Page 4 of 6

Re: Live Stack with Flat Frame Bug?

Posted: Wed Dec 26, 2018 4:14 pm
by admin

just another small update… I did some more testing with the EL panel and I found that by setting a much smaller aperture on my CS mount lens that I could find problems with flat correction when the monochrome flat option was enabled. Basically the correction was pretty good but not excellent when the image was stretched it was possible to still see mottling in the flat corrected image. This didn't happen at all when the monochrome flat option was not selected. I'm going to try to work out what is going on with monochrome flats that stops them from working perfectly – so far having tried to think my way through how it works nothing seems to be amiss, but I'm sure I get there in the end.

Cheers, Robin

Re: Live Stack with Flat Frame Bug?

Posted: Wed Dec 26, 2018 6:21 pm
by donboy
Robin, I have also been following along with this thread and live with Chris (BlackWikkett). I am finding the same issues. I can get a better color balance with a color flat but I can't eliminate vignetting and dust spots. Mono flats have not worked for me.


Re: Live Stack with Flat Frame Bug?

Posted: Wed Dec 26, 2018 9:52 pm
by BlackWikkett
Hi Robin,

Thanks for the update. I'm at your disposal to test next clear night, just let me know how I can assist.


Re: Live Stack with Flat Frame Bug?

Posted: Thu Dec 27, 2018 6:48 pm
by admin
Hi folks,

I have a new version for you to test – ... 3.2.5903.0

Hopefully this will bring the monochrome flat option up to at least the same standard of correction as the non-monochrome option while maintaining the colour balance of the original image.

In my testing I have been working with the following sequence of steps:

1) capture monochrome flat while using an appropriate dark frame
2) apply new flat and dark frame and check that correction is good by stretching resultant image
3) reduce exposure to give a much fainter light level and using appropriate flat for the reduced exposure check that correction is still good by stretching resultant image strongly

I believe that this approach of checking the flat frame correction at 2 different brightness levels ensures that the algorithms being used are correctly adjusting for any black level offset (as previously discussed). With the current code under test conditions I can increase the stretch to the point that I can see camera noise as the biggest deviation from flatness in the output image when stretch which I think is a good result.

If there are still issues with this new version then please capture and pass on to me the images listed below which will let me recreate and try and understand the problems you are still having

1) the master dark and master flat frames created by SharpCap that you are using
2) at least one individual flat frame and at least one dark flat frame
3) at least one individual dark frame that was used to create the master dark that you are using
4) at least one light frame of the target you are trying to capture (captured with no dark or flat subtraction applied so that I can see the raw image)

please also include the capture settings.txt files for items 2, 3 and 4 so that I can fully understand the setting used for each capture.

Hope this helps, Robin

Re: Live Stack with Flat Frame Bug?

Posted: Sun Dec 30, 2018 3:43 am
by Astrogate
Hi Robin

I finally got some clear skies to test my new ASI294MC Pro using my C9.25 # F7.2 (Used a Celestron 6.3 reducer/flattener). I only capture two flat frame one a mono with bias selected and the other a colour with bias selected using your latest SC version you posted here. Target was M33 using 350 gain, offset 10, cooler -15 C (matching dark applied) and bin 2. It seems to me that the dark areas caused by the vignetting is being corrected but where those areas are they are corrected with a colour cast. They colour cast is not as prominent using the flat that had mono non-selected but bias applied. All of my flats and darks where using the exact same settings as my lights. I am also including M1 with and without a flat applied (both have darks applied). You can clearly see that the flat applied is making a big improvement but there is a colour cast.

Cheers and Happy New Year Robin!

Chris A

Re: Live Stack with Flat Frame Bug?

Posted: Sun Dec 30, 2018 3:46 am
by Astrogate

Here are the M1 images with and without flats but dark's used for both. I had a lot of thin clouds so colour balance was hard to determine due to excessive red from clouds.

Chris A

Re: Live Stack with Flat Frame Bug?

Posted: Sun Dec 30, 2018 4:14 am
by Astrogate

Sorry, I forgot to mention that the flat for M1 was captured last week so not with the latest build but applied using the latest build. It was a mono flat with no bias used but again same setting as the lights.

Chris A

Re: Live Stack with Flat Frame Bug?

Posted: Mon Jan 07, 2019 12:08 am
by BlackWikkett
Hi Robin,

Sorry for taking so long with a test of the new version, bad weather, holidays etc.

Used 4 different flat frames for the images below. All flats were 209 ms 300 gain using T-shirt and LED light panel. All images have the described flat applied along with 25 frame master dark. All images were 30 sec exposure at 300 gain, 10 offset (brightness) cooler at -15c.

1. Color Flat
2. Color Flat with BIAS
3. MONO Flat
4. MONO Flat with BIAS

All four flats worked removing vignetting and dust bunnies. The color flats tend to give a better color balance while the mono flats require aggressive color balance with live stack color controls.

I have also discovered what may have been causing the initial issues with the terrible vignetting. Once I made the initial test of the different flat frames I made a few changes to offset and color balance in image controls. While using the MONO with bias flat i tuned offset up to 20 from the original 10 offset. This caused the vignetting to come back but flats were still removing the dust bunnies. Next I corrected the white balance adjusting the red and blue channels in image controls until I had red green and blue channel histogram aligned and mean read out in smart histogram roughly equal. A new live stack with white balance and 20 offset made the vignetting even worse and now dust bunnies start to appear.

My theory of what's happening is image controls may be the culprit. Image controls adjust data from the camera before it's handed off to SharpCap to apply flats, darks and then stacking or other enhancements. Not sure what if anything could be done change this behavior or if I'm totally off the mark.
My guess is you could point to an object, white balance and set offset in image controls. Then with the white balance and offset applied setup your light panel and take your flats. Once you have your flat then slew back to the object and start live stacking. Then isn't very realistic since the whole point, in my mind, of live stacking is to be able to get great images without having to post process and add a lot of complexity to the experience. I don't have a permanent setup so nightly setup entails mount leveling, balance, alignment, focus and flats acquisition. Ideally I'll have one master flat for the evening unless I change filter or major focus change is called for.

For now flats seem to be working as long as you make no adjustments to image controls once you have the flat set.

The images from my testing are below let me know if you need further explanation. I'd like to hear your thoughts on my "theory"

Thanks for all you do!

ImageM45 No Flat Stack_10frames_300s_WithDisplayStretch by Black Wikkett, on Flickr

ImageM45 COL FLAT Stack_10frames_300s_WithDisplayStretch by Black Wikkett, on Flickr

ImageM45 COL BIAS FLAT Stack_10frames_300s_WithDisplayStretch by Black Wikkett, on Flickr

ImageM45 MONO FLAT Stack_10frames_300s_WithDisplayStretch by Black Wikkett, on Flickr

ImageM45 MONO BIAS FLAT Stack_11frames_330s_WithDisplayStretch by Black Wikkett, on Flickr

ImageM45 MONO BIAS FLAT 20 Offset Stack_10frames_300s_WithDisplayStretch by Black Wikkett, on Flickr

ImageM45 MONO BIAS FLAT 20 Offset WB Stack_10frames_300s_WithDisplayStretch by Black Wikkett, on Flickr

Re: Live Stack with Flat Frame Bug?

Posted: Mon Jan 07, 2019 9:04 pm
by admin

thanks for sharing your new results – it sounds like we are homing in on the cause of any problems now.

Certainly you could expect issues with incorrect flat frame correction if you make the sort of adjustment you are describing to the camera controls just before taking your light frames. To get correct correction you must do one of the following options

1) capture the flat with the bias option enabled and not change the camera controls relating to offset and colour balance between capturing the flat and capturing the lights. If these controls are change then no longer knows where the black level is in the light frames and cannot correct properly.

2) capture the flat with the bias option enabled and use a dark frame when capturing lights. The dark frame must be captured at the same camera settings (in particular offset, gain, exposure et cetera) as the light frames otherwise the dark frame subtraction will be incorrect thus making the flat frame correction also incorrect.

3) capture the flat frame using a dark flat (captured at the same settings as the flat) and capture the light frames using normal dark subtraction as described in option 2 above.

Unfortunately there's no real way round the requirements as proper flat correction requires that the 'image' part of the signal is properly separated from the background level so that only the image parts can be scaled according to the flat frame data.

Hope that makes sense, Robin

Re: Live Stack with Flat Frame Bug?

Posted: Sun Jan 13, 2019 3:46 am
by donboy
For last two nights I changed my way of doing flats. What I tried was mono & mono/bias flats with longer exposure with offset at zero and the red & blue channels set to 50 each. The first attempt I used 3 sec exposures and got improved results with the lights taken also with no offset. But last night I tried 8 sec flats and also a few 5 sec flats using 16 frames. The images looked quite flat with with either 5 sec and 8 sec and it was hard to tell if the mono/bias flats worked any differently. When doing longer exposure flats one needs to keep the ASI294 above 120 gain so I added extra diffusing material to reduce the light from the light panel and I adjusted gain setting the white peak to 50%.

Highly stretched image
flat:8sec X 16-mono/bias-Red50/Blue50-zero offset
8sec X 16-mono/bias-Red50/Blue50-zero offset
M82_00_16_22_27frames_810s_stretched.jpg (601.33 KiB) Viewed 520 times

Highly stretchedimage
Flat:5sec X 16-mono/bias-Red50/Blue50-zero offset
5sec X 16-mono/bias-Red50/Blue50-zero offset
M63_02_59_01_10frames_300s_WithDisplayStretch.jpg (611.26 KiB) Viewed 520 times