Flats Capture

[Gordon]

Hi Robin,

I was using the Flats capture in SharpCap 4.1 using my Altair Astro Hypercam 183C (fan cooled) yesterday. I used the auto exposure and gain button in SharpCap to set the right levels. All went well and the white level centered on 50%. It captured the flats and dark flats, saved and I then loaded the Master Flat into Pixinsight.

When I looked at the histogram it was around 75% which was the same level the green channel was at during the flats capture. I did try using the flat but didn't get a great result. I manually changed the histogram to 50% and go the best flat I've ever had. A real game changer for me!

I also noted the exposure maxed out at 1 sec and the gain went up as I tried out different light levels. I note it as in the documentation it says its best to keep the gain low to reduce noise? I was expecting the exposure to go up rather than the gain but that's just my simplistic view on it!

I've attached the log file, camera settings and a snap of the master flat with the histogram in Pixinsight. Apologies if get the post twice, I didn't look like the first sent as it said I needed to be logged in, hopefully you'll just get this one. If you need anything else please let me know.

Thanks for looking.

Regards

Gordon

[admin]

Hi Gordon,

so, first just to explain why the autoexposure sticks at 1s - it's basically set up with a range of settings it is allowed to use for the gain and exposure, and the range for the exposure is limited to a maximum of 1s, so once it gets to that point and the image is still too dim, it will use the gain to brighten the image. There is no real special reason for the limit being 1s, although having some sort of limit helps avoid the autoexposure procedure itself taking an extremely long time.

Looking at the screenshot, I presume that is the original SharpCap created/exposed flat frame loaded into PixInsight, since the histogram peaks at near to 75%. The histogram itself looks fine for a flat frame (everything well clear of both ends of the graph). The small secondary peak on the left side of the main peak probably represents the areas that are affected by dust shadows - you can see this by stretching the image...

Untitled.jpg (335.3 KiB) Viewed 571 times

As to why the second attempt at 50% histogram worked much better, I am at a bit of a loss - in theory the requirements for good flat correction are fairly simple

* Flat frame histogram should be clear of both the far left and far right of the graph so that there are *no* overexposed pixels and as few underexposed pixels as possible (sometimes, with heavy vignetting, you cannot avoid under-exposed corners)
* Bias or dark flat subtraction is used with the same settings for gain/offset/brightness etc so that the remaining signal after subtraction results purely and linearly from the amount of light hitting the sensor
* When *using* the flat, proper subtraction of dark frames (or at the very least a bias level to remove the offset) so that the flat is applied to a dark subtracted image that again represents purely light hitting the sensor. Again, matching dark frame settings to light frames is important

If all of those are satisfied then the exact position of the histogram peak should be have a very minor impact on quality - 50%, 60%, 70% - all should be fine. It's hard to see how that change could/should have a big impact. I suppose it is possible that during the original capture, some inadvertent change or a minor light leak or something else may have sabotaged the capture of the flat and/or dark flat frames.

There is one more thing I suppose to mention - PixInsight - the dark/flat calibration routines in PixInsight are very effective, but also very complex compared to the relatively simple 'subtract the dark frame then divide by the (flat-dark flat) frame approach' that I have been considering in my answer above. If you have been using PI to do the calibration then maybe something in its more complex calculations works better with a flat at approx 50% histogram. I have to admit though that this is pure speculation, since I don't even understand the details of how PI does its calibration calculations.

cheers,

Robin

[Gordon]

Hi Robin,

thanks for taking the time to look at this, its appreciated. You may be right about Pixinsight and if it prefers (at least for my flats) the flats at 50%. Flats have been a real problem. At least with this it seems I've route home.

Thanks again.

Regards

Gordon