Can you tell us how the workflow looks like

right now? So in this case, we don't actually generate the Gaussian Splat, but we can consume it.

So we do support two different FL formats.

Right now, we support PLY, so you can import from any different Gaussian Splat generation software.

we also support custom formats like lcc and we're having a look at lcc2 so the new more optimized version from xgrid so in this case we still cannot rely but you can import them and in this case the way we are focusing it is you know trying to use the gaussian splats in the place where they make most sense so sometimes if you talk about something like a car it's something that is actually a bit against gaussian splatting because of there's you know a lot of reflective surfaces a lot of glass and these things that are really hard to be captured in the gaussian splat and we're trying again to put it where it makes sense so if you think about a customer let's say a car maker that wants to develop a product their main product is always going to be the car it's never going to be things like capturing the environment generating 3D assets.

So it's a lot of things where time was spent but didn't actually bring value to them.

So many of the features that we're trying to work on with Gaussian Splats are actually onto putting them where it makes sense.

Now, is that solution still supported? Are there some lessons that you have taken? Are those two separate teams? How does this XR is driven inside Autodesk

Maybe you should just break down the differences there, and then I have a couple questions for you about that.

There was a missing component in most of the open source models for, for image generation that have been released pretty much since 2023, when Stable Diffusion 1.5 was, was released.

And the problem with these open source models is that they were overly post-trained.

Uh, when you post-train a model, you give the model an opinion, and you make it to forget how to make, uh, like, a wide range of styles or a wide range of things that are considered wrong as you run, uh, reinforcement learning techniques or other kind of post-training techniques.

Uh, the problem, the problem here, the problem with only having post-trained models is that they are very hard to tune, and the coolest part of having access to an open source model is that you are able to tune it, and you are able to make it work really good for a specific niche that it's valuable for your business or that it's valuable for the use cases that you have for this.

That on the one side, and on the other side, it's really hard to innovate on post-training techniques without having access to a raw model.

Because as soon as the model has been post-trained, anything that you can do farther that is just, like, very hard to, to, like, actually be able to do research.

Less performant, faster, cheaper.

What is Krea in the world of among those? Where do you place yourself? Why people, who you serve, and why people should use Krea?

So Krea is a creative tools company.

If there's something that we wanna do is build creative tools, and we believe that cr- future creative tools are gonna be built on top of AI models, and we are exploring all the different ways how AI can be used as a creative medium.

So that's at the most high level, and that's why we never cared about is this open source, is this in-house, is this an API, is this whatever.

We always cared more about the capability and what this technology allows you to do as a creative and how do we put it into an interface that makes sense than really anything else.

So that, that I would say is at the highest level, Krea, the g- the goal that we have as a company is to build a future of creative tools.

I- i- if you wanna see it as a whatever the new Adobe may be called, it may not be like-- it may not look anything like Adobe, but whatever that same equivalent is in the future, that's what we want Krea to be.

Wow.