Apple just refreshed its foundational image pipeline. The change arrives quietly in iOS 27, macOS 27 and iPadOS 27. Yet for photographers who shoot RAW files from mirrorless cameras it marks a noticeable step forward.
The update centers on a new version of the system-level processing engine called RAW 9. 9to5Mac first detailed the improvements on July 6, showing side-by-side comparisons that highlight gains in sharpness, color accuracy and noise reduction. The engine runs on-device. It taps the Apple Neural Engine. And it works on older files too, letting users revisit archives without new captures.
RAW files preserve sensor data before heavy compression. They give editors control over exposure, white balance and tone. Apple has maintained its own pipeline for nearly 800 camera models through Core Image for years. Eight prior revisions refined demosaicing, denoising and color handling. This ninth version stands apart. Apple calls it its biggest update yet.
David Hayward, a Core Image engineer at Apple, explained the architecture during a WWDC 2026 developer session. “[RAW 9] dramatically improves the rendering of RAW files,” he said. “It is built atop a tiled CoreML model, that combines demosaic with denoise for best quality. And the model is run on device using the Apple Neural Engine cores, for optimal performance.” The tiled approach processes images in sections. That design choice preserves memory while delivering higher fidelity.
Three demo images made the case. First, a low-noise shot from a Sony Alpha 7 II showed a vintage dial indicator. Under the prior engine the image looked solid. RAW 9 made it sharper. Fine text became easier to read. Clarity rose without added artifacts.
Next came a high-noise challenge. A Canon 5D Mark III file captured at ISO 51,200 produced a 10x crop of a box of crayons. The raw sensor data contained heavy luma and chroma noise. Colors blurred together. The older algorithm recovered some hues acceptably. RAW 9 went further. It defined each crayon distinctly. Specular highlights on their wrappers appeared. The difference looked obvious at a glance.
The final example tested a tricky sensor layout. A Fujifilm X-T5 image at ISO 12,800 featured embroidery yarn. Non-standard color filter arrays often create demosaicing headaches. Previous processing left color artifacts and soft detail. The new engine reduced those flaws. Yarn texture emerged more clearly. Small printed text on labels gained legibility. AppleInsider covered the same session and noted this represents the largest change since RAW 8 launched in 2017.
Developers gain direct access. Apps such as Halide for capture or Pixelmator Pro for editing can opt into the new pipeline. The session outlined steps to enable it, tune performance during editing and streamline exports. Adoption will take time. But once integrated the benefits flow to professional workflows on iPhone, iPad and Mac alike.
This processing advance sits alongside broader photo tools in the same release. Earlier reporting from Bloomberg in April flagged Apple’s plan for an extensive photo-editing refresh powered by Apple Intelligence. Those features focus on extending image borders, reframing composition after capture and cleaning up distractions. They target everyday users more than pure RAW shooters.
Apple’s official announcement reinforced the Photos focus. “In Photos, Spatial Reframing enables users to improve the composition of a photo after it’s been taken,” the company stated in its June newsroom post. Craig Federighi, Apple’s senior vice president of Software Engineering, added context on the wider platform goals. “We’re delivering the next generation of Apple Intelligence across our platforms; introducing Siri AI, a profoundly more intelligent, knowledgeable, and capable Siri; expanding child safety features with intuitive new tools for families; and making our software platforms faster, more reliable, and more delightful than ever before.” Testing for these capabilities used an iPhone 15 running prerelease iOS 27 and a library of 50,000 assets.
Practical tests of the AI editing tools reveal trade-offs. One independent review at Volatile Inputs found that the Clean Up tool no longer works on ProRAW files in iOS 27, a shift from the prior version that may stem from file-size limits when routing work to Private Cloud Compute. High Quality mode takes 30 to 35 seconds on complex edits. Fast mode finishes in two to five seconds but sacrifices some refinement. The author also noted multi-area selection now works smoothly and that edits embed metadata automatically.
But the RAW 9 changes operate at a lower level. They enhance the initial decode and rendering before any AI magic begins. Photographers who import files from Canon, Sony, Fujifilm or Nikon bodies stand to benefit first. The engine handles demanding cases better. High ISO noise drops without smearing fine detail. Colors lock in more accurately. Shadows retain structure.
Professional reaction has been measured but positive. Forums and early beta users point to improved exports in third-party apps. Some note that the gains appear most striking on images that previously challenged the pipeline. A noisy night shot or a detailed macro suddenly looks cleaner. And because the model runs locally, privacy stays intact. No cloud upload required for the core processing.
Compatibility spans hundreds of camera models. Apple maintains an updated list of supported formats on its support site. The list grows with each hardware generation. RAW 9 builds on that foundation rather than replacing it. Existing apps that don’t explicitly adopt the new version will continue with the older pipeline for now.
Timing matters. iOS 27 betas continue through summer. Public release is expected in September alongside new iPhones. Developers have had access since WWDC in June. Those building camera or editing tools can already experiment. The full impact will surface once updated apps reach the store.
Apple has improved its image stack steadily for more than a decade. This release feels different. It marries classical signal processing with a modern machine-learning model. The combination yields results that earlier incremental tweaks could not match. For industry professionals who live in Lightroom or Capture One but also edit on mobile devices the change could shift daily habits.
Look closer at those crayon colors or the yarn threads. The difference isn’t marketing gloss. It appears in the pixels. And it arrives without forcing users into the cloud or draining battery. On-device efficiency remains a priority. The Neural Engine handles the heavy math. Users see the output. The mechanism stays invisible.
Future updates may extend these gains. For now RAW 9 stands as a focused, technical improvement that professional shooters have waited years to see. It won’t dominate headlines like flashy AI features. Yet for those who scrutinize every pixel it delivers exactly what matters most. Better data. Cleaner results. More faithful reproduction of the original scene.
WebProNews is an iEntry Publication