Some sniffers have found out they use some variation of this http://hhoppe.com/proj/gim/​ and https://pages.jh.edu/~dighamm/research/2004_01_sta.pdf​

That's surely not what they do. I work on some tech similar to GIM for some reasons. If they would use it, they would have to mention longer precomputation times, but they say model import is fast.

Talked in more detail about it here: https://forum.beyond3d.com/posts/2126759/​

(also found post showing discrete LOD clusters: https://forum.beyond3d.com/posts/2127466/)​​

Ofc. Its likely they store stuff in textures to utilize HW compression, but i do not think they mess with parametrization to get displacement mapping.
The clusters from debug view looks more like a result from simple triangle reduction. They seem to have no need for displacement at all.

@joej im intrigued on the GIM project you are working. Can I share something?

I could not share anything on the geometry side that would be very interesting yet. Give me some time, like ‘late 2021’ : )

@joej deal. It's just that I am also experimenting with GIM. Well for now just reading stuffs ;D

trsh said:
Can I share something?

Err, from the context i thought the I was a mistake and you wanted me to share my GIM related stuff.

But if YOU want to share something, i'm ofc. interested.

And i can tell you all about the hell you are planning to bath in :D

No you read my mind correctly. I wrote “I” instead of “you” by mistake :D

The big problem with open source project is lack of coordination and unified design, people bring their expertise without much coordination. As the project gets bigger, the learning curve also gets longer for new comers. Complexity will grow uncontrollable and new features will be difficult to add without revising original code base.

Project integration also becomes big issue. How many electricians needed to fix a light bulb? Exactly.