Originally posted by UncleVanya We appear to be playing STUMP the CHUMP. (Cartalk reference)
I zoomed in and can see that weird patterning on some but not all of the pink. I really wonder what caused that pattern.
Here's my guess:
The black plastic of the mirror box is not perfectly black in the near-IR (which can happen if the "black" is made from tiny particles which would tend to absorb shorter-wavelength light better than longer wavelengths). Thus, the reflection of the sun is blindingly bright in the near-IR.
The IR filter of the camera isn't perfect so a tiny % of the near-IR gets to the sensor. (Remember that the sun is over 100,000X as bright as the scene so if even 0.001% gets through, it's going to be very bright.)
The different colors of the Bayer CFA also let in some near-IR which creates a very strange signal.
The demosaicer is seeing some impossible numbers in the different color channels which it ends up interpreting as bright magenta but the odd-even row-column patterns of the RGB filter pattern end up creating odd-even fluctuations in demosaiced values.