New Feature : Background Subtraction
Re: New Feature : Background Subtraction
Robin,
Here are some (bad, I'm afraid) raw frames I shot recently
https://mega.nz/file/6cFhFKhL#E21Et3h4q ... A0ll9iKL44
Do you need the matching master dark / flat?
Here are some (bad, I'm afraid) raw frames I shot recently
https://mega.nz/file/6cFhFKhL#E21Et3h4q ... A0ll9iKL44
Do you need the matching master dark / flat?
- admin
- Site Admin
- Posts: 13473
- Joined: Sat Feb 11, 2017 3:52 pm
- Location: Vale of the White Horse, UK
- Contact:
Re: New Feature : Background Subtraction
Hi,
if you need to use the dark/flat to make the problem happen then I will need them for testing. If the problem happens without dark/flat then I only need the raw frames (and it would hopefully simplify the tracking down of the issue).
thanks,
Robin
if you need to use the dark/flat to make the problem happen then I will need them for testing. If the problem happens without dark/flat then I only need the raw frames (and it would hopefully simplify the tracking down of the issue).
thanks,
Robin
Re: New Feature : Background Subtraction
Hi Robin,
just checked with the latest public v4.0 release, I can confirm the problem is there with or without darks/flats.
just checked with the latest public v4.0 release, I can confirm the problem is there with or without darks/flats.
- admin
- Site Admin
- Posts: 13473
- Joined: Sat Feb 11, 2017 3:52 pm
- Location: Vale of the White Horse, UK
- Contact:
Re: New Feature : Background Subtraction
Hi,
so what am I missing? I stack those five frames you shared using the folder monitor camera with background subtraction set to 'blended offset' and I get a perfectly ordinary histogram in the live stacking with no negative values. Tried both with default stacking and sigma clipping.
Am I just not seeing the problem, or not managing to make it happen?
cheers,
Robin
so what am I missing? I stack those five frames you shared using the folder monitor camera with background subtraction set to 'blended offset' and I get a perfectly ordinary histogram in the live stacking with no negative values. Tried both with default stacking and sigma clipping.
Am I just not seeing the problem, or not managing to make it happen?
cheers,
Robin
Re: New Feature : Background Subtraction
Robin,
There are two problems actually:
1- as your screenshot shows, the histogram is then highly restricted in dynamic range. This makes for very difficult black-mid-white levels adjustments, event with the shift-key modifier (or was it ctrl). Well, I hear myself say, there's always the Auto Stretch feature. So let's give it a try.
2- consequently, as soon as you auto stretch the Livestack histogram, Sharpcap has very few levels of dynamic range left that it can play with. Even with the slightest (0.1dB) adjustment of one the RGB sliders, the color cast changes so much it's close to impossible to achieve a proper color balance.
Both issues may be due to the little signal these 15s subs contain, but that's quite typical of EAA-style exposures in light-polluted environments, where this feature would be one of the most wanted ones. I could verify this was less of an issue as soon as I used longer (60, 120s) exposures, but this is not always doable (lack of tracking, alt-az mounts, etc).
There are two problems actually:
1- as your screenshot shows, the histogram is then highly restricted in dynamic range. This makes for very difficult black-mid-white levels adjustments, event with the shift-key modifier (or was it ctrl). Well, I hear myself say, there's always the Auto Stretch feature. So let's give it a try.
2- consequently, as soon as you auto stretch the Livestack histogram, Sharpcap has very few levels of dynamic range left that it can play with. Even with the slightest (0.1dB) adjustment of one the RGB sliders, the color cast changes so much it's close to impossible to achieve a proper color balance.
Both issues may be due to the little signal these 15s subs contain, but that's quite typical of EAA-style exposures in light-polluted environments, where this feature would be one of the most wanted ones. I could verify this was less of an issue as soon as I used longer (60, 120s) exposures, but this is not always doable (lack of tracking, alt-az mounts, etc).
- admin
- Site Admin
- Posts: 13473
- Joined: Sat Feb 11, 2017 3:52 pm
- Location: Vale of the White Horse, UK
- Contact:
Re: New Feature : Background Subtraction
Hi,
OK, that makes sense - because the R/G/B histograms have been lined up with each other, the luminosity histogram is much narrower, so control adjustments become more critical. Let me think about how this could be improved.
thanks,
Robin
OK, that makes sense - because the R/G/B histograms have been lined up with each other, the luminosity histogram is much narrower, so control adjustments become more critical. Let me think about how this could be improved.
thanks,
Robin
- admin
- Site Admin
- Posts: 13473
- Joined: Sat Feb 11, 2017 3:52 pm
- Location: Vale of the White Horse, UK
- Contact:
Re: New Feature : Background Subtraction
Hi,
I have added an option to zoom the horizontal axis of the live stack histogram (just a checkbox - zoom/no zoom) which gives much finer control when required. I also understand the cause of the colour balance accuracy issue, but haven't decided on the best solution yet.
thanks,
Robin
I have added an option to zoom the horizontal axis of the live stack histogram (just a checkbox - zoom/no zoom) which gives much finer control when required. I also understand the cause of the colour balance accuracy issue, but haven't decided on the best solution yet.
thanks,
Robin
- admin
- Site Admin
- Posts: 13473
- Joined: Sat Feb 11, 2017 3:52 pm
- Location: Vale of the White Horse, UK
- Contact:
Re: New Feature : Background Subtraction
Hi,
the latest uploaded version - 4.0.7995 - contains two changes that should help
1) The ability to zoom the horizontal axis of the live stack histogram - this gives better fine control over the levels
2) The level that is generated after background subtraction when live stacking is reduced from 5% of max brightness to 0.5% of max brightness. This will make the remaining background much less sensitive to colour balance tweaks.
I don't think that 2) is a perfect solution for the problem, but the best solution I can think of requires a lot more work, so this is a stop-gap improvements for now.
Reports on whether these changes help are very welcome.
cheers,
Robin
the latest uploaded version - 4.0.7995 - contains two changes that should help
1) The ability to zoom the horizontal axis of the live stack histogram - this gives better fine control over the levels
2) The level that is generated after background subtraction when live stacking is reduced from 5% of max brightness to 0.5% of max brightness. This will make the remaining background much less sensitive to colour balance tweaks.
I don't think that 2) is a perfect solution for the problem, but the best solution I can think of requires a lot more work, so this is a stop-gap improvements for now.
Reports on whether these changes help are very welcome.
cheers,
Robin
Re: New Feature : Background Subtraction
Hi,
I am just starting to use Sharpcap for live stacking and am pleased to see a Background Subtraction tool to be applied to each sub-frame. Would it be possible to have a "Manual" mode where the user specifies the offset in camera ADUs by channel? My experience is that the best way to remove light pollution and/or airglow is to measure it first in test exposures on an area of the sky you want to be "black" (I usually set black = 100 ADU), then build a subtraction "dark frame" with the appropriate constant R G B values. This would be more precise than relying on the histograms of each channel, especially if there is a large area of colored nebulae on the frame.
Thanks,
Dan
I am just starting to use Sharpcap for live stacking and am pleased to see a Background Subtraction tool to be applied to each sub-frame. Would it be possible to have a "Manual" mode where the user specifies the offset in camera ADUs by channel? My experience is that the best way to remove light pollution and/or airglow is to measure it first in test exposures on an area of the sky you want to be "black" (I usually set black = 100 ADU), then build a subtraction "dark frame" with the appropriate constant R G B values. This would be more precise than relying on the histograms of each channel, especially if there is a large area of colored nebulae on the frame.
Thanks,
Dan
- admin
- Site Admin
- Posts: 13473
- Joined: Sat Feb 11, 2017 3:52 pm
- Location: Vale of the White Horse, UK
- Contact:
Re: New Feature : Background Subtraction
Hi Dan,
probably not something I'd be looking to add at the moment (unless I start hearing the same request from multiple users). Right now the simple background subtraction modes work by picking the brightness of a very low centile (below the first centile) of the histogram for each channel, which means they should pick up on the darkest parts of the field of view even if the frame is mostly filled with nebulosity, etc.
cheers,
Robin
probably not something I'd be looking to add at the moment (unless I start hearing the same request from multiple users). Right now the simple background subtraction modes work by picking the brightness of a very low centile (below the first centile) of the histogram for each channel, which means they should pick up on the darkest parts of the field of view even if the frame is mostly filled with nebulosity, etc.
cheers,
Robin