[horhay]

I mean I'm gonna say this with the hype settling down. But it's pretty on par with visually Kling 2 and Veo 2, it happens to output sound pretty ok but having it be one general output along with the visuals is the gamechanger. Beyond that, eh. I've kinda seen people try to take it to the limit and it's pretty much what you'd expect still from their last model

[danielbln]

I think veo2 and Kling are very strong models, but the fact that veo3 is end2end video/audio including lipsync and all other sound is definitely a step change to what came before and I think you're underselling it.

I also expect Google to drive veo forward quite significantly, given the absurd amount of video training data that they sit on.

And compared to the cinemagraph level of video generation we were just 1-2 years ago, boy we've come a long way in very short amount of time.

Lastly, absurd content like this https://youtu.be/jiOtSNFtbRs crosses the threshold for me on what I would actually watch more of.

Veo3 level tech alone will decimate production houses, and if the trajectory holds a lot of people working in media production are in for a rude awakening.

[horhay]

I'm not underselling it. I'm reminding people who get swept up by headlines to actually use the products and be an objective judge of quality when it comes to these things. Because when you lose that objectivity, you start saying things like what you just said. Veo 3 level tech is basically Kling 2/Veo 2 fidelity with native sound generation, so was it that the last generation of these things were already decimating production houses? Be for real. With the tech they had 6 months ago, all they needed to do was add sound manually, which they could have pretty much also generated. A new layer of abstraction isn't "decimating" anything. I'd really take it easy from professing things like that. These things are great for what they are, but let's be actual objective consumers and not fall for these talking points of "oh industries are gonna change in x-months".