Bit Depth

Bit depth is the number of bits used to represent each color channel of each pixel in a digital image. It directly determines the number of discrete tonal values available per channel and therefore the smoothness of gradients, the headroom for grading and the resistance to banding. 8-bit per channel (24 bits per pixel for RGB) provides 256 values per channel and is the consumer/web standard — fine for finished display content but inadequate for grading (any aggressive correction will introduce banding in skies and shadows). 10-bit per channel (1024 values) is the broadcast and mastering minimum, required for HDR (PQ/HLG encoding cannot work at 8 bits without severe artifacts) and standard for ProRes 422, DNxHR HQX, HEVC Main 10. 12-bit per channel (4096 values) is used in ProRes 4444 XQ, DNxHR 444, DCP, Dolby Vision and finishing-grade pipelines where extensive grading happens downstream. 16-bit per channel (used in OpenEXR half-float and some VFX intermediate formats) gives extreme headroom for compositing math. Float bit depths (32-bit linear EXR) preserve values above 1.0 for HDR scene-linear compositing. In post-production, the rule is to capture and grade at the highest practical bit depth, only reducing at the final delivery encode where 8 or 10 bits is acceptable. Bit depth applies to both luma and chroma; chroma subsampling can reduce effective chroma bit depth even with deep luma. Reference: ITU-R BT.2100 (HDR system parameters) and the OpenEXR specification.