Sora 2 vs Veo 3.1 vs Kling 2.5 vs Seedance 1.0: What Does Physics Best?

Heat. Heat.

[Music] [Music] [Music] [Music] There's more physics comparisons coming up in just a second. Just letting you know that you can find every single prompt that was used to generate all of these clips.

Those prompts are inside a Google document that I'll pop a link to below in the description to this video.

Secondly, why do some of these clips look really similar and others look really different?

We have Sora 2 to thank for that. Sora 2's censorship policy meant that anything with people in it, I wasn't able to use my own image as the starting frame for that video.

So, the clips with people in them look different because they were generated from the ground up from text.

whereas the comparisons have no people in them.

We're just using my own image as the starting frame in each of the models.

Right, let's get back to the comparisons.

Let me know what model you think does physics best in the comments below.

And at the end of the video, I'll give you my verdict based on what I saw creating all these clips.

Ready? Go. Weave. Good girl. Keep going.

Yes. Ramp up. Down. Down. Nice job. Woo.

Heat. Heat.

[Applause] [Applause] Heat up [Music] here.

Heat. Heat.

[Music] Heat.

Heat.

[Music] Cool.

Let us know which model you thought physics best in the comments below.

Personally, after creating all these clips myself and and seeing these models firsthand, for me, VO 3.1 and Cling 2.5 are like the top two, I would have to choose VO as the best one purely because it generates videos with sound versus cling, which doesn't.

But in terms of physics, uh they're both pretty neck and neck. Luckily, I was using an all-in-one website to generate these clips really quickly. Uh, I'll put a link to that platform in the description, too.

There's 20% off at the moment, so it's one of the cheapest ways to generate all of these clips.

Sora just took way too long to create clips.

It takes like five to 10 times longer than the other models to generate a clip.

Seed dance was just awful.

It made so many mistakes. Like, seriously, go back and watch this video and just watch the seed dance generations cuz they're quite easy to miss on first viewing, but it made so many mistakes. Let me know what else you'd like to see compared in future videos.

I'm thinking maybe I could do a sports comparison or like an exercise body movement comparison video.

Maybe test some different models.

Let me know everything you want to see in the comments below and I'll try and do those in the future. Wasn't sure generating the best possible clips with these models.

You might want to think about turning them into like a long length film.

Here's a tutorial I made in the past on how to quickly generate clips using these tools and turn them into long films all in the same platform.

Uh so give that video a watch next.

Subscribe to the channel and I'll see you in the next one. See around.