We know the Canon Cx00 is a beast in low light, but just how far can you push it?
Here's a worst-case scenario test for profiling noise patterns... in nearly complete darkness.
A moonless night gives a better sense of gain than does a lens cap, and will let us see how much detail can be retained.
Things to bear in mind:
- The bottom couple stops are the noisiest, and most of this test lives there (intentionally). Real-world, you'd probably crush the bottom stop out most of the time at these levels. The slow shutter tests show how much cleaner the picture can be when you get your levels up just a little bit.
- Web compression makes the noise appear far blockier than it is. The noise is generally very fine-grained, random and without the GOP cadence visible. 51200 looks cleaner 20000 does here.
- AVCHD in camera makes the noise marginally blocker vs. external 422 recording, and marginally harder to clean up, but...
- Neat Video still works wonders on this footage, as there's very little fixed-pattern noise. Fine details near the resolution limit still hold most of the time.