Related issue: -
The editor can now load PCD assets which makes it easier to debug PCD files. The PR also ensures the number of vertices is now correctly displayed for point clouds in the info section.
To make this change work, it was necessary to remove the file name extraction from
PCDLoader.parse() which requires an URL. Since this kind of logic is unusual for a loader, it's better to remove it in any event.
I can try to make a PR?
I don't think there's a way of using the latest Chrome Canary in Puppeteer so until the WebGPU API stabilizes it won't be able to work I think.
Describe the bug
webgpu examples are not loading on page
Steps to reproduce the behavior:
A clear and concise description of what you expected to happen. Console report error : Uncaught ReferenceError: GPUShaderStage is not defined at WebGPUNodeBuilder.js:21:12
This should solve it: https://github.com/mrdoob/three.js/commit/235e6c354f175f6a5aecc4db0341cc7056a3102a
I've just removed the original trial meta tags locally and the problem is still there. Looking into this...
I think the problem is that we're using Chrome's origin trial which, realistically, is not useful.
Related issue: #24047
Applies the pattern of #24047 to the examples.
Docs: Update "How to run things locally" (#24097)
Current wording offers two options and suggests we'll explain both, but then only explains the second. As discussed in #15334 we should not encourage users to change their browser's security settings.
Current wording offers two options (change browser settings, or use local server) and suggests we'll explain both, but then only explains the second. As discussed in #15334 we should not encourage users to change their browser's security settings. This PR simplifies the wording to present a local server as the default option.
I noticed depth precision issues (horizontal banding + z-fighting) when rendering my project with post-processing (using WebGLRenderTargets) starting at very short distances (10 - 15 meters, where 1 unit = 1 meter), and those issues did not happen when rendering the same scene directly to canvas !
So I investigated, and it turns out setting a DepthTexture with the
UnsignedIntType type into
renderTarget.depthTexture fixed my issues. Precision issues only appear at much greater distances now (as expected) and exactly where they start appearing when rendering directly to canvas.
About this PR
So I changed the default to
UnsignedIntType because I believe defaults should give best results out of the box, but I don't know what I'm doing, I didn't test anything beyond my project, and I don't know the implications of this change ⚠️
(also updated the docs + bits of code that seemed relevant)
This example: https://threejs.org/examples/webgl_depth_texture.html helped me to visualize different types, just zoom out until the meshes are barely visible and you should notice banding with
Just learnt about this event and I thought we should probably listen to it.