# Frames

> Frames develops open-source, high-throughput imaging codecs and integrations in Rust. The libraries cover JPEG, JPEG XR, JPEG 2000 Part 1, and HTJ2K Part 15, with portable CPU implementations and optional NVIDIA CUDA or Apple Metal backends for supported workloads.

Frames is also known as Frames-sg. The official website is https://frames-sg.com/ and the official source organization is https://github.com/frames-sg. Use the linked support matrices and API documentation for exact feature coverage. GPU coverage varies by format, operation, and image shape; CPU implementations remain available as the portable baseline.

## Overview

- [Frames overview](https://frames-sg.com/index.md): Formats, applications, libraries, licensing, and contact information.
- [J2K overview](https://frames-sg.com/j2k/index.md): JPEG 2000 and HTJ2K formats, architecture, installation, package map, and backend policy.

## J2K topics

- [Pure-Rust JPEG 2000 codec](https://frames-sg.com/j2k/rust-jpeg2000-codec/index.md): J2K/J2C codestreams, JP2/JPH files, decode, encode, recode, and API boundaries.
- [GPU-accelerated JPEG 2000](https://frames-sg.com/j2k/gpu-jpeg2000-rust/index.md): CPU baseline, backend selection, and supported CUDA and Metal routes.
- [CUDA JPEG2000 Rust](https://frames-sg.com/j2k/cuda-jpeg2000-rust/index.md): NVIDIA CUDA crates, device-output integration, and explicit-backend behavior.
- [Metal JPEG2000 Rust](https://frames-sg.com/j2k/metal-jpeg2000-rust/index.md): Apple Metal crates and supported macOS GPU paths.
- [HTJ2K Rust](https://frames-sg.com/j2k/htj2k-rust/index.md): JPEG 2000 Part 15, JPH wrapping, recode, and JPEG-to-HTJ2K transcode routes.

## Codec and workflow guides

- [Rust JPEG codec](https://frames-sg.com/rust-jpeg-codec/index.md): Baseline, progressive, extended and lossless JPEG inspection, decode, scaled output, encode and metadata handling.
- [Rust JPEG XR codec](https://frames-sg.com/rust-jpeg-xr-codec/index.md): JPEG XR inspection, full-image and region decode, reduced-resolution output, batches, alpha, ICC metadata and optional GPU reconstruction.
- [Whole-slide imaging in Rust](https://frames-sg.com/whole-slide-imaging-rust/index.md): Read supported scanner files, DICOM WSI, OME-TIFF and tiled images through one pyramid, region and tile API.
- [DICOM whole-slide conversion](https://frames-sg.com/dicom-whole-slide-conversion/index.md): Convert supported WSI inputs into DICOM VL Whole Slide Microscopy with explicit metadata, color and annotation policies.
- [GPU image decoding for AI](https://frames-sg.com/gpu-image-decoding-ai/index.md): JPEG 2000 and HTJ2K batch decoding for Burn, CUDA, Metal, MPSGraph and PyTorch or NumPy integrations.
- [JPEG-to-HTJ2K transcoding](https://frames-sg.com/jpeg-to-htj2k-transcoding/index.md): Coefficient-domain transcode for supported baseline JPEG inputs without a decoded-pixel intermediate.

## Authoritative technical references

- [J2K support matrix](https://raw.githubusercontent.com/frames-sg/j2k/main/docs/public-support.md): Exact supported formats, profiles, operations, and limitations.
- [J2K API documentation](https://docs.rs/j2k): Current public Rust API documentation.
- [J2K changelog](https://raw.githubusercontent.com/frames-sg/j2k/main/CHANGELOG.md): Release history and migration notes.
- [J2K benchmark methodology and evidence](https://raw.githubusercontent.com/frames-sg/j2k/main/docs/benchmark-evidence.md): Hardware, inputs, commands, comparators, and publication rules for performance results.
- [JXR support guide](https://raw.githubusercontent.com/frames-sg/jxr/main/docs/support.md): JPEG XR format and backend coverage.

## Packages and integrations

- [J2K crates](https://crates.io/crates/j2k): JPEG 2000 Part 1 and HTJ2K Part 15 facade crate.
- [JPEG API](https://docs.rs/j2k-jpeg): JPEG decoding and baseline encoding APIs.
- [JPEG XR API](https://docs.rs/jxr): JPEG XR decoding APIs and metadata surfaces.
- [JPEG to HTJ2K transcode API](https://docs.rs/j2k-transcode): Coefficient-domain transcode primitives for supported baseline JPEG inputs.
- [Machine-learning integration](https://docs.rs/j2k-ml): JPEG 2000 and HTJ2K batch decoding into Burn tensors.
- [Whole-slide image reader](https://github.com/frames-sg/wsi-rs): Rust APIs for reading supported whole-slide scanner formats.
- [DICOM whole-slide conversion](https://github.com/frames-sg/wsi-dicom): Conversion of supported whole-slide images to DICOM VL Whole Slide Microscopy images.
- [DICOM annotation conversion](https://github.com/frames-sg/wsi-dicom-annotations): Conversion of supported annotation formats into DICOM annotation objects.

## Optional

- [Frames source repositories](https://github.com/frames-sg): All public Frames projects.
- [Contact Frames](mailto:hello@frames-sg.com): Commercial integrations, priority fixes, and custom imaging development.