CyberTimon/RapidRAW v1.5.2

RapidRAW v1.5.2 - AI Denoising, Image Analytics, Parametric Masks & More!

on GitHub

NIND Denoising

Previously, RapidRAW only had a traditional BM3D denoiser. It produced decent results, but it was never really competitive with the newer AI-based denoising methods found in other software.

So I'm excited to introduce a new denoising method based on nind-denoise. It can be applied directly to RAW images to naturally remove noise while preserving detail.

The results speak for themselves:

Original	BM3D	NIND

Due to dependency limitations, NIND denoise is not yet GPU accelerated, so it will take some time to process each image. In future versions, once things have matured, I plan to support batch denoising and move the processing over to the RapidRAW AI Connector, which supports GPU acceleration.
Credits for this denoising method go to @trougnouf's nind-denoise.

Image Analytics Panel

RapidRAW now features a switchable image analytics panel, giving you multiple ways to exactly visualize what's going on with your edit in real time.

Supported modes:

• Luma Waveform
• RGB Waveform
• RGB Parade
• Vectorscope
• RGB Histogram

Parametric Color & Luminance Masks

I've added two new mask types that let you quickly target specific color ranges or luminance zones in your image.

Single Point Subject Masks

This release ships with updated, more accurate, and higher resolution AI subject detection models based on SAM-2. On top of that, I've implemented single point masking - just click on your subject and it gets selected. No need to draw a bounding box around it first.

JPEG XL, WebP, and Additional Format Support

This update adds support for a bunch of new export formats, including JPEG XL, WebP, and CUBE LUTs. The LUT export is something I'm particularly happy about: You can screenshot a frame from your video editor, color grade it in RapidRAW, export the result as a LUT, and then apply that same color grade back in the video editor at full quality.

LRU Cache

RapidRAW now keeps recently viewed images in a cache, so switching between them is instant and seamless. This is especially useful when you're culling and need to quickly flip back and forth to check focus.

ROI Rendering

Previously, dragging sliders while zoomed in would lag and stutter because the entire image was being reprocessed on every change. Now RapidRAW only processes the region that's actually visible in the viewport, which gives a huge speedup when editing zoomed in.

• Improved tonemapper & shader, particularly how it behaves within masked regions
• Refined keybinds across the app for a smoother editing workflow
• Various UI polish and visual tweaks
• A ton of bug fixes under the hood - too many to list individually ;)