Skip to content

OpenEXR

The professional standard for VFX/compositing handoff and archival. Stores floating-point data with no gamma curve baked in — the most information-preserving format Transduce offers.

Colour Space

Which set of primary colors the numbers are measured against, and what (if anything) gets converted before writing.

  • ACEScg (Linear, Wide Gamut) — default. A direct passthrough of Transduce's internal working space. Nothing converted, nothing clipped — you get back exactly what the app computed, including very saturated colors some source footage might carry. Any OCIO-aware compositing or grading tool recognizes it automatically.
  • Linear Rec.709 (Standard) — converts down to the same color triangle as standard HD/sRGB video. The "works everywhere, no setup needed" choice — After Effects, Premiere, and any non-color-managed tool reads it correctly out of the box. Tradeoff: colors outside that standard gamut get pulled in/clipped, so some of ACEScg's wider range can be lost. Best for general-purpose interop when you don't know the recipient's setup.
  • ACES 2065-1 (Production/AP0) — the official interchange color space of the full ACES pipeline many studios and post houses standardize on. Embeds a flag telling ACES-aware software to route it correctly. Outside a properly configured ACES pipeline, this is the option most likely to look wrong if someone just opens it casually. Only use this if a post house has told you they work in ACES.

Compression

All lossless options preserve every pixel value exactly; only DWAA/DWAB throw away information to save space. Every option works with either bit depth.

  • None (uncompressed) — no compression at all. Fastest to write/read, largest files. Useful for scratch work, or matching a reference file that was itself delivered uncompressed.
  • ZIP (lossless) — the standard general-purpose choice. Compresses in 16-scanline blocks. Default, and a safe one.
  • ZIP1 (lossless, single-scanline) — same compression as ZIP, but line by line instead of in blocks. Some streaming/partial-read tools prefer it; occasionally handles grainy footage slightly differently than block ZIP.
  • PIZ (lossless) — wavelet-based, historically the strongest choice for photographic content with film grain or sensor noise.
  • RLE (lossless, fast) — simple and fast, but weak compression ratio. Rarely the best choice unless speed matters more than file size.
  • DWAA (lossy, compact) — the one lossy option here. Uses a JPEG-like approach in 32-scanline blocks. Often produces much smaller files than any lossless option with no visible quality loss at default settings — an increasingly common choice for final delivery in modern pipelines.
  • DWAB (lossy, tiles) — same idea as DWAA but in larger 256-scanline blocks. Slightly better compression, but block boundaries can be more visible on hard scene changes.

Bit Depth

  • Half (16-bit float) — default, and what almost every VFX pipeline actually uses.
  • Float (32-bit float) — doubles file size for precision headroom that's rarely visible, useful mainly for extreme compositing/HDR work where 16-bit banding has actually been observed.

Include Alpha Channel

Adds a 4th (transparency) channel to the file. Needed for compositing an element over other footage.

Alpha Type

  • Premultiplied — default. RGB is pre-scaled by alpha before writing. This is the convention OpenEXR's own documentation calls "the norm," and what Nuke, Resolve/Fusion, and Flame assume by default when they load a generic RGBA EXR — pick this unless you have a specific reason not to.
  • Straight — RGB keeps its full value regardless of alpha; alpha rides along as independent coverage data. Use this only if the alpha channel is meant purely as garbage/holdout-matte data for a keyer or mask input downstream, not a displayable beauty-with-alpha image.

See Alpha Channels for the full explanation of why this choice matters and what a mismatch actually looks like in a compositor.