GLTF v/s S: Casual Questions

[TrainGuy6550]

Four contestants: Uncompressed .S, Compressed .S, GLTF with PBR materials, GLTF without PBR materials

Theoretically, without considering the development state of Open Rails, and without caring about piracy, which is better for:
1) Best performance
2) Best visuals
3) Balanced performance and visuals

4) Is GLTF without PBR materials better than compressed .S ? I mean I'd always like a performance boost on my crappy PC even if it's very subtle...

[pwillard]

Without knowing any facts and basing it on complexity, I'm going to guess that Uncompressed S is more performant than Compressed S (one time, until it's in RAM), GLTF without PBR is more performant than full GLTF with PBR, but in both cases, GLTF is going to be better visually than anything an S file can produce. GLTF without PBR is likely to be the most balanced between visual quality and performance, but why only go halfway (other than not being familiar with PBR techniques)?

To me, it's time to stop being so restrained by the best practices of the distant past.

[TrainGuy6550]

The only problem is my computer is from the past (good grandfather Pentium with intel HD Graphics iGPU about 13 yo still serving faithfully) 😂
(Well, PBR materials are heavy and I can't afford to have them anyway. But if gltf without PBR will give better performance than compressed .S, I can atleast have my own models performing as I want them to

P.S. Google seems really confident that gltf must be better than .S

When comparing the Kuju Legacy .s (Shape) format from Microsoft Train Simulator (MSTS) and glTF 2.0 (without PBR) solely in terms of rendering and engine performance, glTF completely outperforms the Kuju .s format.

Copy-pasted from Google's new AI mode