Why are my darks suddenly different?

[BrendanC]

Hi all,

I decided to extend my darks library while the clouds are still rolling in.

I know, I know - I shouldn't be using a darks library with a DSLR but it seems to work for me.

Anyway, my past MO has been to put the camera in the fridge and build up the library for 60s, 120s, 180s and 240s exposures, all at ISO800. Then I group them according to temperature and build master darks from 50 subs for each temperature.

Here's an example of one at 29C that I took a while ago, stretched in APT to bring out the details:

I've used these darks for a while and they produce good images (or at least I think so).

The difficulty using this method is getting the higher temperatures (dur, cos it's in the fridge, right?), which is where the dark files are most handy. So I decided to put the camera in a sealed box outside, with the lens cap on. I got this (ignore the square top left, that's just an APT thing):

It's the exact same exposure time - 240s - and EXIF again reports 29C.

Looks like light leakage maybe?

Realising the LED display was still on, I slapped my forehead, switched it off, thought that would be it. Nope.  So I draped a thick towel over the box. Nope, still there. So I wrapped the camera in a big hat, and then in the towel. Absolutely no way light's getting in there.

Still the same.

And now I just put it back in the fridge, exactly as per the, er, darker dark, and I'm getting the same result - a sea of purple noise. Nice title for a Prince song maybe, not so great for calibration.

So, I'm thinking this must be thermal noise, despite the EXIF reporting the same temperature. Is this the classic thing of the EXIF reporting the Digic 3 temperature, not the sensor temperature? In which case, am I stuffed?

Thanks, Brendan

Edited June 30, 2021 by BrendanC

[Astro Noodles]

I can't help asking. When are you expecting it to be 29C at night in the UK?

[BrendanC]

I'm not - it's just that the dark came through at that temperature. I was hoping it would go down during the night, but I was testing it now.

Also, given that my camera typically is around 10 degrees above ambient when shooting, I can easily imagine an evening of 19C. I do have subs at 29C and above, believe it or not.

The problem still remains though. Why is one taken at 29C markedly different from another?

Edited June 30, 2021 by BrendanC

[BrendanC]

Well, after leaving it in the fridge for a bit longer, I started getting the same as the darker darks.

So it could be something thermal.

If so, it would be an object lesson in how EXIF temperatures differ from the thermal state of the sensor.

Hmmmm.

Edited June 30, 2021 by BrendanC

[BrendanC]

Just a little bump of this one cos I still don't really know what's going on.

I have a theory that it's evidence of how the Digic 3 processor, from which the EXIF temperature is derived, is demonstrably a different temperature from the sensor in certain conditions.

It would be great if anyone could confirm this one, ideally by having experienced this themselves. I know that in theory this could account for it, but whereas in theory, theory and practice should agree, in practice, they don't!

[Starflyer]

The LCD screen generates a lot of heat, the temperature sensor used in EXIF isn't on the imaging chip, but close to it.  It's entirely likely that the heat from the screen running heated the sensor enough to change the darks.

I ran a 450D in a cooler box for a few years and always used temp matched darks, the banding typically seen in canon DSLRs completely disappears at an EXIF temp of 10C.

[BrendanC]

Thanks - that's why I was asking this, because I'm aware that the EXIF isn't from the sensor, it's from the processor.

I've discovered that just leaving it in the fridge for a few minutes gets rid of the issue. So, I guess it's as I thought: an object lesson in the difference between sensor temps and processor temps. I'm just glad I spotted this before ruining my darks library!