What Marques Thinks About OpenAI's Sora

Author Avatar

Waveform Clips

Joined: Mar 2024
Spread the love

What Marques Thinks About OpenAI's Sora


In this clip, everyone weighs in on OpenAI’s new AI video generation tool Sora.

Watch full episode:https://youtu.be/LEdYbuwn7DE?si=vJvdZCW8iW-iqzDO

Shop the merch:
https://shop.mkbhd.com…

source

Reviews

0 %

User Score

0 ratings
Rate This

Sharing

Leave your comment

Your email address will not be published. Required fields are marked *

43 Comments

  1. Unable to see vertical images maybe because the encoded orientation of the image isnt at 0 degrees, It may look correct when opened on a desktop, but the AI could be seeing it sideways if its not doing the orientation adjustments like most photo apps does.

  2. 0:02: 🤖 Discussion on OpenAI's new project Sora, potential security concerns, and comparison to Runway features.
    3:59: 🤯 Insight into the capabilities of Transformer and diffusion models for sequential and non-sequential data processing.
    6:21: 🎥 Discussion on innovative AI technology in video production and unique features observed in Sora videos.
    9:22: 🐀 Discussion about a viral image of a rat in a tuxedo with outstretched arms on Twitter.
    12:09: ⚡ Implications of AI-generated drone footage and energy consumption comparison between drones and servers.

    Timestamps by Tammy AI

  3. I don't agree with Marques that Sora's shows massive improvement. The slide-like walk of the lady in the premiering video had put me off in like 3 seconds into the video followed by that leg mixing mentioned at 10:40

  4. People using AI outright for cheap filler content in their productions already leaves a very bad taste. Admitting as much on a podcast like this can be very detrimental to your brand, consider that. I also really wish you would have touched on at least SOME of all the incredibly slippery slopes this development leads us toward. For one guy trying to make a living, this looks like a neat new tool, but for humanity, it's a nightmare, no matter who you are.

  5. I think it should be stressed more just how fast these things have improved. Also, I think a more detailed explanation of the physics 'world model' that is being used to train this thing. Rumor has it, they are using Unreal Engine to train this thing on physics. The coffee cup pirate ship one is a great example of just how amazing these juiced diffusion models are getting.

  6. It'd be not "completely indistinguishable from reality" for another 1-200 years, but it could easily became completely indistinguishable from other digital media like photo or video in a few years.
    Not arguing, just saying.

  7. We are already at the point where some Ai images are indistinguishable from reality. Based on the progress so far there's little doubt that video will get there also. Look out for your boomers, the internet is gonna get even more deceptive.

  8. people who think the vfx/anim industry is done, have no clue how exact, how precise and picky creative choices are made in these productions. simply writing detailed text prompt really is laughable to think it'll give exactly what a director wants. yes, the tech will get better and better but vfx/anim won't be done for a while. it'll be the perfect concept/previs tool for the industry tho, but these productions need every tiny minute detail of control to give exactly what a director wants.