1 hour ago · Tech · 0 comments

I’ve been curious about what rendering intents do geometrically. ICC documentation is qualitative. Looking at ordinary photographs doesn’t completely tell the story, because image content dominates the visual result. I wanted to see the transforms themselves. I built a MATLAB framework around a regular RGB cube image‚ in this case Bruce Lindblom’s RGB16Million.tif ‚ with a ProPhoto RGB profile applied. The basic workflow: Start with a regular RGB cube. Convert it through a CMYK profile using a selected rendering intent and color-management engine. Convert the result back to ProPhoto RGB. Compare the transformed cube geometrically. With the Lindblom image source sampling is regular and known. That makes it possible to visualize the rendering-intent transform as a deformation of color space itself. The first useful plots were 3D in Lab slices selected by source L*. Those revealed that perceptual rendering is not merely clipping or radial compression. The transforms are anisotropic,…

No comments yet. Log in to reply on the Fediverse. Comments will appear here.