HDR Format Guide

Convert from HDR

Convert to HDR

About HDR Format

HDR (High Dynamic Range) is a raster image format developed by Greg Ward at Lawrence Berkeley National Laboratory in 1985 as part of the Radiance lighting simulation system. The format uses RGBE (Red, Green, Blue, Exponent) encoding to store high dynamic range image data in a compact 32-bit-per-pixel representation, where each pixel consists of three 8-bit mantissa values for the red, green, and blue channels plus a shared 8-bit exponent. This clever encoding scheme allows the HDR format to represent luminance values spanning many orders of magnitude -- from the darkest shadows to the brightest highlights -- while keeping file sizes manageable. HDR files use the .hdr or .pic file extension and are identified by the "#?RADIANCE" magic string in their header. The format supports 32-bit floating point per channel (96-bit RGB) effective precision through its RGBE encoding, making it capable of storing real-world lighting conditions with far greater fidelity than standard 8-bit-per-channel image formats. HDR images are widely used in 3D rendering, architectural visualization, photography HDR workflows, game development for image-based lighting (IBL) and environment maps, and VFX compositing pipelines where accurate light representation is critical.

History of HDR

The Radiance HDR format was created in 1985 by Greg Ward Larson as part of his Radiance lighting simulation and rendering system at Lawrence Berkeley National Laboratory. Radiance was developed to provide physically accurate lighting simulation for architectural and lighting design, and it required an image format capable of storing the full dynamic range of real-world illumination -- something no existing format at the time could do. Ward designed the RGBE encoding as an elegant compromise: by sharing a single exponent across the three color channels, each pixel could be stored in just 4 bytes while still representing luminance values from approximately 1e-38 to 1e+38. The format also includes Run-Length Encoding (RLE) compression to reduce file sizes, particularly effective for images with large areas of similar luminance. Throughout the late 1980s and 1990s, the HDR format became the standard for exchanging light probe images and environment maps in the computer graphics research community. Paul Debevec's landmark 1997 work on recovering high dynamic range radiance maps from photographs brought HDR imaging to broader attention, and the Radiance HDR format became the de facto standard for distributing HDR light probes and environment maps. The format gained widespread adoption in the visual effects and game development industries during the 2000s as real-time HDR rendering and image-based lighting (IBL) techniques became mainstream. While newer HDR formats like OpenEXR (developed by Industrial Light and Magic in 2003) offer additional features such as arbitrary channels, multiple compression options, and deep image support, the Radiance HDR format remains widely used due to its simplicity, broad software support, and the enormous library of existing HDR environment maps and light probes available in this format.

Key Features and Uses

The HDR format's RGBE encoding stores each pixel as four bytes: red mantissa (8 bits), green mantissa (8 bits), blue mantissa (8 bits), and a shared exponent (8 bits). The actual floating-point color value for each channel is reconstructed as mantissa/256 * 2^(exponent-128), providing an effective dynamic range of over 76 orders of magnitude. The file header is ASCII text containing format identification ("#?RADIANCE"), optional variables such as exposure, color correction coefficients, pixel aspect ratio, and the image resolution. The pixel data section uses an adaptive RLE compression scheme that operates on each scan line independently, first separating the RGBE components into four separate sequences and then applying byte-level run-length encoding to each. This approach achieves good compression ratios for typical HDR content, often reducing file sizes by 50-70% compared to uncompressed storage. The format natively supports both latitude-longitude (equirectangular) panoramic images and standard rectangular images, making it particularly well-suited for environment maps and panoramic HDR photography. Modern image libraries including Pillow (Python), OpenCV, stb_image (C/C++), and ImageMagick all support reading and writing HDR files, ensuring broad compatibility across programming languages and platforms.

Common Applications

The Radiance HDR format is used extensively across multiple industries where accurate light representation is essential. In 3D rendering and architectural visualization, HDR environment maps provide realistic lighting for scenes through image-based lighting (IBL), where a single HDR image of a real or virtual environment can illuminate 3D objects with accurate reflections, shadows, and color bleeding. Game developers use HDR environment maps for skyboxes, reflection probes, and ambient lighting in real-time rendering engines such as Unreal Engine and Unity. In photography, HDR files serve as the intermediate format when merging multiple exposure brackets into a single high dynamic range image, preserving detail from the darkest shadows to the brightest highlights before tone mapping to a displayable range. VFX compositing pipelines use HDR images to match CG elements with live-action footage by capturing on-set lighting conditions in HDR light probes. Product visualization studios capture HDR studio environments to light 3D product renders with photorealistic accuracy. The academic and research communities continue to use the format for lighting research, perceptual studies, and tone mapping algorithm development. HDR environment maps are also commonly used as backgrounds for automotive visualization, jewelry rendering, and any application where realistic reflections on specular surfaces are required.

Advantages and Disadvantages