SharpCap Forums

Hi robin ,
I just changed from a 183mc to a 533mc pro. I ran the sensor analysis and no errors. I had a lot of trouble with flats over correction. I made sure temp, color space, offset, binning were all correct. Then I used the flat wizard in sc. then I changed to longer exposure flats. I also tried various average values on histogram. I finally found a way to make it work I set all flats, dark flat, darks and lights to 0 offset. I originally set everything to 11 which was the value suggested for lights in smart histogram. Another thing I noticed smart histogram tries to give me 6-8 sec for subs. I tried several different exposure and 30 gives best results. Gain is at 101 to take advantage of hcg. Should I remove the sensor analysis and go with the default? If so how do I remove analysis?

Thanks for the help

Hi,

there should be no need to remove the analysis, as it's likely to be very similar to the default analysis file included with SharpCap (unless the illumination brightness varied a lot during certain parts of the measurement when it should be constant). If you do want to wipe the measured analysis file then you can either wipe all analysis files from the SharpCap settings 'Reset' option (see https://docs.sharpcap.co.uk/4.1/#Saving ... 20Settings) or you can wipe individual ones by deleting them from the folder where they are saved (see https://docs.sharpcap.co.uk/4.1/#Sensor ... ta%20Files).

I would need to see a full set of files to try to analyse properly whether the issue is one with the settings used or the way flats/darks are being used or is a bug in SharpCap. That set would be

* One or more unprocessed light frames (without dark/flat applied, not stacked, straight from the camera).
* The master dark created in SharpCap
* The master flat created in SharpCap

All of these with the settings files that are created alongside them that record the camera settings used to capture.

Setting the brightness/offset to zero is basically minimizing the scale of the problem without really fixing the cause - it also opens up the possibility of other issues in the end due to black level clipping of images.

cheers,

Robin

Robin,
I re ran the sensor analysis and used a brighter setting on my light panel. The analysis went much quicker so I think the light source was to dim before. I used the smart histogram and the results seemed closer to the needed exposure. The gain stays close to the hcg value of 100 with 13.7 sec. Also the offset is now 5 instead of 11. I used these settings for lights and darks. I used the flat wizard for the flat with a 5 offset. The over correction isn’t showing anymore. Looks like offset of 5 is magic number.
Thanks for the help

Hi,

interestingly, I had a go at a similar setup last night with my PlayerOne Poseidon camera (using flats and darks and a very strong stretch). I also got over-correction, and I have most of the data I need to try to analyse why it might have happened. I do need to go back and re-do the flats before coming to any conclusions though - it's possible that the flat frame illumination was off, and that could easily have caused the issue, so I need to be sure that isn't the case before digging into the processing algorithms.

Anyway, good to hear that you are getting good results now - clear skies!

Robin

Robin,
Just for your info I was shootng through an optolong lquad filter. This camera seems to be very sensitive to offset settings. Also for your info im in a bortle 7 zone. I attached a link for last night files.

https://www.dropbox.com/scl/fo/7c8hwp8h ... w6x92&dl=0

Hi,

thanks for those files - any chance you could add in the master dark file that was being used in the stack so I can have a look at that too? According to the capture settings file, it should be found at
The effect is very subtle in those particular images, but I think I can see a brightening in the corners of the final image when I stretch the background.

cheers,

Robin

Robin,
Sorry I had already changed to another set of darks. I uploaded last night’s complete set to the same link I sent before.

Thanks

Hi,

just to post an update on this issue...

I have been trying to collect some data of my own to check this sort of problem out, since I can take care to try to do all the 'best practice' things to get good flat frame correction. Unfortunately, my first go at flats weren't any good (an artificial light on a piece of white paper over the end of the telescope), so I had to get hold of a proper flat panel to do things the 'right way'. With the flats from the flat panel, I got better results, but still some artifacts under strong stretching - see below...
Here you can see what looks like a failure to correct for a darkened top-right corner along with over-correction of a dust shadow in the bottom right. Compare that with the same process done without flats...
Without the flats, the darkening on the right is worse, and there is darkening at the top-left corner, so the flats are definitely helping, but not 100%

Now, for comparison, I ran the same stacking (same frames, darks and flats) in DeepSkyStacker, and got these results (I tried to get the stretch level similar to that in SharpCap)
Now, interestingly we can see the same sort of problems that we saw in the SharpCap result - still dark in the top-right corner, and over-correction of the dust shadow at the bottom right.

Ok, time to get out the big guns... Pixinsight - the 'simple/fast' version of the weighted batch pre-processing script (again colour balanced and stretched to look similar to SharpCap)
We still have the darkening to the top-right, but the over-correction of the dust shadow is not present...


My current anslysis of these results is

* Similar patterns of remaining dark areas in all three applicaitons suggest that :
1) The flat frames are still not 100% good - perhaps because I captured the flats several days after capturing the light frames - maybe something shifted slightly.
2) The basic 'subtract dark and divide by flat' algorithm is working properly in SharpCap and DSS (since they are giving roughly equivalent results)

* PixInsight is (perhaps unsurprisingly) doing something cleverer, even in the 'fast' processing script. The trouble is that it's hard to be sure exactly what is going on there that is making the difference.

I'm going to have another (even more careful) go at collecting data when we next get a clear night and then re-check. I'm also going to see if I can work out a way of making PixInsight do the equivalent of the simple subtract dark/correct for flat that is going on in DSS and SharpCap to validate that gives similar results.

cheers,

Robin

Robin,
I’m trying a different group of flats tonight. I found using folder monitor on previous raw data. Flats taken at 100 gain and 2 sec exposure with a 20000 adu seems to work best. I noticed the corner areas in my pics have a tendency to be red in color. This last set of flats made sure I used 50/50 for wb. The darks are 50/50 as well. I will use 50/50 on lights and see what happens. I saw a video about pixinsight where you can use pixel math to do simple math functions to frames. I’ll let you know how it goes.

Robin,
I uploaded files from another night. The flat files are named 533 20k (gain) fd for flat darks. I found the 0 gain with no dark flats seemed to work best. The over correction is there but less. I have a bottom right corner which seems to be the worst area.
I was thinking about the math involved with flats. Since you are dividing pixel value of light by same pixel on flat. What about a slider in sc where user could multiply flat value by a selected factor to basically adjust the strength of the flat correction?

Thanks for the help