Adaptive local contrast and dynamic range compression on CUDA (ALTMapper)

Updated: July 20, 2026

ALTMapper is the adaptive local‑contrast and dynamic‑range compression module of the Fastvideo SDK, running entirely on NVIDIA CUDA. It lifts shadows, strengthens local contrast and preserves detail from a single frame — on high bit‑depth HDR sources and on ordinary 8‑bit images alike — fast enough for real‑time video, with no exposure bracketing.

adaptive local tone mapping on NVIDIA CUDA: source frame and ALTMapper result

Source frame on the left, ALTMapper result on the right — recovered shadow detail and controlled local contrast.

How the engine works

The algorithm transforms image intensity using information from local regions of every frame. It divides each image into a grid of rectangular tiles, computes a small set of local parameters per tile, and combines local and global parameters into a set of local tone curves applied per pixel. Keeping the parameter count small is what makes the processing fast.

Capabilities

Fast, precise adaptive local tone mapping for HDR management
Local and global contrast preservation and enhancement
Shadow recovery, local gain control, detail preservation
ISP with 16/32‑bit accuracy, compact footprint, no AI
Compatible with FastVCR; Windows 10/11, Linux Ubuntu, L4T (Jetson Orin)
Supported NVIDIA GPUs: Jetson, GeForce; CUDA‑12.6

Why not CLAHE/AHE

Many local tone mapping tools are based on AHE or CLAHE. They can look acceptable on still images but are not stable: CLAHE tends to cover uniform‑brightness areas with unnatural shadows, and in video the output is often marred by dark noisy spots and visible tile borders from the internal partitioning. ALTMapper is built to avoid these artefacts in video.

Benchmarks

Hardware	Resolution (16-bit RGB)	Time	Throughput
GeForce RTX 4090	9344 × 7000 (65 MPix)	6.3 ms	10 GPix/s
Jetson AGX Orin 32GB	4128 × 3008 (12 MPix)	5.4 ms	2.3 GPix/s

Timing includes GPU computation only, with all data resident in GPU memory (RTX 4090: Windows 10, CUDA‑12.6; Jetson: JetPack 6.2.1). These numbers matter most for workloads that process enormous volumes of images or video — exactly what ALTMapper was designed for.

Applications

Digital photography and cinema, display devices, computer vision, medical imaging, industrial cameras, broadcasting and streaming, and street‑view, mobile mapping and aerial imaging.