Help with Spotty Darks

[iceman (Mike)]

Is this normal or am I doing something wrong?

I took some lights and darks last night with the (unmodded) 350D on the ED80/EQ6. It was reasonably mild, about 18deg-19deg ambient.

The lights look fine (although washed out from light pollution and low altitude), but the darks are incredibly spotty - so when they are subtracted from the lights, I end up with dark spots everywhere.

They're best seen in the full-size versions, so I uploaded a sample image to show you. They're just converted to 8-bit and saved as medium-quality jpegs (around 500kb each).

1. Raw light image (300s, ISO800)
2. Raw dark image (300s, ISO800)
3. Stack of 5 x median combined darks (all at 300s, ISO800) using ImagesPlus
4. Calibrated light from the stacked dark. You can see the black spots.

Ignore the jpeg compression artifacts.

I've attached a 800px resized version of #4, but as I said it's much more obvious in the full size version.

So, what am I doing wrong?

[h0ughy (David)]

had the temperature changed from when you started to image until you took the darks?

[iceman (Mike)]

Not significantly.

[Dennis]

I once saw these on images calibrated with Dark Frames from our SBIG ST7 and figured that if hot pixels are white, then cold pixels must be dark as they would be less sensitive to the incoming photons. I was able to remove them (automatically) using the “Remove Cold Pixels” function in CCDSoft.

However, I was very surprised to find these in the (cooled) Dark Frame as the ST7 is temperature regulated. The repair operation reported back that it had repaired a number that was well over 100 from memory.

Looking at your example, you seem to have 100’s – I wonder if this indicates something going wrong with the calibration routine?

Does Images Plus have a cold pixel repair tool?

Incidentally, your 300 second Darks look significantly less noisy that those from the ST7, even when it is cooled to 25 deg C below ambient!

Cheers

Dennis

[avandonk]

Mike the mere act of taking a long exposure will heat up the sensor from its idle temperature. I always use ICNR (in camera noise reduction) and the first light frame will have more and larger dark spots than subsequent light frames. This is due to the sensor heating up slowly during the first exposure and by the time the dark is taken the dark will overcompensate and produce dark spots. It is almost impossible to have the sensor at the same temperature for both darks and lights.

I have never had any joy out of taking darks before or after a series of exposures and correcting in the computer later. I have tried every possible method even working with all fits files. To this day I have found NOTHING and I mean NOTHING beats ICNR. Canon do something very cunning at the individual sensor pixel level. ICNR also eliminates amp glow.

It doubles the time needed for data collection but at least the data is as free of noise as is possible.

There is nothing wrong with your cameras sensor. These spots are more noticable as summer weather arrives.

Here is a little gif 300k

http://members.optusnet.com.au/~fmlee/mk4.gif


Bert

[Dennis]

That’s very useful information Bert – thanks! You’ve saved me a lot of potential heartache and wasted experimentation time, as I’ve yet to use DSLR’s for long exposure DSO work and subsequent image calibration with separate Darks.

I’ll be sticking with ICNR!

Cheers

Dennis

[iceman (Mike)]

Thanks Bert - I started out using ICNR a few months ago, but was convinced by some that it's a waste of time and that you could be getting twice the number of lights if you have a good master dark library.

I'll have to give it another try. But surely not every DSLR user gets these black spots in their calibrated images?

I'm not sure what the animated gif is showing?

[avandonk]

It is alternating between a brightened crop of your dark and the exact same area from your 'corrected' light. Can you see the correlation?

Bert

[rogerg (Roger)]

Mike,

I find it the norm that, from my simplistic terminology, the dark frames for my DSLR will "subtract too much". I do my dark frame subtraction manually in photoshop, and find if there is a problem, it's easy to fade the subtraction (only subtract 80% for example), and this gets rid of them for me. There's always a balance where there's no significant number of dark specs left.

I'd say about 65% of the time I would need to fade the subtraction of dark frame, the rest of the time it's good enough.

I put it down to the difference in time between taking the images and taking the dark frames. I usually take a dark frame about once every hour.

With the ST7 there's often a few specs (say, 10) that I clone out once I've finished processing the image. Problem is no where near as bad with the ST7 presumably because of it's lower noise level and it's stable temperature.

It's one of those things.... 99.99% of the time people only ever show their good processed images, so unless you're "in the game" doing the imaging, you don't know just how the raw images and the rest of the processed images that the people don't show, turn out

Roger.

[Aster]

I get the same effect as Mike, taking darks with my DSI Pro. Anything longer then 30sec is just about useless.

But still classing myself as a beginner I have no idea how to correct that, although Rogers info of subtracting in photoshop looks like something I may try.

[turbo_pascale (Rob)]

Originally Posted by iceman

Is this normal or am I doing something wrong?

My experience so far (especially on really dim stuff like this) has not been very good with my unmodified camera. I've always put it down to huge light pollution and low signal causing my poor results. I don't seem to have the same kinds of issues on brighter objects though.

Mike, although this should not make any difference (in theory), because you have the darks all scaled correctly, you did not mention taking/subtracting BIAS frames. It is very unlikely to make much (if any) difference, but it is a very quick process to take them.

In terms of your darks, how did you physically block the light? (ie camera off the scope, body cover on, or did you just cover up the incoming light path on the OTA)? The reason for the question is to eliminate any possibility that there may have been some kind of partial light leak during the dark taking process.

Neither of these is likely to be the culprit, but it's a process of elimination!


Turbo

[sheeny (Al)]

When I had a play for a while taking some DSO images with the ToUcam, I found the same thing as Roger - the darks subtract too much.

I was stacking and subtracting darks in both Registax and K3 and saving them as TIF files. They looked fine in the software, but when I viewed the saved files later the background was full of psychedelic colours, which I guessed were the result of the dark subtraction producing a negative number. I worked around it by stacking multiple darks and then darkening them by about 80% (remarkable that we used about the same number Roger!) before using them as darks.

I didn't persevere long enough to conclusively prove my theory but the effect seemed to work - no psychedelic pixels, but maybe a little more noise...

Al.