top of page

LMR: Line Memory Reduction

A very low latency perceptually lossless on-chip line/tile memory reduction codec delivering up to 60% line/tile memory reduction 
Works in RGB, RAW, YUV and FP16 domain processing
All CFAs supported
Pin Out Block Diagram.png
Cores can support one pixel/multi pixel per clock processing

Compression ratio is fully configurable: Lemur typically recommends 50% compression for Human consumption; 60% for Computer Vision Applications

Original

Lemured - Visual

8x zoom

Original

Lemured - CV

8x

Lemured CV

 Lemured Visual

8x

Core Sizes

Core size is dependent on type and bit depth of data to be encoded
One encoder is required per filter/scaler and one decoder per line

Screenshot 2022-08-27 at 15.39.18.png
bottom of page