It was a long road to get here, but after much trial and error (mostly error), and then realizing my code was full of mistakes and I should install the #vulkan SDK for it's validation layer, I have finally got working 32bpp textures in my ray marching engine!
Spent the better part of the evening devising a hopefully non-branchy procedural UV generator

#3dgraphics #glsl

Writing a post-processing shader to upscale my compute program output was a pretty decent experience in #vulkan. Just create a new pipeline using the same descriptor set layouts, drop in a float array as the framebuffer, sprinkle a few commands for dispatch and blammo! Finally a respectable FULL-HD™® experience. I really need to fix those glitchy too dark/light pixels though.

#graphics #3dgraphics #compute

I'm working on a #vulkan raycasting engine in a compute shader without the burden of an entire rasterization pipeline and neeed to figure out the upscaling situation. I tried to get my swapchain --> #SDL3 surface to scale but the documentation is not clear how to accomplish that. It implies scaling is automatic if the surface is larger but that's not happening.

To preserve my sanity I'll create another compute shader for the task instead of trying to guess how swapchain image scaling is supposed to work. Yeah I tried setting all combinations of VK_PRESENT_SCALING_*_BIT_EXT and VK_PRESENT_GRAVITY_*_BIT_EXT, no dice.

prototype raymarching SDF engine experiment #003.

Using a smoothMin function let's us combine geometric primitives to form complex surfaces. Now to figure out how to upscale the final swapchain image with a single #vulkan #compute "shader" because my poor cheap minipc struggles to cast more than 900x600 rays@30fps. #screenshotsaturday

Got back into #gamedev a bit recently and this is one of the most wholesome bugs I've ever encountered. #signeddistancefunction and #raymarching is so exceedingly bodacious, we can have accidental shadows on our debug #octree grid lines.

#sologamedev #solodev #indiedev #vulkan