Newbie question about darks

[Diane Miller]

I'm just starting on this journey, shooting with a Canon 7D Mk II.  During a typical run out here in the west, the temperature can drop by up to 10 or even 15 deg F.  But last night it was only 3 deg. so it can vary.

If I cover the camera very carefully (lens hood and black tape over eyepiece, than a dark blue blanket over it all) and shoot blacks before it is fully dark enough to image, I have them at a warmer temp than the run, and if I shoot after, it's colder.  I don't know how much ambient air affects sensor temp, though.

Am I better off doing a set of each and then deciding which to use, or not doing darks at all?

I've heard of making a master scalable dark (do I have the name right?), but haven't learned how yet.  Would that be better?  It sound a lot more convenient.

I understand the Canon 7D Mk II (and other Canon bodies) may be problematic for darks.

Thanks for any information!

[speach]

Hope you get a reply as I'd like to know too.

[dnault42]

I've had good success at taking darks at a higher temperature and then using the dark optimization option during calibration to scale them.  I process the darks normally otherwise.  For that to work well, you need to create a bias frame as well.  Then you can either calibrate you darks with the bias before stacking them to make the master dark or turn on the calibration option under the master dark section on the ImageCalibration form.

I believe the PI team is working on a new dark scaling method, but the current method works reasonably well for now.  I usually find that the scale darks do not correct everything so you have to relay on statistical rejection during stacking your light frames to remove the errant pixels or do a CosmeticCorrection pass after calibrating the light but before registration.

Regards,
David

[dnault42]

I have a tutorial on how to do a full manual integration of your data, including weighting the individual frames to target specific metrics, like FWHM.  You can find it here.  Kayron also has some really excellent tutorials about this.  You can see his forum post about that here.

Regards,
David

[CraigNZ]

I would recommend taking say 10 darks at different temperatures, call each of them a 'Set'.  For each set combine them to create an average, so one for each set.  Then use statistics to look at each average image and compare them.  If the mean and deviation are pretty much the same then I would not worry about temperature.

It is always best to bracket your light frames with dark frames.  In other words, say a set of 10 dark images before you take your light frames and 10 after.  Again average each set and then take the average of the two set averages which then represents somewhat the average dark frame during the time of your light frames.

Dark frames are also dependent on the exposure time.  A longer exposure will have more thermal noise (Dark frame) in it than a short exposure.  Again, for a given exposure time repeat the tests above and determine if you need to do specific dark frames given the exposure time, or as mentioned above, take a very long dark frame (usually 2x longer than your longest light frame) and use 'optimization' in PI when calibrating an image.

Each sensor and observing session is different so begin by simply taking dark frames in varying conditions and see how your camera does.  If the thermal noise is very low and the exposure times are short than a single dark frame will suffice.  But if you are doing precision photometry work or very critical 'pretty pictures' then you will need more sophisticated calibration and dark frames.

Start slowly and learn, but most of all have fun.  Take pictures and study each one, you will have plenty of time to master it all.  Going too technical can quickly become a downward spiral where we spend more time trying to squeeze that last bit of perfection out of a pixel than we do in simply enjoying the hobby.

I have seen many times that a uncalibrated image of a galaxy or nebula will bring the same 'wow' from family and friends as did several hours, days and weeks trying to get that picture to match the Hubble.

[pfile]

also one minor note is that a canon camera samples the temperature of the camera at the start of an exposure. so when the camera is 'cold' your first (few) image(s) will show some very low temperature in the exif header, which can be misleading. once the camera is running for a while the temperature will stabilize.

adding to the problem is that the temperature in the exif file is not necessarily the sensor temperature - it's just the temperature somewhere in the body. there is a correlation between the sensor temperature and the recorded temperature, but it may also take a while for the recorded temperature to be accurately reflective of the sensor temperature.

rob