Neural block texturing example

A developer demonstrated a neural block texturing method that compresses a full PBR material stack — normal, roughness, albedo, and ambient occlusion maps — into a single 320-bit record per 6×6 block. The approach pairs shared 4+4-bit selectors with an 86×86×4 bilinearly sampled block latent and a 1,335-weight 8→27→27→12 MLP decoder trained via evolution strategies and central finite differences rather than backpropagation. It achieves 32.27 dB weighted reconstruction quality (normal 28.94, roughness 35.40, albedo 32.52, AO 36.09 dB) at 2.24 bits per pixel raw and 2.16 bpp entropy-coded.

- Material: normal, roughness, albedo, AO PavingStones070, 512×512 padded to 516×516 - Selectors: 516×516×2, 4 + 4 bits per pixel, shared by all four maps, exact search each iteration - Block latent: 86×86×4 6×6 blocks , 8 bits/value, sampled bilinearly, ES-trained - Decoder: 8→27→27→12 MLP, 1335 weights, decodes all four maps from one record - 8000 iterations, ES 256 pairs then central finite differences, lr annealed, no backprop - 32.27 dB weighted; normal 28.94 / roughness 35.40 / albedo 32.52 / AO 36.09 dB - 320 bits per 6×6 block for the whole stack: 2.24 bpp per texture raw, 2.16 entropy-coded - Selectors: 516×516×2, 4 + 4 bits per pixel, shared by all four maps, exact search each iteration - Block latent: 86×86×4 6×6 blocks , 8 bits/value, sampled bilinearly, ES-trained - Decoder: 8→27→27→12 MLP, 1335 weights, decodes all four maps from one record - 8000 iterations, ES 256 pairs then central finite differences, lr annealed, no backprop - 32.27 dB weighted; normal 28.94 / roughness 35.40 / albedo 32.52 / AO 36.09 dB - 320 bits per 6×6 block for the whole stack: 2.24 bpp per texture raw, 2.16 entropy-coded Decoded 512x512 material: