Apple introduced RAW 9 with iOS 27, macOS 27, and iPadOS 27, calling it the biggest update yet to its system-level RAW image processing pipeline. The engine powers Core Image and supports nearly 800 camera models, handling demosaicing, denoising, white balance, exposure, and tone adjustments for third-party RAW files.

The new version is built atop a tiled Core ML model that combines demosaic and denoise operations in a single pass. It runs on the Apple Neural Engine cores for on-device performance, eliminating cloud dependency. At WWDC26, Core Image Engineer David Hayward demonstrated RAW 9 against RAW 8 using crops from a Sony Alpha 7 II, a Canon 5D Mark III at ISO 51,200, and a Fujifilm X-T5 at ISO 12,800 with its non-traditional sensor pattern.

In each case, RAW 9 produced sharper detail, cleaner color separation, and fewer artifacts. The Canon example showed visible specular highlights on crayons that RAW 8 muddied, while the Fujifilm sample resolved fine yarn texture and small text previously lost to color fringing.

For developers, Apple detailed how to enable RAW 9 and optimize editing and export workflows. Consumers benefit automatically when apps adopt the updated Core Image pipeline, meaning older RAW libraries can be reprocessed with significantly better noise handling and detail retention — no new hardware required.