Built In LCD Calibration

Started by chrisfrancia, April 16, 2018, 03:21:19 AM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

chrisfrancia

Is there any way we could adjust the built in LCD's calibration. Mine seems a bit off, not much but off.

(just in case theres any confusion, I'm wondering if theres a way to adjust the RGB of the LCD or something)

I suppose its also worth asking if it would be possible/easier to have the cam load an .icm file on boot, then  a color calibrator would be used and the .icm transfered onto the card?
[5D3 1.2.3 ML] [6D 1.1.6 ML]

a1ex

All the adjustments performed by the LCD controller are already available in ML menu, including RGB adjustments; one "just" has to document what they do, mathematically (e.g. extract the response curves). There is test suggested in post #20 that could be useful for that, although I'll add a few more requirements to make things easier:

- the test image should have a color chart somewhere, but should not be limited to it (other things should be in there, see below)
- the test image (displayed by the camera) should have both dark and slightly overexposed areas in YUV space
- a RAW (CR2 or DNG) version of the test scene should be available (set the camera in PLAY mode, reviewing the static image, while recording its display)
- the captured image (video recording of the camera screen) should be a little out of focus (just enough to avoid seeing the LCD pixel grid)
- the frames containing the same (static) scene should be averaged into a DNG (to have both a smaller size and a better SNR)
- of course, I'm only interested in registers and values useful for color adjustments (not interested in DNG frames with display screwed up)
- no light changes (the test is probably best done under artificial light, or maybe even in total darkness)



The end result will be similar to this clip, but I need it to be raw video in order to work with linear measurements of the LCD brightness. Ideally, static (averaged) DNG frames matched with register settings, with some consistent naming pattern (e.g. the message displayed on the screen during the test should be transferred into the file name).

As you can see, the task is a bit tedious and time-consuming, but there's no programming required.

Once we have a mathematical description of what these registers do, it should be easy to extract the curves from an ICM profile and tweak the registers to get a close match.

The "best" part: the registers are model-specific, so the test should be repeated for every single camera where this feature is desired.

chrisfrancia

Thanks for the reply! I don't currently own a color chart so this will have to be a future project for me but, just to clarify what would be needed

1. Take test image with chip chart and stuff
2. Record Raw video of the back of the screen while the test image is being shown (very slightly out of focus)
3. convert to DNG
4. send to you

do you care about resolution/bit of the raw video or how the DNG is made from it?

(sounds quite annoying for you as well :/ )
[5D3 1.2.3 ML] [6D 1.1.6 ML]