Processing Moon Photos: The Colour Almost Everyone Leaves Out

Almost every processed Moon photo online is some shade of grey, cream or pale yellow. That is not what the Moon actually looks like in colour data - it is what happens when real, subtle colour gets discarded on the way to a sharp, high-contrast image, because sharpening and colour are usually optimised for separately and colour loses.

The Moon’s surface genuinely has colour. It is just an order of magnitude fainter than the brightness variation between crater floor and crater rim, so any processing that treats brightness and colour the same way stretches away the one while chasing the other.

Where the colour actually comes from

The lunar maria - the dark “seas” - are basalt, and what tints them depends on what is mixed into that basalt.

Iron oxide (FeO) shows up as an orange-brown tint. Titanium oxide (TiO2), alongside the iron, pushes the tint toward blue. Mare Tranquillitatis, higher in titanium than most other maria, reads distinctly bluer than its neighbours once the colour is actually processed rather than averaged away - the same basalt plain, different mineral mix, different colour, and it is real geology rather than a processing artefact.

None of this is visible by eye at the eyepiece, and it barely survives a normal stretch either. It needs to be pulled out deliberately, and it needs the one thing most lunar workflows do not do: treating detail and colour as two separate problems.

Why the normal workflow loses it

A typical Moon processing pass stacks the video, sharpens hard with wavelets to bring out craters and rilles, and applies whatever saturation adjustment is left over at the end. That order is the problem.

Wavelet sharpening operates on luminance-scale detail - it amplifies local contrast at chosen spatial scales, which is exactly right for crater rims and ray systems. Applied to a colour image without separating the channels first, it also amplifies noise in the colour channels, which is many times fainter and mostly noise to begin with at that point. Boost saturation afterwards and you are amplifying that noise along with the two or three percent of it that was real mineral tint.

The fix: split luminance from colour before you sharpen

Treat the Moon like an LRGB target, not a single greyscale-plus-tint image, even from a one-shot colour camera.

Luminance carries the detail. Take it from the green channel, or from a proper luminance derivation if your capture allows one - this is what gets the aggressive per-scale wavelet sharpening, pushed as far as the data supports, because noise here is at its lowest and detail at its highest.

Chrominance carries the colour, and only the colour. Keep the colour channels at native resolution rather than sharpening them, since there is no fine colour detail to sharpen - the tint varies smoothly across a mare, not crater by crater - and boost saturation here specifically, separately from anything done to luminance.

Recombine, and the craters stay sharp while the maria pick up colour that a single combined-channel sharpen would have scrubbed out as noise before saturation ever got a chance to show it.

This is the same principle Akastroid applies automatically for deep-sky photometric colour calibration - measure and set colour from the data itself, on its own track, rather than as an afterthought to whatever the detail pass left behind - just applied to a much smaller, much subtler colour signal.

What this needs to actually work

A clean base stack. Colour this subtle does not survive being pulled out of a noisy or poorly-aligned stack. Grade for sharpness and keep a smaller, better fraction of frames rather than a larger, noisier one - the usual lucky-imaging trade-off, and it matters more here than for a monochrome-only result.

Accurate white balance. Every mineral tint is measured relative to the frame’s overall colour balance. A white balance that is off by even a little pushes the whole result warm or cool and the mineral difference between maria - which is what you are actually trying to show - gets lost in a colour cast that has nothing to do with geology.

Restraint on saturation. The colour is real but small. Pushed too far it reads as an obvious processing choice rather than a discovery, and a Moon photo that looks like it was run through a Instagram filter undermines the thing that made this worth doing - showing something true that a casual look does not reveal.

The short version

• The Moon’s maria have real colour: orange-brown from iron oxide, blue from titanium oxide mixed in with it.
• A combined sharpen-then-saturate workflow amplifies colour noise faster than it reveals real colour, because the tint is far fainter than the surface detail.
• Split luminance (sharpen hard) from chrominance (leave at native resolution, boost saturation separately), then recombine.
• Needs a clean stack and accurate white balance - fix those first, or there is no real colour left to pull out.