Try

Hi @dreimer1986 ! Thanks for report. Apparently newly introduced `shadow_resolve` pass doesn't compile for DX - looking into it

[7dbd988](https://github.com/Try/OpenGothic/commit/7dbd9887edc48d4e9dc1578616848c12bdd6ec6b) - fixes crash on my machine. Please don't close ticket, as I need to develop proper solution + add unit test

> Loading times are quite a bit longer though.. This is probably not related to fix - load time become slower, for the sake of mesh-optimization in 37f7ac68eda50734240b4d96636570adae953604 :D

related: https://github.com/KhronosGroup/SPIRV-Cross/issues/1987 Need to implement masking for DX and probably for metal as well

Tested on code above: ``` [zpy]: Time: 2025-11-16 19:19:37 ``` `grow` is triggered and feature seem to work well. > mutable reference to it which could be stored somewhere As,...

AMD commercial driver has fully broken mesh-shading unfortunately. If you have time - please send bug-reports to them, as we need a driver fix.

> I think they disagree with that being their fault, even though... Even though OpenGothic runs on same hardware with opensource driver just fine :) > Could you elaborate on...

> debugging the project in Windows. Does anyone have experience with it? cmake + VulkanSDK + windows-sdk should be enough. No need to install glslang or set environment variables -...

Thanks @hwnde , for investigating it! > I can create a PR with this solution unless there are better ideas. This is an unease split: * with new found evidence...

Started a post on AMD for forum: https://community.amd.com/t5/newcomers-start-here/driver-crash-when-using-vk-ext-mesh-shader/m-p/667490#M2963 Feel, free to join discussion :) > I'd argue that doing the shader workaround is the approach with least downsides right now....