Jetraw Core

Jetraw Core is a high-speed, high-ratio RAW image compression solution developed by Dotphoton, designed for OEMs and system manufacturers who need to integrate efficient compression directly into their acquisition or processing pipelines. It is available as an FPGA IP Core or as a software library, making it suitable for camera manufacturers, automotive Tier 1 suppliers, aerospace and satellite system builders, and scientific instrument makers. Jetraw Core delivers a typical compression ratio of 6:1 (ranging from 5:1 to 10:1) while maintaining strict RAW image quality guarantees, enabling robust AI and ML performance without sacrificing data integrity.

At its core, Jetraw Core works by embedding a custom noise model based on the sensor's physical properties, turning raw image data into metrologically validated measurements. It compresses noise lossily and the signal losslessly, achieving market-leading compression ratios that match the file sizes of lossy formats while preserving the reliability of lossless ones. The result is output that is indistinguishable from RAW, with no artifacts, no bias, no artificial correlations, and no low-pass filtering — delivering a tightly controlled image quality equivalent to a 1.2 dB SNR increase (ISO 100 to ISO 130).

Key Benefits for Optical Systems

• Up to 6× faster data capture and transmission — enabling higher throughput even over lower-bandwidth connections.
• Higher sensor resolution and dynamic range without increasing bandwidth requirements.
• Extended battery life through reduced energy consumption on-device.
• Efficient IT infrastructure — maximises processing and storage capacities while reducing data storage expenses and boosting data access speed.
• Reliable AI/ML data — preserves critical image information with pre-processing flexibility and artifact-free output, improving the robustness of AI and ML models.

FPGA IP Core Specifications

• Up to 6.4 Gpx/s throughput with 32-pixel parallel processing.
• Low latency of approximately 60 clock cycles (~1 line; ~2 lines for Bayer sensors).
• 1 to 32 pixels per clock cycle; up to 200 MHz clock frequency.
• Parallel computing of multiple pixels for higher compression throughput.
• AXI4-stream data interface with backpressure support.
• LUT footprint from 3,968 (single-core compressor) to 70,790 (32 pixels per clock).
• Supports Xilinx FPGAs; Intel available on request.
• Supports 16-bit images with configurable image dimensions.
• Accepts RAW image buffer or common file formats.

Software Library Specifications

• 6.2 GB/s on Intel Core i9-14900K (Windows).
• 200 MB/s per core with multi-threading support.
• Callable from C, C++, C#, Java, and Python.
• Available as a software library or codec.
• Supports Intel, AMD, and ARM CPUs.
• Compatible with Linux, Windows, and macOS.

Camera and Sensor Compatibility

• Supports CMOS, sCMOS, and CCD cameras.
• Compatible with Mono, Bayer, and other CFA image sensors.
• Handles 16-bit images with configurable image dimensions.
• Interoperable with RAW image buffers and common file formats.

Target Industries and Use Cases

• Automotive — compression solutions for Tier 1 suppliers and automotive OEMs.
• Aerospace — compression for satellite and geospatial system manufacturers.
• System manufacturers — compression for camera and scientific instrument manufacturers.
• Life Sciences — enabling data-centric AI workflows with metrologically validated image data.

Jetraw Core is developed by Dotphoton AG and can be integrated into existing acquisition pipelines via FPGA IP Core or software library. Dotphoton offers a discovery workshop and quantified benefits analysis for prospective OEM customers to evaluate the solution against their specific optical system requirements.