11 releases (7 breaking)

0.10.1 Sep 1, 2021
0.7.1 Dec 31, 2020
0.7.0 Jul 21, 2019
0.6.0 Jun 15, 2018
0.5.0 May 28, 2017

#4 in Multimedia

Download history 17/week @ 2021-06-06 13/week @ 2021-06-13 7/week @ 2021-06-20 13/week @ 2021-06-27 12/week @ 2021-07-04 11/week @ 2021-07-11 21/week @ 2021-07-18 21/week @ 2021-07-25 30/week @ 2021-08-01 14/week @ 2021-08-08 19/week @ 2021-08-15 11/week @ 2021-08-22 31/week @ 2021-08-29 6/week @ 2021-09-05 10/week @ 2021-09-12 8/week @ 2021-09-19

65 downloads per month

BSD-3-Clause

4MB
14K SLoC

OpenEXR

Bound version: 3.0.5

The openexr crate provides high-level bindings for the ASWF OpenEXR library, which allows reading and writing files in the OpenEXR format (EXR standing for EXtended Range). The OpenEXR format is the de-facto standard image storage format of the motion-picture industry.

The purpose of EXR format is to accurately and efficiently represent high-dynamic-range scene-linear image data and associated metadata, with strong support for multi-part, multi-channel use cases.

OpenEXR is widely used in host application software where accuracy is critical, such as photorealistic rendering, texture access, image compositing, deep compositing, and DI. OpenEXR is a project of the Academy Software Foundation. The format and library were originally developed by Industrial Light & Magic and first released in 2003. Weta Digital, Walt Disney Animation Studios, Sony Pictures Imageworks, Pixar Animation Studios, DreamWorks, and other studios, companies, and individuals have made contributions to the code base.

OpenEXR is included in the VFX Reference Platform.

The openexr crate is maintained by the vfx-rs project.

Quick Start

To use the included C++ OpenEXR source:

cargo add openexr
cargo build

While this method is supported and easiest to get starter, it is strongly recommended that you build and install the C++ library separately and build the crate against it like so:

cargo add openexr
CMAKE_PREFIX_PATH=/path/to/cmake/configs cargo build

Note that you must take care to ensure that the version of OpenEXR you are pointing it to is the same as that for this version of the crate, otherwise you will encounter linker errors since all OpenEXR symbols are versioned.

The prelude pulls in the set of types that you need for basic file I/O of RGBA and arbitrary channel images:

use openexr::prelude::*;

fn write_rgba1(filename: &str, pixels: &[Rgba], width: i32, height: i32)
-> Result<(), Box<dyn std::error::Error>> {
    let header = Header::from_dimensions(width, height);
    let mut file = RgbaOutputFile::new(
        filename,
        &header,
        RgbaChannels::WriteRgba,
        1,
    )?;

    file.set_frame_buffer(&pixels, 1, width as usize)?;
    file.write_pixels(height)?;

    Ok(())
}

fn read_rgba1(path: &str) -> Result<(), Box<dyn std::error::Error>> {
    use imath_traits::Zero;

    let mut file = RgbaInputFile::new(path, 1).unwrap();
    // Note that windows in OpenEXR are ***inclusive*** bounds, so a
    // 1920x1080 image has window [0, 0, 1919, 1079].
    let data_window: [i32; 4] = *file.header().data_window();
    let width = data_window.width() + 1;
    let height = data_window.height() + 1;

    let mut pixels = vec![Rgba::zero(); (width * height) as usize];
    file.set_frame_buffer(&mut pixels, 1, width as usize)?;
    file.read_pixels(0, height - 1)?;

    Ok(())
}

Beyond that, types related to deep images are in the deep module, and tiled images are in the tiled module.

The Reading and Writing OpenEXR Image Files document is a great place to start to explore the full functionality of the crate. It contains example usage for nearly everything.

Math Crate Interoperability

OpenEXR (and much of the rest of the VFX ecosystem) relies on Imath for basic math primitives like vectors and bounding boxes.

Rust already has several mature crates for linear algebra targetting graphics such as cgmath, nalgebra, nalgebra-glm and glam. Rather than adding yet another contender to this crowded field, we instead provide a set of traits that allow any of these crates to be used with openexr in the form of imath-traits. By default, these traits are implemented for arrays and slices, so you will find that the examples in this documentation will tend to use e.g. [i32; 4] for bounding boxes:

use openexr::prelude::*;
fn read_rgba1(path: &str) -> Result<(), Box<dyn std::error::Error>> {
    use imath_traits::Zero;
    let mut file = RgbaInputFile::new(path, 1).unwrap();
    let data_window = file.header().data_window::<[i32; 4]>().clone();
    let width = data_window.width() + 1;
    let height = data_window.height() + 1;
    Ok(())
}

To use your preffered math crate instead, simply enable the corresponding feature on openexr, which will be imath_<name>, for example:

cargo build --features=imath_cgmath

Now you can use types from that crate together with openexr seamlessly. In the case that the math crate does not provide a bounding box type, one will be available as imath_traits::Box2i and imath_traits::Box3i.

use openexr::prelude::*;
#[cfg(feature = "imath_cgmath")]
fn read_rgba1(path: &str) -> Result<(), Box<dyn std::error::Error>> {
   use imath_traits::Zero;
    use imath_traits::Box2i;

    let mut file = RgbaInputFile::new(path, 1).unwrap();
    let data_window: Box2i = *file.header().data_window();
    let width = data_window.width() + 1;
    let height = data_window.height() + 1;
    Ok(())
}

Features

  • High dynamic range and color precision.
  • Support for 16-bit floating-point, 32-bit floating-point, and 32-bit integer pixels.
  • Multiple image compression algorithms, both lossless and lossy. Some of the included codecs can achieve 2:1 lossless compression ratios on images with film grain. The lossy codecs have been tuned for visual quality and decoding performance.
  • Extensibility. New image attributes (strings, vectors, integers, etc.) can be added to OpenEXR image headers without affecting backward compatibility with existing OpenEXR applications.
  • Support for stereoscopic image workflows and a generalization to multi-views.
  • Flexible support for deep data: pixels can store a variable-length list of samples and, thus, it is possible to store multiple values at different depths for each pixel. Hard surfaces and volumetric data representations are accommodated.
  • Multipart: ability to encode separate, but related, images in one file. This allows for access to individual parts without the need to read other parts in the file.

Dependencies

~4.5–6.5MB
~157K SLoC