https://turbowarp.org/1369144128 what makes it unique? - it stores only 1 integer per pixel in its framebuffer, packing both 24bit z-depth and 24bit triangle index. - its triangle function is one of the fastest on scratch, drawing roughtly 20k triangles per second on my laptop at 480x360p resolution. - it supports writing into multiple canvases of different dimensions at the same time. - using the [set current canvas(id)] block, you can determine into which canvas the triangles are supposed to be drawn into. - it natively supports shaders. SGL 1: - first prototype. - deferred pipeline. SGL 2: - complete rewrite. - forward pipeline. - multi-canvas system - no rasterizer. SGL 2.1: - rasterizer implimented. SGL 2.2: - vertex data for the rasterizer. SGL 2.3a: - fast deferred pipeline - visibility buffer - shader support - faster triangle filler - experimental Tri3D SGL 2.4a: - 2x faster triangle filler - more memory efficient SGL 2.4a: - 1.3x faster triangle filler SGL 2.4: - functioning Tri3D function with near-clipping - 3D transformation matrix utilities SGL 2.4.1: - better Tri3D SGL 2.4.2: - fully functioning Tri3D - Tri2D bug fixes - perspectve-correct barycentric coordinates - 3D Demo SGL 2.4.3: - immense performance speed up - 3x more triangles per second SGL 2.4.4: - optimizations to the fragment reconstruction
