I have noticed when I updated three.webgpu.min.js to r180 the frame rate dropped? The r179 was faster.
Did you compare loaded texture assets? If the culprit is quality levels, you can differ app-level resources per-release. My hero was 13 seconds slower, because a (broken) linked resource url became requested. Sometimes you can just disable textures/features to bisect the “bug”, but detailed app logs would be more informative and resilient.
Thanks for the feedback. I want to clarify a few points so we don’t talk past each other.
Regarding proportions, in my current implementation the interior parallax remains visually correct even when the room is stretched. I’m not claiming the original approach is mathematically ideal for non-cube rooms. Only that in practice it holds up well for the use cases I’m targeting.
For reference, here is the interior parallax component I’m using:
https://github.com/kylebakerio/aframe-interior-parallax-mapping
This is an A-Frame interior parallax mapping implementation, and the behavior I’m seeing is consistent with its intended usage.
On the texture inversion, I’ve already investigated the library side and verified that the issue is not coming from the source itself. The GitHub link was provided purely for reference.
Visit location: https://theneoverse.web.app#serini&&serini&&loc=platform1_edge001,-1247.7,6.3,-3200.1,-171.7
Visit location: https://theneoverse.web.app#serini&&serini&&loc=platform1_edge002,-1242.5,6.3,-3248.6,-127.7
Visit location: https://theneoverse.web.app#serini&&serini&&loc=platform1_edge002,-1244.5,6.3,-3243.6,124.5
As for the repositories, the minimal GitHub repos are intentional. This one, for example, serves as a portfolio entry and gateway to the commercial assets rather than a full open-source codebase:
https://github.com/VeinSyct/theneoverse
If someone wants to inspect the source code, it is already accessible through the browser. The demo can be examined via page source or downloaded using Chrome DevTools. There is no single public repository because each project is different and tailored to its own requirements.
There is no intent to mislead or market aggressively in the thread. The Showcase post was meant to demonstrate the visual result, not to request a code review. That is why the demo is minimized.
I appreciate you taking the time to respond, and I’m happy to keep the discussion focused on the technical behavior of the interior parallax itself.