Closed 78Gausen closed 4 months ago
Using the usual kernels for N=2 and N=3. 4, 8 and 16 phases normalized for 128 and 256 (10bit). Meant to reduce horizontal blur on N64 core when VI bilinear is in use. Less phases produce a closer to original output while adding some diffusion.
Using the usual kernels for N=2 and N=3. 4, 8 and 16 phases normalized for 128 and 256 (10bit). Meant to reduce horizontal blur on N64 core when VI bilinear is in use. Less phases produce a closer to original output while adding some diffusion.