Dark frames are not correctly subtracted

[Bobinius]

Hi everybody,

I am having this problem which seems to be reccurent during different sessions. I am using an Altair Hypercam 294C and Altair capture. After the integration using Batchpreprocessing, the image still has the sensor sideglow present and it seems that the dark frames did not subtract this signal completely.

Any idea what could be the cause? Pixinsight can correctly determine the duration of the darkframe (I capture fits). The glow goes away when I uncheck "optimize darks" but the image seems noisier (probably because the flats are no longer correctly calibrated).

Thanx

B

[pfile]

well when you have sensor artifacts like amp glows or if you are using a CCD with RBI preflash turned on, you can't use dark optimization. there's no way around that.

i don't use BPP regularly so someone else would have to chime in, but i think if you added some flat darks that BPP would try to use those to calibrate the flats. i think BPP does try to understand what the duration of the various darks loaded are and intelligently apply them. but not 100% sure.

rob

[Bobinius]

Thanx. The thing is that it doesn't seem to be systematic. I tried to do it manually and it stuck. One way around it is to remove optimize darks (is dark optimisation done only for the flats? by definition darks have the same length as the lights).

As you can see in the image the glow (apparently it is called Starburst and it is not ampglow), was partially subtracted but not completely.

I was wondering if the flat settings could influence this. I am trying to take them using the same gain I used for the lights and darks, in colour, with the B/G bell around the middle (the flat panel I use is blueish).

I don't know how to take dark flats.

[Bobinius]

I just saw that there are a few threads on this topic with the same problem. I'll do some experimenting and tell you if it works. The thing is I didn't notice this problem on all my deep sky shots, and I always stick to the same protocol.

If I understand correctly, if I remove "optimise darks" I don't need bias frames?

[pfile]

yes, i should have linked to those threads. one poster even said that from camera boot to boot the starburst can be different, which if true, makes taking darks a little like flats - you would have to do it before or after each imaging session.

if you don't scale your darks, bias frames are not necessary. in order to scale the dark, the bias signal must first be removed from the dark (since the bias signal does not change with frame duration.)

scaling darks, even if they match the duration of the lights, may still be beneficial. what PI does is to iteratively scale the dark and trial subtract it from the light until the noise in the calibrated light is minimized. it does this over a small preview of the light. however when there are sensor artifacts like amp glows or this starburst that are independent of the dark frame duration, you just can't use dark optimization as the artifacts are invariably under-subtracted from the lights...

rob

[Niall Saunders]

I don't know how to take dark flats.

A FlatDark is just a Flat taken with your lens cover on, at the same exposure time and temperature as used when you expised yourFlats. Take the same number of Flats and FlatDarks and, persomally, although I also acquire Biases, I never use them during calibration.

Hope this helps.

[Bobinius]

Thanx Niall and Rob. Maybe the flat dark will work.

Here is a comparison with "optimize darks" enabled and desabled in the batch process. The left image has unoptimized darks and the right one optimized. The Starburst has disappeared, but the image looks really noisy (even though the Standard deviation of the background is slightly lower!). I would not want the process to impact the quality of the final integration. Last night I did the process on another shot by doing it manually for the Master flat with dark optimization, and no optimization for the batch integration process. It worked without Starburst.

Could the method of taking flats influence the subtraction process?

PS: The integration still has vignetting so I am wondering if the flats are correctly subtracted... I'll have to try without flats at all.

[wvanreeven]

Did you use the same ScreenTransferFunction settings on both images or did you let STF compute the optimum settings for each image individually? If the latter, how does the left image look if you apply the right STF to it? Perhaps that could be a more objective way to evaluate the differences between the two.


Wouter

[Bobinius]

Good point. Those where stretched images, so I redid a histogram transformation using the same STF parameters. They look more similar in noise and no Starbust. I can still see some vignetting.

Background: 0.0081for optimization and 0.0085 for no optimization

[wvanreeven]

The vignetting is similar in both images though slightly worse in the left image. I guess you need to choose between slightly worse vignetting and starburst. Could you explain how you shoot your darks? And perhaps DBE can reduce the vignetting?

[Bobinius]

Darks, the usual way. usually in the same session with the cam cap on, same gain same temperature. Some dark series the next day, same temp and gain. Starbust does not seem to depend on that

[wvanreeven]

Thanks. And sorry, but I meant flats. How do you shoot those?

[Bobinius]

Well, the flats are still an issue with the CCD for me, I'm still trying to find the optimal way of taking them. I am using an Aurora flat field. The settings that seem to work best are when I shoot them in RGB (I have a colour camera 294C), same gain as the lights and the histogram is balanced at 50% for the green/blue channel, while the red is to the left (the flat field is predominantly blue).  I tried mono with the histogram in the middle (SharpCap uses 8 bit for those) using the same gain but the results on the final image are catastrophic and the vignetting/dust is actually increased.

[wvanreeven]

It seems that you're doing the right thing. What exposure times do the flats have? How do you create the master flat? Do you use filters when imaging? If yes, do you take flats for those as well?

Lots of questions, I know

[Niall Saunders]

The settings that seem to work best are when I shoot them in RGB

You should be shooting all your frames the same way you appear to ve shooting your Lights - i.e. in RGB mode. You should not deBayer until you have fully calibrated all of your (still in grey-scale) Lights - just before you Align and Combine the Calibrated Lights. You should only expect to see the beginnings of a colour image if you unlock the channels duriing STF.

Hope  this helps.