Exactly How "Gamechanging" Was AI Video In 2024?

[@rosscads]

What a year! And don’t forget Kling 1.6 - the last and best publicly available video generation model of 2024 🎉

[@krisDag17]

Kling 1.6, Prompt coherence is significantly improved from 1.5. Lip Sync actually works better than Runways lip Sync. Kling output quality was already better than Runway's so, yeah, it is truly great. The best I am seeing at the moment.

[@DJTechnosapien]

Udio and Runway have changed my life! It’s so much fun to create on the fly, and got me back into physical art as well! 2024 is a year to remember, 2025 is the year of forever

[@torrancejones7294]

Thanks!

[@ArtificiallyCreative-AI]

Nice overview, thanks for the "year in review" - and 👍 for the RUSH Permanent Waves album on your mantle this time. Excellent taste in music!

[@TheoreticallyMedia]

Haha- that was the morning listen! Although not my favorite Rush album, still a killer listen! (I mean, they all kind of are!)

[@LouisGedo]

Instead of SORA, we got SORTA 🤔 😉

[@GabrieloTomorrow]

That's actually very clever! 😂

[@LouisGedo]

@GabrieloTomorrow I sole that from someone else........ because it's clever! 👍

[@GabrieloTomorrow]

​@@LouisGedo _"sole"?_ Bro, you're so good at stealing that you're even stealing letters from the alphabet! 😂👍

[@LouisGedo]

@@GabrieloTomorrow 😆 🤣 😂 Keyboard issues!

[@Edit9-dj9vc]

Thanks for a great year mate! Great content as always, have a great Christmas

[@afrofutcha]

This is crazy, the amount of things that came out this year. my favourites have been Hailuo, Udio, Kling by far.

[@TheoreticallyMedia]

2024 has for sure been good to us! Can’t wait to see what surprises 2025 has in store! To think at the start of 24, no one knew those names!

[@narottamzakheim5051]

Great video Tim... as always. Amazing how far we have come in one year. The cherry on top of course being Kling 1.6 which i huess was too recent to make the cut... but its showing pretty impressive results. Look forward to 2025,,, im sure this will all just expand exponentially... as per usual lol. Have a great holidays mate... see you next year!!!

[@gumvue.studio]

thanks for another great video .. Merry Christmas and Happy New Year. lets see we can get access to VEO 2

[@Sasha_sementsev]

Thanks for your videos! Happy holidays 🖖

[@TheoreticallyMedia]

Happy Holidays and...well, via that emoji: Live Long and Prosper as well!!

[@Escelce]

Thanks for the great videos of 2024. Looking forward to next year's. All the best, Tim.

[@TheGalacticIndian]

Massive kudos to the creator for building such an amazing AI channel and being an incredible AI influencer! His name is TIM🎖 Tim, your work truly stands out as one of the best out there. Keep inspiring us in 2025 and beyond please!

[@biancablackwhite4723]

Thank you! Merry Christmas and a happy new year, kind regards from good old germany 😊

[@PeterStrmberg007]

A great premiss for a video Tim. Would have loved to see how the man in a blue suit progressed through the year. Lots of omissions. I'm trying to say this should have been longer! Amazing to think Flux was only 5 months ago. It really did raise the bar. Vidu also came back strong, and LTX made local video generation practical. Honestly, if I could I would step back from video generation for a while, knowing that before long it will be much, much better and easier. VEO 2 has set a new level of realism due to it's inbuilt adherence to physics, but still lacks some of the features we know are possible from other platforms. Hopefully Flux video will be just as transformative as Flux image generation was.

[@morpheus2573]

An overwhelming number of AI contenders & competing products. I never thought I'd wish for... "One ring to... bind them."

[@TheoreticallyMedia]

It’ll for sure happen. There are a number of platforms already starting via APIs, but I think that cost will ultimately weigh them down. But I do think in 2025 we’ll see a major swing from someone, maybe Meta? With a “one ring” platform. It’ll be the hotness for a bit, but ultimately fizzle once everyone’s outputs look “same-y” In the end, I think we’ll see a few “do it all” spots, each serving a different audience skill level. The real powerhouse players will be the creators who bash all of them together to create something super unique!

[@morpheus2573]

@ Sage reasoning, friend. And when someone says... "Look to my coming at first light on the 5th day. At dawn, look to the East." ...believe them.

[@MentalParadox]

@@TheoreticallyMedia "... ultimately fizzle once everyone’s outputs look “same-y” " Or people get sick of being completely restricted in terms of allowed prompts. These corporate generators are insanely censorious

[@High-Tech-Geek]

I can't believe this all just happened. Seems like it's been a few years. Great recap.

[@TheoreticallyMedia]

I know, right?! When you kind of lay it all out, it's pretty mindboggling at how quick everything has happened. 2025 is going to be insane!

[@Bittermandl]

The speed of new developements in AI is overwhelming and frightening!

[@asociacioncolibri]

Happy Holidays and regards from Mexico City. ✌️

[@TheoreticallyMedia]

Happy Holidays to you as well!! Please eat a few extra Tamales for me!!

[@videosfeoscomotucara9038]

Gold, very informative

[@TheoreticallyMedia]

Appreciate it!! Have a great New Years!!

[@thorstenkiecker5593]

Your passion for staying ahead in the fast-paced AI world is truly commendable. The time and effort you invest in testing platforms offer your followers invaluable insights. That said, many platforms and reports about the AI world often lack transparency regarding data handling and licensing agreements. While new features and tools grab attention, details on how user data and creations are managed or what rights users retain are frequently unclear. I’d encourage you to highlight these aspects more in your updates. By shedding light on licensing and data usage, you would help your followers make informed decisions while navigating this complex landscape.

[@apoemaday]

Am I mistaking or you left LTX Studios out? I think it really is up there and deserves attention! Thanks for the vids though. I always like your stuff and point if views. Happy New Year 2025 to you!

[@harrygoldhagen2732]

Thank you Tim. Happy Holidays!

[@jaysonp9426]

I loved this, great to really think about how far we've come. Here's to 2025 🎉

[@TheoreticallyMedia]

I seriously can’t wait to make this video again in a years time! I’ll have to come back and watch this like it’s a time capsule!

[@solidreactor]

Thanks for 2024, looking forward to your content throughout 2025!

[@JonathanStidwill]

Thanks for that great summary of the year. What an incredible year for AI image and video generation. I can’t imagine where we’ll be this time next year. I just hope that some of these sites continue to offer (limited) daily credits to allow people to try things out and experiment before subscribing (rather than one-off credits when you sign up)

[@roricam]

Merry Christmas and I hope you have a great 2025. +1 on voting Kling 1.6 as the best right now.

[@TheBadRandolph]

Thank you as well! All the best to you, your loved ones, and A.I. Tim; no matter where he is, he's probably having a blast right now ;).

[@GrizzlyBossTV]

Happy 2025 🤞 🎉. Where is Seaweed ai? I thought it was gonna release before sora.

[@TheoreticallyMedia]

Same! I’ll check in with Mr.X, my guy at ByteDance and see what I can pull out of him!

[@marcs7847]

Three minutes into video and I laughed so loud my wife came in and made me explain the "Talking Heads" joke. (She has no idea what EMO is lol) But that was classic... Awesome idea for a video BTY... Because as Maya Angelou said: “You can't really know where you're going until you know where you've been”. I can't say why... but I have a feeling 2025 is going to be NUTS... for some reason I smell AI Remakes hiding in the shadows that could actually have family watching TV together again....Just kidding lol ... not even AI could that... But it's a nice dream. 😃 Thanks for the great content and the laugh!

[@ssotkow]

Suno V4 model is already online and available for free testing. Imho, it's better for vocal songs but less so with instrumental only after several days of testing. Suno is definitely the top AI music maker on the market right now.

[@ssotkow]

Thank you Tim for a great year of contents from your channel. May you stay well as 2025 hits us with more fantastic AI products/services. Cheers Tim!

[@TheoreticallyMedia]

Yeah, I called out the release of v3, but v4 did drop in November! Still waiting on a big Udio update, but I'm sure it's coming!

[@TheoreticallyMedia]

SUPER excited for 2025! Can't wait to share all the insanity that is sure to come!

[@LouisGedo]

10:50 I will never get tired of watching this generated video! 👀 👍

[@MartinZanichelli]

That should be the thumbnail of every video Tim makes :)

[@onlineispections]

Hi. How do you create images of two people talking, shots, reverse shots, close-ups of the same room? What tools to use? Thanks

Great reacap, Bro :)

[@TheoreticallyMedia]

Appreciate that! Might be a bit early, but have a great New Year!!

[@dotdotmusic-xt6gw]

Merry Christmas / Holidays! crazy to think so much has happened so fast 😵‍💫 chatgpt only released like 5 years ago. Going from the wright brothers to F15s took 60 something, so it makes you wonder where’ll we be in ten…

[@adamholter1884]

Great synopsis

[@MA_SK_ID]

“Taking a knee before sprinting” is the best turn of phrase

[@TheoreticallyMedia]

gotta catch your breath before that big burst of speed! (Actually, I think the olympic runners would disagree...but, just another reason I would never be in the olympics. Wait, is there an AI Olympics? I might be up for that!)

[@MA_SK_ID]

@theoreticallymedia obviously it is not my wording but I interpreted it slightly different. I do like both of our interpretations. I saw it like bending the knee is the final preparations nobody pays attention to, the runner down and setting self up, ready to leap to action. Less about rest but more the work that leads to success. The sprint is the public forward piece which looks tons of speed. However,in reality it is just the public announcements that come fast and furious - the work has been going on for a while. WOW I really over thought that one- tell me you’re autistic without telling me you’re autistic 😂

[@agnesslovehealz]

Happy healthy safe holidays everyone ❤ 2025 is gonna be lit

[@TheoreticallyMedia]

Same to you!! Can NOT wait for the ball to drop and to get cranking on '25!!

[@CyberwizardProductions]

really fast year, but wait till next year - what happens next year is going to make this year seem like it was crawling.

[@Kenb3d1]

Thanks for all the great content Tim. Out of interest, what would be your 2025 AI image/video wish list?

[@MentalParadox]

"a certain Chinese flair" *showing Winnie the Poo* hey man, I still want to access Hailuo and Kling at some point 😅

[@brakasoft]

Have a nice Christmas. And Miss TM too. Actually ... that´s all for the comment this time 🙂

[@nadernaderi1017]

Same as it ever was! 😊

[@TheoreticallyMedia]

Here comes the twister!

[@GabrieloTomorrow]

I predict (and hope) that 2025 will be the year of _Text-to-3D games._

[@СлаваЛакки]

I guess it is yet too early.. we need another 2-3 years.. but maybe first neural nets for 2d games. I mean not diffusion simulations like google showed, but real games that are possible to play and replay.

[@GabrieloTomorrow]

@@СлаваЛакки 2 or 3 years is way too conservative in my opinion. I'm sure that before 2025 ends, we will be able to generate some 3D games. It won't be perfect yet, but it will still be impressive. I'm talking about a "Sora Moment" for AI generated games. I think that 2025 will be for AI games like how 2024 was for AI videos. We will see some AI generated games that will make our jaws drop, but at the same time there will still be some limitations.

[@Elwaves2925]

It was a great year all round but this was the year for video. It saw advances at a pace that image generation had seen in previous years. Flux was my big one for the year though, even though it's significantly slower on my 3060 than SDXL, it's people loras and training are way ahead of others. I'm sure some still use it but I have to ask, did SD3 and 3.5 essesntially fail in your opinion? Did many really care by the time 3.5 hit as I hardly ever hear 3.5 talked about now. I wonder if audio will see the same pace of advances next year. It's had a good year but still has a long way to go IMO.

[@theblackbag8531]

Thanks for your videos. If you ever want to have a substack or the sort, I only have so much time and a limited AI budget. It'd be nice to know what the best audio, video, and image AI subscription is at any given moment, live updated. Would pay for that advice for sure...

[@JuanVargas-rb1vj]

No "My name is Tim" at the end with the Spock hand wave? Disappointed :(

[@High-Tech-Geek]

But how much will Veo 2 cost?

[@TheoreticallyMedia]

unknown as of now-- but being Google, it might be (dare I say) free? It's supposed to be integrated into KZbin, but that might be a "Lite" version. But, I don't actually know. I've got my fingers crossed for low/free though!

[@High-Tech-Geek]

@TheoreticallyMedia that would be incredible!

[@Mooooty]

@ 11:58 who the FU%$ is that hunk? And R.I.P. SORA!

[@zrakonthekrakon494]

Overall AI video as we know it was conceived in 2024, it went from incoherent motion and abstract morphing to coherent and usable genuinely quality 5 second clips. 2025 will be the year of consistent characters and long term coherence, the models released in 2024 were very much proof of concepts, runner ups, and major reconstructions to make models functionable and desirable to use. Now that much of the slats are in place, we will see the first generations of true polish, coherence, more features, higher quality for lower prices and longer form video. In 2026 we should see the advent of the first true fully AI generated video media likely in the form of 5-10 minute stories with an identifiable story arc.

[@BiblicalBoldness]

2025 HERE WE COME! 😊

[@TheoreticallyMedia]

I’m soooo fired up for it!

[@BiblicalBoldness]

@ Me too my friend! Love your channel btw!

[@bhavenjain07]

❤❤

[@mysticmaze2268]

I took a page out of your detective performance playbook and just uploaded my FIRST true blue AI shortfilm using Act 1. The funnies part of my video is when I roll credits and ALL the credits are ME. :)

[@mik3lang3lo]

😂😂😂

[@Mooooty]

@ 11:58 who the FU%$ is that hunk?

[@TheoreticallyMedia]

Haha, AI Tim does ALL the working out for me! I'm gonna eat some potato chips while my AI gets all those reps in for me!

[@agnesslovehealz]

Awsome video looking back didn't know needed lol and few forgot what companies overprimised didn't deliver but like tim says dont count them out takes time to get the goodies and some worth the wait and love the ones that pop up n blow our minds so many ai vid gem and then sora drop which is def held nack i hope there's gonna be ai video for flux and grok taking it out had me unsubsidize that was my go to and I love glif w it and krea w flux

[@GRNKRBY]

The cost of the video generation has priced me out of using them as a casual consumer, especially at the level they're at (still very much trial and error and randomness).

[@fredrik241]

What happened to EMO? We saw nothing as good after that.

[@OakSummitNursery]

but wait I didn't catch your name... /s

[@TheoreticallyMedia]

Haha-- I keep thinking about retiring that line-- kind of waffling on it for 25-- we'll see!! Hope you have a great New Years Eve and Day!

[@raymond_luxury_yacht]

Who is this?

[@themightyflog]

Midjourney is so behind now. No control even with poses. No real training or consistency vs stable diffusion. Sora waited too long and everyone jumped ahead. It was a meh release. And where is EMO talker. Still waiting for that…gotta go runway.

[@TheoreticallyMedia]

Agreed that MJ feels behind the ball right now-- but, v7 is coming sometime soon. Hoping that it packs a wallop.

[@michail_777]

TOPS 2024: A team of enthusiastic programmers work on Comfy/AI:video/image/text Kling AI Hailuo AI FLUX SUNO 4 Disappointment 2024: SORA! P.s.- Also KZbin channel Theoretically Media gained more than 130.000. Congratulations! Two Minute Papers has garnered over 1.5 million followers, a testament to the growing interest in our subject matter. Have a great 2025 and good generation, everyone!

[@TheoreticallyMedia]

Two Minute Papers is what I hope to grow up to be one day! haha, such a good channel! And yeah-- it's funny, the narrative of 2024 was for sure set by Sora. It kind of became the through-line of the stories of 24, and when it finally arrived? Well, let's just say the 3rd act was a bit of a letdown. That said, a 2.0 version in 25 could wash all that bad taste away!

[@pdjay8912]

kzbin.info/www/bejne/o5TKamCjrq6XoJIsi=pIVaJhbW0FDV4hvK SIR!!! thank you always for sharing such great news! I tried creating a music video myself using AI. It took me weeks of hard work with countless tools. Except for the music track, I made 99% of it with AI. I hope you enjoy it. Thank you!

[@TheoreticallyMedia]

That’s wild! And great job with that music video! I can for sure see the work you put into it! That is hardly: “prompt and voila” Seriously, great job there and I see the grind! (Solid track as well!!)

[@petercarrick2678]

all i want to know ang tell us true why bother veo 2 if its not available till September as google is saying that's even stupider that sora time rubbish

[@BybbyCreme]

Looking forward to another year with my parasocial AI buddies/team mates, including you Tim. Cheers!