Skip to content

DPX Sequence

The dominant format for Digital Intermediate (DI) and VFX plate delivery — many post houses are DPX-first, and Nuke reads it natively. Unlike EXR, DPX almost always carries a log curve rather than raw linear values (a holdover from film scanning, still the industry convention).

Encoding

What curve gets applied before the numbers are written.

  • Cineon log (ACES 2065-1) — default. The classic Kodak film-log curve, applied to ACES 2065-1 data. This is the standard, expected DPX encoding for DI and VFX delivery — use this unless told otherwise.
  • Scene Linear (ACES 2065-1) — no curve at all, straight linear values. Least common for DPX specifically (EXR is more usual for linear delivery) but valid if requested.

Camera-native log curves (ARRI LogC, RED Log3G10, Sony S-Log3, etc.) aren't offered here — real DPX delivery doesn't involve picking a camera-specific curve. Use CineForm's own Log mode for camera-log plate handoff instead.

Bit Depth

  • RGB 10-bit (default) — the traditional DPX standard, tightly packed per the SMPTE spec, and what most facilities expect.
  • RGB 12-bit — more precision, some VFX pipelines request this specifically.
  • RGB 16-bit — maximum integer precision. There is no true "16-bit half-float" DPX — it isn't part of the core SMPTE spec, so this writes ordinary 32-bit float underneath rather than a mislabeled format.
  • RGBA 8-bit (Alpha) — the only bit depth at which DPX reliably carries an embedded alpha channel across real-world tools. Higher-bit-depth alpha embedded in the same file (e.g. 10-bit RGB + alpha) is technically legal in the DPX spec but not reliably read by mainstream applications — use the matte option below instead.

Write Separate Matte Sequence

Off by default. When the frame has an alpha/matte and a 10, 12, or 16-bit RGB profile is selected (not the RGBA 8-bit profile, which already embeds alpha), this writes a second DPX sequence alongside the main one: same frame numbering, in a sibling <name>_matte/ folder, with the matte value written straight-linear into all three RGB channels. This is the real industry-standard way high-bit-depth mattes are delivered with DPX — a second file set, not an embedded channel.