Updated

Updated · PetaPixel · Jun 10

Apple Unveils RAW Processing 9 for 784 Cameras, Boosting Sharpness and Noise Reduction

Updated

Updated · PetaPixel · Jun 10

Apple Unveils RAW Processing 9 for 784 Cameras, Boosting Sharpness and Noise Reduction

Q: Apple's new RAW engine is power-hungry. Will older iPhones and Macs be able to keep up with its AI demands?

Not fully. Apple’s new RAW 9 engine is built around on-device AI, and the best results depend on the Neural Engine in Apple’s A-series and M-series chips. That means newer iPhones and Apple Silicon Macs should handle it well, while older hardware will be slower and, in some cases, may not get the full experience. Apple says the new pipeline uses a CoreML model that performs demosaicing and denoising together, a much heavier workload than the older multi-step RAW engine. It is explicitly designed to run on the Neural Cores of Apple chips. That is a strong signal that Intel Macs and very old iPhones without a dedicated Neural Engine are at a disadvantage. The performance gap is large. Apple’s Neural Engine has grown from about 0.6 TOPS in the A11 Bionic to roughly 35-plus TOPS in recent iPhone chips, and from 11 TOPS in the M1 to around 38 TOPS in the M4. So an M1 Mac or older A-series iPhone may still run RAW 9 if supported by the OS and app, but processing will likely be slower, with more heat and battery drain on phones. Apple appears aware of that trade-off. RAW 9 is not the default for third-party apps; developers must opt in because of its higher processing demands. Apple also provides tuning tools such as lower-resolution previews and memory controls to keep editing responsive on capable devices. The likely outcome is that newer Apple Silicon devices will benefit most, while older iPhones and Intel-era Macs will struggle or fall back to older RAW processing paths.

3 articles · Updated · PetaPixel · Jun 10

Version 9 of Apple’s Core Image RAW processing brings major image-quality gains to RAW and ProRAW workflows across iPhone, iPad, Mac and Vision Pro, with rollout tied to iOS 27, iPadOS 27, macOS 27 and visionOS 27.
A new CoreML pipeline now performs demosaicing and denoising in a single step, which Apple says improves sharpness, color and noise handling versus the older multi-stage approach.
The upgrade is the first major Core Image RAW revision since RAW 8 in 2017 and runs on-device using Neural Cores in Apple A- and M-series chips, though it demands more computing resources from apps.
Photos, Pixelmator Pro and other third-party apps that use Apple’s RAW APIs stand to benefit, while Apple says its engine currently supports 784 cameras and is still expanding format coverage.