Beta nearly over (but some new features snuck in too)

[admin]

Hi folks,

This week's update 4.2.15290 is maybe the last beta version if all goes well - soon we will be moving to release candidates and then the full SharpCap 4.2 release.

There is currently only one known issue in 4.2 that I am still trying to track down and ensure it is fixed (auto stretch somehow feeding negative black levels into the live stack stretch). Changes this week might fix the issue, or at least add enough diagnostic information that they will let me fix it in the next update.

A few last minute new features have snuck in to this week's release too...

CLAHE Enhancement for Deep Sky Live Stacking
There is now a new CLAHE enhancement available in the Deep Sky Live Stacking enhancements tab. This applies 'Contrast Limited Automatic Histogram Equalization' to the image being displayed - effectively this is an extra stretch that rather than being constant across the whole image is variable, stretching the faint areas of the image more while leaving the brighter areas more or less alone. It can help bring out faint detail without needing to use a very strong stretch globally, and gives an extra option for image processing.

You have two adjustments for CLAHE - tile grid size and clip threshold. Higher clip thresholds give a stronger effect, while low values for that setting keeps the effect of CLAHE subtle and limited. The tile grid size controls how many sub-areas the image is broken down into when applying the algorithm - higher numbers mean a finer grid of more, smaller squares, making the effects of the CLAHE transform more variable across the image. Some images respond well to low values for tile grid size, others seem to do better with larger values.

An AllSky Reticule
This is a very simple overlay in the reticules drop down on the tool bar - it overlays the compass directions on the image on screen (not the saved image). You can move North around with the right mouse button to match the orientation of your camera and that orientation will be saved for the next time you come back to this reticule.

Improvements to Light Pollution Measurement
I have re-added the ability to select colour channels for measurement as well as the default of measuring the sum of all three channels when using a colour camera. Measuring all 3 channels is the best representation of your true light pollution level, but in some cases it will be interesting to see how the values break down per channel.

I've also added information about the solar altitude and how dark it is (astronomical darkness, twilight, etc) to this tool so that you can be sure that it is properly dark when testing.

New/Changed sequencer steps
There is a new step to select a reticule and changes to the steps that work through a series of RA/Dec co-ordinates so that you can opt to skip points where the co-ordinates are currently below the horizon. Note that this check for the horizon evaluates against any custom horizon that you have set up, so it is not just a simple altitude > 0 check.

Improvements to Solar System Live Stacking
There were a few features in Solar System Live stacking that worked on their own but caused issues when combined with other features - several of these are now fixed including

* Corona boost for solar images now compatible with rotating the viewed image
* Drizzling the image 2x2 as part of planetary stacking no longer causes issues with frame quality history, rapid updates when changing settings and a few other things


cheers,

Robin

[Borodog]

Great to hear, Robin. Is the drizzle a true drizzle, or is it a 2X up-scaling prior to aligning and stacking?

I have some questions about automated planetary imaging, specifically around the meridian flip, which is the only remaining sore spot I have. To manage the meridian flip, I use the following set of steps manually:

1) Turn off Feature Tracking for the main camera
2) Use the Offset Goto functionality on my electronic finder & camera (a 30mm guide scope) to re-slew to the planet, with position refinement enabled; that is always close enough to put the target planet on the main camera sensor
3) Reset the main camera's ROI to the full sensor
4) Turn Featuring Tracking back on, which guides the target back to the center
5) Change ROI to the capture size, which turns off Feature Tracking automatically (is there a reason for this?)
6) Turn Feature Tracking back on

Can all of this be done in the Sequencer, or is python required?

[MarMax]

WOW . . . Robin . . . CLAHE . . that is awesome so thank you. Can't wait to test it out.

Mike

[Gfischer]

Robin,

Thank you for the new sequence step to unlock ui. I installed in sequence and when clouds leave I’ll give it a try.

[Gfischer]

Robin,

Thank you for the new sequence step to unlock ui. I installed in sequence and when clouds leave I’ll give it a try.

[Gfischer]

Hi Robin,
The new sequence step works great. Now I can zoom and recenter view during sequence run! Clahe works great on nebula.

Thanks for the help

[admin]

Hi folks, glad the new features are popular

The drizzling is not true drizzling but a 2x upscale before alignment/stacking. That gives you an increase in alignment accuracy and a finer control over wavelet sharpening due to the doubled image scale (although the adjustments can be tricky to get right). I think a full drizzling algorithm (requiring a droplet from each input pixel to be shared over a number of stack pixels) is just too computationally intensive to run at live imaging speeds (with the exception of very small ROI).

I'm afraid the meridian flip is beyond the power of the sequencer at the moment, but it would probably be do-able via scripting since that can now control feature tracking. The only other thing that is complex/odd is to turn on the offset goto, but selecting the "Alignment Mark" reticule via scripting does that.

Changing the ROI breaks the assumptions that the feature tracking code needs to make about the image - if you are tracking with phase offset or image features then the ROI change means that the things that you were tracking are now moved or potentially gone from the image, etc. There are too many cases where the feature tracking would potentially do weird stuff after an ROI change to allow it to stay active.

cheers,

Robin

[MarMax]

Robin,

I was able to try the CLAHE enhancement last night for the first time. Here's the CN post on it:

https://www.cloudynights.com/forums/top ... t-14871435

My only reference for a "CLAHE-type" enhancement is the Photoshop Shadows/Highlights.. tool which I've been using for 20 years. When CLAHE was first presented on CN, even though it was over my head technically, it sounded a lot like the Shadows/Highlights.. tool. Now that I've played a bit with CLAHE it does have similarity to the Shadows/Highlights.. tool.

As always you amaze me with your skill and dedication to SharpCap and willingness to put forth new suggestions, so thank you!

Mike

[Borodog]

Hi folks, glad the new features are popular

The drizzling is not true drizzling but a 2x upscale before alignment/stacking. That gives you an increase in alignment accuracy and a finer control over wavelet sharpening due to the doubled image scale (although the adjustments can be tricky to get right). I think a full drizzling algorithm (requiring a droplet from each input pixel to be shared over a number of stack pixels) is just too computationally intensive to run at live imaging speeds (with the exception of very small ROI).

I'm afraid the meridian flip is beyond the power of the sequencer at the moment, but it would probably be do-able via scripting since that can now control feature tracking. The only other thing that is complex/odd is to turn on the offset goto, but selecting the "Alignment Mark" reticule via scripting does that.

Changing the ROI breaks the assumptions that the feature tracking code needs to make about the image - if you are tracking with phase offset or image features then the ROI change means that the things that you were tracking are now moved or potentially gone from the image, etc. There are too many cases where the feature tracking would potentially do weird stuff after an ROI change to allow it to stay active.

cheers,

Robin

Thanks, Robin. That is about as expected. I have some time to figure out how to handle the meridian flip via python, as Saturn will shortly be transiting before midnight, so handling it manually is not a big deal, and it will be quite a while before Jupiter is rising early enough to even be a target for me again.

Can a python script be run from within the sequencer? Perhaps having a script that just does the flip and reacquires the target that could be run from the sequencer would be the simplest way to go.