be the answer here). But I do like SORA and other video generators, and still very early days - exciting times :D

[@joelleyendecker1536]

Reminds me of « Je danse le Mia », one of Michel Gondry’s early music videos.

[@TheoreticallyMedia]

I can 1000% see that. Actually, a lot of Sora stuff kind of has that Gondry vibe come to think of it. There's a good amount of Eternal Sunshine I see in the overall look here as well. I mean, Gondry is one filmmaker I'd love to see play with Sora. I'd imagine he'd get pretty insane with it!

[@tantarantaran]

Exactly, the whole video is so Gondry. A lot of what’s coming out of Sora is very reminiscent of the more edgy early 2000s music videos/ads from Jonze, Cunningham, Fincher and Gondry. Cool way to utilise all the AI weirdness and glitching. Still skeptical if this is ever going to be economical when you pay the real price in production

[@tantarantaran]

@@TheoreticallyMediaGondry would just probably get pissed and quit after an hour or two haha. Slot machine filmmaking is not for the perfectionist

[@tantarantaran]

Fincher’s ‘constant change’ for HP or Spike Jonze’s Apple Homepod would be spots that they would try to make with heavy generative ML l, if they would do them today. That style was in for a few years. Perhaps it comes back if these tools become available, then people get bored with it. ML will pass the test if/when it’s invisible and seamless - I can’t see how filmmakers would ever achieve that without advanced knowledge in traditional filmmaking, editing and compositing. 99% of generated art, you can see straight away that whoever made it had no idea of basic photography, filmmaking and editing, storytelling.

[@vainezaiven6677]

The 10% angle isn't a big deal to me. As a writer, I've always followed what I call the "iceberg formula," in that 90% or more of the work I do is never seen by the audience, similar to how most of an iceberg is below water and can't be seen by those above water. I consider that to be pretty typical.

[@TheoreticallyMedia]

Couldn't agree more. Even on these videos: I think probably would be shocked to find how much editing went into even just this video. Most people figure: 9 minute video, probably took him 9 minutes...NOPE! Iceberg formula is a great way of putting it!

[@noobicorn_gamer]

KZbin really won't be the same when sora does ever gets released to the public lol

[@TheoreticallyMedia]

haha, it'll be a mess for a bit for sure-- but I think the general audience will be able to pick up on Sora Generations fairly quickly. From there, you'll have to have really stellar editing and a ripping story-- otherwise the YT algo will just bury you. ...trust me, the Algo has buried me plenty of times...and that's for videos I've actually SHOT! haha

[@noobicorn_gamer]

@@TheoreticallyMedia Let's think of it this way; we subscribers are lucky to have found you so we feel special like we unearthed an unexpected treasure that both informs us with trust worthy info and entertainment :) Thanks always for your uploads.

[@johntnguyen1976]

Gentle reminder...green fill light is fine...if you're Gore Verbinski! LOL. Jk jk. As always...thanks for keeping the community up to date, my good sir!

[@TheoreticallyMedia]

Haha, I’ll have to keep that green light saved as a preset for the next time I do some J-Horror examples!

[@spinninglink]

No i can't all be done with what we have now lol. ugh can't wait to be able to use sora

[@rw9207]

This is the guy that would stand at the side of the field, watching the Wright brothers and be like, " That's never going to get me to Europe for my Vaca!"

[@TheoreticallyMedia]

Also, complain that the flight has been delayed, luggage was lost and the WIfi on the flight isn’t working!

[@rw9207]

@@TheoreticallyMedia HAA.. yus! Well, it was only 66 years between the wright bothers and landing on the moon... there is hope yet.

[@kon_radar]

480p is weirder than 576p 576p is when you divide 720p by 1.25 576p is still a real 16:9 format. 480p is not. Because a real 16:9 480p would be 853.33333 x 480, and that is not possible.

[@TheoreticallyMedia]

Someone mentioned that 576p was a PAL format? Either way, your math checks out! Or is it Maths in the UK? Blimey!

[@tantarantaran]

720x1.5=1080, 1080x2=2160. You stretch 4:3 720x576 to 1024x576 16:9, with a 1.42 widescreen pixel ratio. 720x576 (SD) - 1920x1080 (HD) 3840x2160 (4k). 576p is the perfect resolution to render now that you have advanced upscale

[@cuya1985]

very big improvement. actors can be chosen by us:) or old movies can be changed to modern times:) or for exampe japenese movie can be changed as a west movie:) especialy the old ones:)

[@claudioagmfilho]

🇧🇷🇧🇷🇧🇷🇧🇷👏🏻, Sora is just so awsome, i am always inpressed with it!

[@EllenVaman]

👏🏻😁

[@TheoreticallyMedia]

Appreciate you!

[@CineSolutions]

It wears thin to hear updates on something that isn't even out yet.

[@TheoreticallyMedia]

I know-- That's why I've got that bit toward the end...Everytime I do a video on Sora, there's a lot of salt. Trust me, I get it... I'm starting to turn the corner a bit-- maybe as we get closer to Sora (?) where, as we see more of the limitations and frustrations-- I'm actually getting more excited about it. I know I'm in the minority there, though!

[@lenderzconstable]

New trick? I thought it wasn’t even out yet.

[@mehface]

That music video gives me the creeps xD After looking at AI stuff too long i have a bit of a crisis and lose my sense of reality, especially when I look at real people again.

[@TheoreticallyMedia]

Yeah, I'm not going to say that wasn't an uncommon statement around the era of MJv4. I was doing tutorials on that during that era (ha, era...meaning a year ago) and your comment was pretty common. Something in that wax-character/uncanny valley thing. I totally get it. But, like MJv4 (which again, for some reason, I actually LIKE that look)-- it'll get better as time goes on.

[@erwinmeisel2545]

San Fancisco was on purpose I guess

[@TheoreticallyMedia]

SIM Francisco!

[@ishimurabeats6108]

This rly should never go into public hands

[@zenmasterjay1]

Dont pay attention to sora hype until it is accessible

[@fiebelk]

What happened to your audio recording lately? Since your new location the sound quality is indeed terrible! Keep the good work anyway!!

[@TheoreticallyMedia]

oh, god-- I can't wait to get back to studio A. Mostly, it's like starting all over again and I'm back on the learning curve. I was JUST talking about this: This room kind of sucks compared to the wood walls in Studio A. I haven't quite found the sweet spot on the mic-- though, I think I figured it out after this video. Oddly, I'm sitting-- which I think actually affects the recording. I mean, ever so slightly. Mostly, I think it's the mic. I need to angle it a little better. I should have it dialed in next video. ....but yeah-- I can't wait to get out of this room!

[@twilightfilms9436]

The ratio was 300 to 1…..and in other news, Apple mentioned for the first time the word AI in a launch presentation……

[@TheoreticallyMedia]

300 to 1 was the ShyKids video, I believe. And HA! Apple has finally given in! I know they were going for the whole "Think Different" in trying to brand Machine Learning...and I get it, but-- man, no one was drinking that Kool-Aid. Sorry, Tim.

[@armondtanz]

How the heck are adobe gonna cover the $500 for 30 seconds in a $20 a month subscription 😂😂😂

[@TheoreticallyMedia]

Haha, creative bookkeeping? Honestly, I have no idea how all of this is going to shake out...

[@LouisGedo]

👋

[@TheoreticallyMedia]

Heya Louis!!

[@sveinnrafnsson851]

Need 10 clips to get something useful. 1 hour generation per clip. Need 50 clips for a short video. OpenAI: "That would be 500$"

[@wally3436]

Damn you AI overlords! I'm truly getting fatiqued by the constant teasing of video AI in it's different forms. It's not only SORA, but almost almost all of the truly cool features like Animate Anything, Microsofts VASA-1 ect. I mean, Pika and RunawayML are cool tools and all, but if ya wanna make something coherent, it takes alot of trial and error and compromise and frankly that stuff just isn't that interesting anymore. One option would be to blast 5k on a computer and truly start learning Automatic 1111....

[@wally3436]

Oh, but cool video tho and I appriciate your content! Keep up the good work! :D sry for the rant

[@TheoreticallyMedia]

Haha, I know, I know. We're in a weird "growing pains" lately-- you can see that it's right around the corner, but not quite there yet! I think by this summer we'll see a lot of improvement on the AI Video front. We've already seen Vidu, and I have the sneaking suspicion that Runway is going to drop something big soon.

[@TheoreticallyMedia]

Oh, no worries-- I feel it too. I'm right with you!

[@LouisGedo]

Sora is the benchmark tool for text to video AI......all the others are far behind IMO

[@TheoreticallyMedia]

It's frustratingly true-- although others are catching up! I think by this summer we'll see a big upswing. Fingers crossed!

[@nicheva417]

I'm keen but have no audio skills or experience and I absolutely ha te adobe as a company.

[@High-Tech-Geek]

I can't stand the nauseating infinite zooms. I want to know why this is so popular. Let's try doing something unique guys.

[@TheoreticallyMedia]

Yeah, that's a pretty established look at this point. To Paul's credit, he's apparently wanted to do this video for 10 years (pre-AI), and it narratively fits the idea-- but, I do hear you. Let's maybe give this one a pass, but put the infinite zoom thing to rest. Next up, I'm thinking Bullet Time!! haha

[@johnpapadopoulos5687]

I think at this point, everyone is asking in their head the same question. How much is all this going to cost? But even more so... who is this for? It is an easy guess that the vast majority of AI image generators, are kids who play around, mainly creating images for social media etc. And a very helpful factor in this, is the generation time, which is now less than a minute. Kids today can wait for a minute, but an hour for video..? And do what with it? I am going to take a wild guess that this is going to be expensive and addressed to serious users, who like in Midjourney are a minority. Even a mucisian who would like to make music videos for his songs.... how many videos will you make? Does it worth a monthly subscription? Personally I would love to make two or three videos for my music, but do I want to be part of a monthly paying model for that? Probably not. So do we have any news, rumors, or suggestions about the pricing plans of Sora yet? Also.... you are killing me man... it might be a good idea to leave your grandmother's favorite frame out of KZbin... still looks better than nothing, but... com' on!

[@TheoreticallyMedia]

My guess is (and this is pure speculation) but that's what the Adobe partnership is all about. There will be a free (fun) version of Sora available. That will be short generations, maybe 6 seconds at 720p. Video will be emailed to you when it's done. But, there will also be the "pro" version: Which will be...like, $40/mo to OpenAI, but only available to you with Adobe Premiere-- so you've got that Creative Cloud sub to contend with. It's really the only way I can see the cost being offset. Naturally, the CC users will have priority generation and hopefully it'll take less time. Again, total speculation.

[@johnpapadopoulos5687]

@@TheoreticallyMedia It sounds about right... we'll see. Thank you for taking the time to reply to your comments.

[@paiaam]

Why do you keep dangling Sora in our faces, while we all know that we won't have it?! :(

[@TheoreticallyMedia]

I still think we will eventually get it. It'll just be a fairly nerf'd version of it. Maybe 8 sec generations, and possibly at 720p. That said, since they're partnering with Adobe, we might see a "Pro" version integrated there. I can't see how you couldn't if you've got professional editors working on it.

[@paiaam]

@@TheoreticallyMedia Thank you for the great work.

[@Guest_1138]

Critical of Sora?? Sora what? We can’t even use it. It doesn’t exist until it comes out. And how does Sora allow you to modify and edit what it creates? Isn’t it going to be the same as every other AI app that does great magic 8 Ball videos, but no way to change what you have and keep the rest of the image. Even your video example changes the entire image and camera move on every variation. And 15 min to 1 hour to generate a video, but at a 300:1 ratio that a LOT of wasted time. None of these apps will be able to be used for industry work.. so at the moment it’s just for fun. Like an AI fidget spinner.

[@TheoreticallyMedia]

So, literally everything you said is what I'm critical about Sora on. Seriously, dig through my previous videos on it. A lot of which was a reaction to that initial wave of wonder (which was justifiable) but also overhyped by a lot of other channels. From an audience perspective, there was a lot of "I'm not going to work on my project, I'm just going to wait for Sora" and I really wanted to push against that. The Industry stuff will eventually have super niche tools (I've seen a few Film Restoration AI generators, SUPER impressive-- but we're talking frames that it generates, not even seconds)-- but to be honest, I'm not particularly focused on that area. Namely, because they've got much bigger problems than Sora. (I say that with love, as I still have many friends in the industry...but c'mon, The Fall Guy bombs and you're scratching your head wondering why?)

[@LynneHand]

AI fidget spinner. Stealing that one. 😂

[@Guest_1138]

@@LynneHand hehe.. it's true though.... re-roll over and over. Trying to get what you want... Never get what you want ... Re-prompt over and over... nothing accomplished, End of day.. but sort of fun doing it somehow.

[@QuantumEX]

All facts! Show us just one clip that doesn't hurt our eyes. The only clip that comes close is the woman walking in Tokyo; but even that is not that great. If Sora is so powerful, why not create a one simple coherent clip. For example, Will Smith eating spaghetti. So far, all released clips make me feel like I am high; like Katt Williams high high.

[@Nemsis19]

be patient and trust the process compare that to windows movie maker on windows xp

[@Crazy_Truth]

Are u millionaire!? 👀

[@TheoreticallyMedia]

Ha. Hardly. You'll know I'm doing well when I introduce a line of energy drinks.

[@Crazy_Truth]

@@TheoreticallyMedia coz I wanted PS5 😭

[@unitednorthpole]

At this point... according the social pressure between men and woman.. we soon have only virtual friends...Women are too much selective.. virtual women always are free to talk and they will be perfect..

[@robertruffo2134]

It still has that nauseating AI look. It's just offputting. And as a note: yeah, it's fine for a certain kind of slurpy surrealism, but most films are not that, and empty slurpy surrealism is only interesting in small doses. As a final note: this video has no copyright. So if you want to use the images in your own music video, orin a hard core porn even, anything, go ahead, because you can't copyright "AI" output..

[@gr33rtheartist]

Sora has not leveled up... Sora has done nothing other than be a rumour for some time now. Sora is doing nothing much at all.

[@TheoreticallyMedia]

well, it's more than a rumor-- I mean, it exists. But, I hear you. Like I said in the video, anytime I mention Sora, I fully expect some extra salt. I get it.

[@gr33rtheartist]

@@TheoreticallyMedia No salt from me... I love your work. But everything we see from Sora and other apps with similar claims of greatness hold no value until real people can try them out and see how they actually work. All we have seen from Sora are curated highlight reels. And we all know from experience the expectation vs reality when companies, even very well-established ones, present us with Highlight Reels rather than real tangible hands-on products. I guess I am guilty of being somewhat too dismissive at times of these claims. But all I want to hear as far as news about Sora is when I can lay my hot little hands on it and give it a go. Then I will know what it can and can't do.

[@ashlayriot4135]

Text to video is good, video to video is not good . 비디오 투 비디오가 제대로 되려면 텍스트 투 비디오가 소라까지 오는데 걸린 시간에 비해 꽤 많은 시간이 필요할듯.

[@QuantumEX]

I didn't realize that incoherent crap is considered a style. Do you understand what it means to have a style!? Randomly generating incoherent images is not a style.

[@johngordon4118]

Stop showing the capabilities, just bring it out .... I BET THE CHINESE ARE WORKING ON SOMETHING MORE ADVANCED

[@TheoreticallyMedia]

Haha. Already on it, did you see this yet (ok, maybe not more advanced…but still): New AI Video Goes Hard At Open AI! kzbin.info/www/bejne/qGPHd6SEh9Wbe7c

[@a.akacic]

cringe

[@TheoreticallyMedia]

Eh. +2?

[@1980daxian]

STOP REPORTING ON SORA!!! ITS NOT OUT..WHO THE FUCK CARES

[@TheoreticallyMedia]

Haha, I hear ya, all caps and all. I don’t know, it’s still interesting. I mostly like to showcase that it’s not going to be some magical one-prompt movie maker. I think a lot of unreasonable expectations have been set up, and I just like to show what it really looks like. Because it is amazing. And a mess. At the same time.

[@gracegoce5295]

you talk about something you dont know. You0re just talking about the videos sora's presenting.

[@a.akacic]

cringe

[@TheoreticallyMedia]

Eh.

[@TheoreticallyMedia]

Eh, +1