Change Depth Node

Change Depth lets you force the depth of the RGBA channels to 8, 16 or 32-bit (per channel). In a pro pipeline, you often work in 32-bit float for maximum precision, then downconvert at the end of the chain for delivery or optimization.

The Dither parameter is crucial when downconverting: it adds controlled noise to hide the banding that would otherwise appear in transition areas (skies, gradients).

• Effects Library: Tools > Miscellaneous > Change Depth
• Quick add: right-click in the Node Editor → Add Tool > Miscellaneous > Change Depth

8-bit

256 values per channel (0-255). The JPEG and legacy TV standard. Visible banding on gradients. Light on memory.

16-bit

65536 values. The standard for HDR / higher quality delivery. A quality / memory compromise.

32-bit float

Unlimited floating-point precision (including negative values or values >1 for HDR). The pro compositing standard. Heavy on memory but keeps every operation lossless.

• Input (orange) - the 2D image to convert.
• Effect Mask (blue) - optional.

32-bit pipeline with a final downconvert:
MediaIn (32-bit float)
   → [all compositing in float]
   → Color Corrector
   → Change Depth (Depth = 16-bit, Dither = Error Diffusion)
   → MediaOut (16-bit)

Memory optimization:
Loader (32-bit) → Change Depth (16-bit) [early in the chain]
              → faster / lighter processing
              → MediaOut

8-bit delivery preparation:
32-bit compositing → Change Depth (8-bit, Dither = Additive Noise)
                  → Saver JPEG/PNG

Depth (default: Keep)

Target depth. Keep: keeps the input depth. 8-bit / 16-bit / 32-bit: forces the chosen bit depth.

Dither (default: Error Diffusion)

Noise algorithm used to hide banding when downconverting. Error Diffusion = the recommended standard. Additive Noise = an alternative. None = no dithering (banding may be visible).

• Downconvert 32→16 after color - the classic pipeline: color in float, downconvert to 16-bit for delivery.
• Memory optimization - on very long projects or huge resolutions: downconvert early to save RAM / disk.
• Converting to an 8-bit format (JPEG, PNG) - web delivery or 8-bit archives: Change Depth with dither for visual quality.
• Upconvert for processing - 8-bit footage with heavy operations: Change Depth → 32-bit to avoid clipping during processing.
• Pipeline standardization - force every source to the same bit depth for a consistent pipeline.

• Always dither when downconverting - otherwise banding shows up in gradients. Error Diffusion = a safe default.
• 32-bit for HDR - if values > 1 or negative values are expected (HDR, log), 32-bit float is required.
• Downconvert at the end - keep float during compositing and only convert at the end for delivery.
• Keep mode as the default - leave Keep in templates so you do not accidentally force a downconvert.
• Additive Noise for low detail - scenes with few fine details: Additive Noise can give a more organic look than Error Diffusion.
• Check the Saver output - the Saver can impose a bit depth depending on the format (PNG = 8-bit, EXR = 16/32). The Change Depth before the Saver should match.
• For archive workflows, keep at least 16-bit to allow for future reprocessing.