Library

oiio v3.2.1.0

Versionv3.2.1.0
Stars★ 2,375
Released2026-10-01

Reading, writing, and processing images in a wide variety of file formats, using a format-agnostic API, aimed at VFX applications.

Release notes

Release 3.2 (Oct 1, 2026) -- compared to 3.1
-------------------------------------------------------
- Beta 1 (3.2.0.3-beta1): Sep 1, 2026
- Beta 2 (3.2.0.4-beta2): Sep 15, 2026
- Release Candidate 1 (3.2.0.5-RC1): Sep 27, 2026
- v3.2.1.0 supported release: Oct 1, 2026

**Executive Summary / Highlights**
  - JPEG support for writing Ultra-HDR images.
  - FLIP perceptual image comparison via `ImageBufAlgo::flip_diff()` and
    `oiiotool --flipdiff`.
  - Many color management improvements: use of Color Interop Forum (CIF)
    staandardized names throughout, support for CICF codes in file formats
    that allow them, better ICC profile R/W support.
  - Improved "thumbnail image" support by several image file formats, and via
    new oiiotool commands for adding and extracting thumbnails.
  - Dozens and dozens of security/bug improvements to harden the various file
    format readers against corrupted or malicious image file input.
  - Switch Python bindings to [nanobind](https://github.com/wjakob/nanobind)
    by defalt.
  - New ABI version namespacing scheme makes 3.2 fully API, ABI, and link
    compatible with OpenImageIO 3.1.

### ABI backwards-compatibility -- NO CHANGE

Unlike previous yearly releases. OpenImageIO 3.2 is 100% backward compatible
with 3.1 releases in its API and also ABI / link compatibility. There may be
new functionality added, but any API call supported by 3.1 still exists and is
link compatible.

This means, for example, that if you are making a large appication that
currently embeds OpenImageIO 3.1 and exposes its APIs to user-written dynamic
plugins/modules (so that they can make calls to the built-in OIIO), your app
can upgrade to OpenImageIO 3.2 without losing compatibility with those plugins
-- all the link symbols that existed before continue to exist with the same
call signatures.

### New minimum dependencies and compatibility changes:
  - The deprecated **icc compiler** is no longer supported. (3.2.0.0)
  - **CMake**: Minimum 3.18 -> 3.23. [#5394](https://github.com/AcademySoftwareFoundation/OpenImageIO/pull/5394) (3.2.0.3)
  - **fmt**: Minimum 7.0 -> 9.0.
  - **libjpeg**: Minimum 8 -> 9. Note that libjpeg-turbo 2.1+ still satisfies this. [#5432](https://github.com/AcademySoftwareFoundation/OpenImageIO/pull/5432) (3.2.0.3)
  - **LibRaw**: Minimum 0.20.0 -> 0.21.0
  - **libtiff**: Minimum 4.0 -> 4.1. [#5414](https://github.com/AcademySoftwareFoundation/OpenImageIO/pull/5414) (3.2.0.3)
  - **DCMTK** (optional, DICOM support): Minimum 3.6.1 -> 3.6.2. [#5438](https://github.com/AcademySoftwareFoundation/OpenImageIO/pull/5438) (3.2.0.3)
  - **SIMD baseline**: Builds targeting x86-64 now default to an SSE4.2 instruction baseline (`USE_SIMD=sse4.2`) instead of inheriting the compiler default (often plain SSE2), so OIIO's SSE4 fast paths are used out of the box. Every x86-64 CPU since 2008 has SSE4.2. Use `USE_SIMD=0` to disable SIMD or `USE_SIMD=sse2` for the old behavior. `USE_SIMD` is also now architecture-aware, so x86 tokens no longer leak into ARM builds. [#5411](https://github.com/AcademySoftwareFoundation/OpenImageIO/pull/5411) (3.2.0.3)
  - **Python bindings**: The default binding backend is now nanobind rather than pybind11; configure with `-DOIIO_PYTHON_BINDINGS_BACKEND=pybind11` to keep the old backend. [#5430](https://github.com/AcademySoftwareFoundation/OpenImageIO/pull/5430) (3.2.0.3)

### ⛰️  New features and public API changes:
* *New image file format support:*
  - *jpeg*: Support writing Ultra HDR images (previously only reading was supported). [#5335](https://github.com/AcademySoftwareFoundation/OpenImageIO/pull/5335) (by @mugulmd / Loïc Vital)
  - *openexr*: Add support for the `lj2k` (lossy, JPEG2000-based) and `zstd` OpenEXR compression modes, when OIIO is built with OpenEXR 3.5 or later. [#5495](https://github.com/AcademySoftwareFoundation/OpenImageIO/pull/5495) (by @QuantaDude / Abhirup Bhattacharyya) (3.2.0.5) [#5520](https://g…

Share this resource


Discovered 2026-10-02 Source GitHub Archive 2026-10 →