High-dynamic-range format with 32-bit floating point — stores layers, alpha, metadata losslessly. Industry standard for VFX renders and comp pipelines.
Anyone rendering or compositing in VFX can't get around OpenEXR — the format is the de facto standard in professional pipelines because it can do things that JPEG or PNG simply cannot. The decisive advantage: 32-bit floating-point per channel. This means you don't just store 0-255 brightness values, but an enormous range of values — overbrights exceeding 1.0 are preserved. This is worth its weight in gold for compositing: you can recombine lights without them already being clipped.
The technical strength lies in its flexibility. An EXR file can store multiple render passes in one file — Diffuse, Specular, Normal, Depth, Occlusion, Shadow Matte — all in one container. In the compositor (Nuke, After Effects), you load one file and immediately have access to all layers. This saves storage space and makes management clearer. In addition, there is the alpha channel with full resolution, not compressed as with some other formats. Metadata — camera parameters, render settings, custom attributes — can also be embedded. This greatly aids documentation and later adjustments.
In practice: You render your 3D scene from Maya, Arnold, or V-Ray directly as an EXR sequence. 16-bit or 32-bit, depending on the requirements. The file sizes are larger than JPEG, but that's not a real problem with modern storage. What's important is the compression — ZIP, PIZ, or DWAA — which helps without losing quality. In the editing workflow, you can also use EXRs as a proxy format to edit 4K material more smoothly.
A practical note: Not all video players understand EXR natively. For quality control on set or in review, you need specialized tools or have to convert to ProRes/DNxHD. But within the post-production pipeline itself, EXR is the working material — from rendering to the final composite. Those working with high-quality visual effects cannot do without it.