Matching Temperature of Lights and Darks

[bendiddley]

I hear people say that you have to use darks that are the same temperature as the lights but during the course of a session the temperature often changes so you end up with darks and lights that are different temperatures. What do you do about this and how do you match them? I'm using a DSLR.

[vlaiv]

Not much to be done when using sensor that doesn't have set point cooling. There are some things that you can do:

1. Create dark library with darks created at different temperatures - do one in unheated room to get about 15C, do one during the day outside in cold weather to get 10C, do one on cold night to get 5C (or 0C depending on weather) ... and afterwards select darks that are "closest" match to temperature at which your lights were shot and use dark optimization (you need bias files as well, and your camera needs to have good bias, but I think DSLRs have it).

2. Shoot darks at the end of each session - it will be close to temperature of lights (but not exact). Again you can use dark optimization to try to fix this.

3. Do some sort of cooling mod - like cold box or something (again dark optimization won't hurt if mod is not giving you 100% temp control).

4. Switch to cooled astro cam?

[bendiddley]

Great thanks, I'd wondered about doing a dark library. I thought they needed to be at every temperature though, but its good to hear they only need to be close, I'll check on the dark optimisation, is that a setting in DSS? Also do bias files need to be close in temperature as well? I definitely want to get a cooled CCD cam at some point in the future, but not sure when, need to do some saving I think. That would make it nice and easy to match the temperature.

[MarsG76]

This might be pushing against the grain, but why use darks? especially when using a DSLR for astro imaging... I never use darks and the results seem ok.... in actual fact way back when, when I did use darks, it gave no benefit, actually created "black holes" where the noise/overactive pixels/cosmic rays were.

In contrast when converting the Canon RAW files to TIFF, the Camera Raw plugin does a very good job are removing nothing but noise and for the single pixle meshy noise that is left over... running a depixel filter (Pixel size:2 pixels & threshdold: 30) removes the reminder of the noise.. now I did compare this to the original and it removes NONE of the actual data mistaking it for noise. This can be automated using "Batch" and Actions to do to all of your subs in one click.

Bias and flats is another matter... those do help.

[vlaiv]

13 hours ago, bendiddley said:

Great thanks, I'd wondered about doing a dark library. I thought they needed to be at every temperature though, but its good to hear they only need to be close, I'll check on the dark optimisation, is that a setting in DSS? Also do bias files need to be close in temperature as well? I definitely want to get a cooled CCD cam at some point in the future, but not sure when, need to do some saving I think. That would make it nice and easy to match the temperature.

Yes dark optimization is feature of DSS - there is a check box on darks parameters tab in settings dialog. There is also manual "coefficient" next to it that you can use, but I would not recommend that.

Dark current doubling temperature is around 6C (some cameras more, some cameras less than that, but in general it is around that value). This means that if you shoot at 10C and at 16C - darks at 16C should have roughly double ADU values compared to 10C one (with bias removed, just dark current values). Dark optimization uses clever algorithm to try to guess this factor between ligths and darks (you don't need to tell it anything about temperatures) when subtracting dark current values.

Bias should not depend on temperature since it does not contain any dark current signal - just read noise / offset level. In practice there might be slight difference between let's say -20C and 20C in bias file, but question is if it's going to even be detectable next to read noise. So you can take them at any temperature.

I don't have experience with DSLRs, but people tend to recommend using bias only for calibration and skipping darks. In principle this is wrong calibration, but you should give it a try because so many people claim it works well (just take some bias and some dark subs and try one and the other and see what you like the best). I'm never sure that "general rules" for sensors work with DSLRs because there is always possibility that DSLR manufacturer did some "internal optimization" of their own (like bad pixel maps or something with dark current - factory calibration of some sort).

[mvas]

With most DLSR's, they don't report he temperature of the Image Sensor.
The Canon temperature probe is embedded in the Canon Digic Processor Chip, not the Image Sensor.
Subtracting Darks that have a "higher temperature" than your lights, can cause black holes.