You can't color calibrate deep space photos

https://news.ycombinator.com/rss Hits: 16
Summary

Why you can't color calibrate (deep) space photos: Jul 21, 2025 (Photography) Color is the most disputed part of astrophotography. It seems like no two images have the same colors, and the internet is full of disputes over what colors are correct or natural. Human vision and perception are very complex. This description is hugely simplified as to fit in a blog post. If you want details, consult a textbook. Roughly speaking, our eyes have four types of photoreceptors. The most abundant, rods, provide a black-and-white view of the world: very detailed but with no color. The others are cone cells, each type responds to a particular part of the visible spectrum: Left: All that you will ever see.Right: Infinite abyss of the IR. The peaks of these wavelength ranges roughly correspond to the colors red, green, and blue. Because there’s a lot of overlap between the red and green cones, our brain subtracts some green from red, yielding this spectral response: Yes, this results in red having negative sensitivity @500 nm. The perceived color depends on the ratio between the signals from each type of cone, which corresponds to the ratio of red, green and blue light entering your eye. The pixels in a camera produce a signal based on the brightness of incoming light, producing a monochrome image. To make a color camera, we overlay a grid of alternating RGB filters over the pixel grid: Typical Bayer filter matrix The camera’s computer converts the brightness ratio between neighboring pixels into color. The resulting color accuracy depends on how close the camera’s filters are to the cones in the human eye. Here’s what the spectral response of a typical camera sensor look like: ohno The most striking difference is what happens from 800-1000 nm: To our eyes, these wavelengths are mostly invisible. However, the camera sees them quite well, and because the organic dyes used as filters don’t work in the infrared, they trigger all the color channels equally. After white balancing (dimmi...

First seen: 2025-07-23 00:52

Last seen: 2025-07-23 15:55