New Feature : Planetary Live Stacking

All the latest news about new features and improvements to SharpCap
Post Reply
DrBobAZ
Posts: 71
Joined: Thu Feb 25, 2021 11:41 pm

Re: New Feature : Planetary Live Stacking

#51

Post by DrBobAZ »

Finally figured out the details of this great new feature. Since I have a lightweight mount it is difficult to track planetary disks. Feature tracking took care of this.

Scottsdale AZ 18-22:00 hrs 11/25/2023 just messing around. With the new sharpcap capability of live stacking of planets.

My hurdle was understanding the feature tracking in sharpcap. Once I did, I was able to keep the planet (actually its center of gravity) in the middle of the ROI, despite my rickety ole SW az-gti mount and tripod.

C6/asi678mc/uv-ir discovered that when the on-screen color was 'good' then the captured video coloration was a bit off. Still working on that.

Here is a screenshot from one of my experiments

8msec, 184 gain 8000 frames stacked in sharpcap, best 25


Seeing was my typically poor rating, and this was the 'backside' of Jupiter, i.e. without the GRS.

Perfect for debugging the workflow and study results.
LiveStack_Sharpcap.png
LiveStack_Sharpcap.png (260.33 KiB) Viewed 140561 times
Bob
Borodog
Posts: 324
Joined: Fri Jan 01, 2021 7:25 pm

Re: New Feature : Planetary Live Stacking

#52

Post by Borodog »

Robin,

Why does ROI-based feature tracking need a calibration step? Wouldn’t you just check to see that the COG has reached the edge of the dead zone and then just recenter the ROI on the COG? It’s seems like with ROI based you should be able to just immediately click Guide and you’re off.
Borodog
Posts: 324
Joined: Fri Jan 01, 2021 7:25 pm

Re: New Feature : Planetary Live Stacking

#53

Post by Borodog »

DrBobAZ wrote: Sun Nov 26, 2023 3:07 pm Finally figured out the details of this great new feature. Since I have a lightweight mount it is difficult to track planetary disks. Feature tracking took care of this.

Scottsdale AZ 18-22:00 hrs 11/25/2023 just messing around. With the new sharpcap capability of live stacking of planets.

My hurdle was understanding the feature tracking in sharpcap. Once I did, I was able to keep the planet (actually its center of gravity) in the middle of the ROI, despite my rickety ole SW az-gti mount and tripod.

C6/asi678mc/uv-ir discovered that when the on-screen color was 'good' then the captured video coloration was a bit off. Still working on that.

Here is a screenshot from one of my experiments

8msec, 184 gain 8000 frames stacked in sharpcap, best 25


Seeing was my typically poor rating, and this was the 'backside' of Jupiter, i.e. without the GRS.

Perfect for debugging the workflow and study results.

LiveStack_Sharpcap.png

Bob
If you're talking about the saved time lapse, I think that's a bug/oversight. The saved time lapse does not include the display stretch. I reported it to Robin.

Also, be careful switching the time lapse save format to PNG. It will crash SharpCap. Also reported.
DrBobAZ
Posts: 71
Joined: Thu Feb 25, 2021 11:41 pm

Re: New Feature : Planetary Live Stacking

#54

Post by DrBobAZ »

Borodog wrote: Mon Nov 27, 2023 4:29 am
DrBobAZ wrote: Sun Nov 26, 2023 3:07 pm Finally figured out the details of this great new feature. Since I have a lightweight mount it is difficult to track planetary disks. Feature tracking took care of this.

Scottsdale AZ 18-22:00 hrs 11/25/2023 just messing around. With the new sharpcap capability of live stacking of planets.

My hurdle was understanding the feature tracking in sharpcap. Once I did, I was able to keep the planet (actually its center of gravity) in the middle of the ROI, despite my rickety ole SW az-gti mount and tripod.

C6/asi678mc/uv-ir discovered that when the on-screen color was 'good' then the captured video coloration was a bit off. Still working on that.

Here is a screenshot from one of my experiments

8msec, 184 gain 8000 frames stacked in sharpcap, best 25


Seeing was my typically poor rating, and this was the 'backside' of Jupiter, i.e. without the GRS.

Perfect for debugging the workflow and study results.

LiveStack_Sharpcap.png

Bob
If you're talking about the saved time lapse, I think that's a bug/oversight. The saved time lapse does not include the display stretch. I reported it to Robin.

Also, be careful switching the time lapse save format to PNG. It will crash SharpCap. Also reported.
Borodog...I haven't tried the time lapse yet. This was something I did with adjusting the WBR and WBB of the camera. I deviated away from my usual settings and the video capture had a blue tinge that I couldn't get rid of even when the live stack looks pretty decent color wise.

Here is a question for Robin and yourself: Is it possible to pause the time lapse (briefly < time between frames) to recenter the COG in the ROI with feature tracking?

Bob
Borodog
Posts: 324
Joined: Fri Jan 01, 2021 7:25 pm

Re: New Feature : Planetary Live Stacking

#55

Post by Borodog »

Bob,

If you mean pause to re-adjust the location of the planet and ROI on the chip, then I don't think so. But you can always use PIPP to assemble multiple timelapses into a single one after the fact. Very easy.

I'm no sure what you mean by "recenter the COG in the ROI with feature tracking" since the COG is always stabilized in the ROI and the purpose of feature tracking via ROI is to track the planet across the chip with the ROI.
DrBobAZ
Posts: 71
Joined: Thu Feb 25, 2021 11:41 pm

Re: New Feature : Planetary Live Stacking

#56

Post by DrBobAZ »

Borodog wrote: Mon Nov 27, 2023 5:06 am Bob,

If you mean pause to re-adjust the location of the planet and ROI on the chip, then I don't think so. But you can always use PIPP to assemble multiple timelapses into a single one after the fact. Very easy.

I'm no sure what you mean by "recenter the COG in the ROI with feature tracking" since the COG is always stabilized in the ROI and the purpose of feature tracking via ROI is to track the planet across the chip with the ROI.
Mike

Thanks for reply.

With my setup (only rough polar align, and az-gti carrying beyond weight limit) it takes about 10 min to cross the chip dimension (asi678). I want to do a timelapse of say 60 min with frames added every 2 min.

I use ROI/COG feature tracking at a small ROI until it reaches the chip boundary, then I have to expand the ROI to max and then recenter the planet, and reduce the ROI again and reestablish feature tracking where it left off.

If I do that faster than the interframe time, It seemed like I might be able to continue with the same time lapse. But I think you are suggesting that I can splice several videos together with PIPP in several time lapse segments. Is that correct? Does sound doable, and I am familiar with PIPP.

Bob
User avatar
admin
Site Admin
Posts: 12984
Joined: Sat Feb 11, 2017 3:52 pm
Location: Vale of the White Horse, UK
Contact:

Re: New Feature : Planetary Live Stacking

#57

Post by admin »

Hi folks,

I should be getting a new update out today with some further improvements, but will try to answer questions/points above...

@starLights - I think that in theory modern phones are probably powerful enough to do this sort of thing if the phone was well mounted over the eyepiece. I'm afraid I have precisely zero interest in writing phone apps though (and no time either - SharpCap for windows fills all the time I have available).

On pausing the stacking, I think that would be possible, although trying to maintain the stack alive and paused while changing camera ROI would be an interesting challenge!

@Borodog, I have found and fixed the PNG bug already. I'm not currently planning on changing the timelapse to use display stretch as it would be awkward to code and have several drawbacks, in fact I may drop the display stretch from the 'save as PNG' button for consistency as you can always use the button to save the display stretched image. This would let me save the PNG (I'm thinking a TIFF option too) as 16 bit, which has advantages if you plan to do further processing. I have already added an extra brightness boost slider into the planetary stacking which covers a lot of what I suspect you were using the display stretch for.

The requirement to 'calibrate' the ROI tracking was a bodge to let it get developed quickly without risk of breaking the existing tracking. Looking at it again, I could probably do that internally when the 'Monitor' button is pressed, removing the need to press the 'Calibrate' button.

Stabilization to leave the planet in position rather than center it - yes, on the list to happen at some point.

As I was a little late to the party with this feature (Jupiter already past opposition when I released it), I am going to focus on getting the basics working well over the next couple of weeks to benefit the most people with a working feature. Beyond that we can start looking at some more fine tuning and some of the nice to have features.

cheers,

Robin
User avatar
admin
Site Admin
Posts: 12984
Joined: Sat Feb 11, 2017 3:52 pm
Location: Vale of the White Horse, UK
Contact:

Re: New Feature : Planetary Live Stacking

#58

Post by admin »

Ok, further update...

I have uploaded a new version this morning 4.1.11388, which has some hopefully useful improvements to the planetary stacking.

Compared to the 4.1.11368 interim version, the changes are
  • Add option to halve wavelet scales
  • Add colour channel alignment
  • Fix crash on PNG timelapse
  • Add TIFF option to save timelapse
  • Save TIFF/PNG timelapses as 16 bit images to allow for further processing
  • Allow AVI timelapse when camera in 16 bit mode
  • Add a brightness adjustment for the planetary stack
  • Remove final wavelet level 5 (have not found it useful, removing it speeds up procesing)
  • Add checkbox to view unprocessed frames
Some of these deserve a little more discussion...

Halving wavelet scales

I found that the sharpening worked pretty well on over-sampled images, but couldn't be tuned to give good results on smaller images that were either under-sampled or near critical sampling.

I suspect that in time I will have to re-do the sharpening code, which is a little crude, but an effective way to get much improved results turned out to be to double the size of the image, wavelet sharpen the zoomed image, then rescale back to the original size. This is what the 'Halve wavelet scales' option does behind the scenes - it will slow down processing, but will probably only been needed for relatively small ROI captures anyway, so that should be OK.

Colour alignment

This option will digitally re-align the Red and Blue channels with the Green as the image is being processed (before sharpening). It is only enabled if the checkbox is checked *and* the measured offset of either the blue or red channels from the green channel is at least 1 pixel. As well as helping suppress blue-one-side-red-the-other fringes on the image, it will also help with general sharpness as it will align the image features better across the image.

Note that an ADC will still do a better job, since the image processing approach cannot do anything about blurring within the colour channels caused by dispersion across the range of the colour channel.

Brightness Adjustment

Brightness is a bit of an inaccurate name for this, but I struggled to find an accurate name that wasn't either too long or potentially confusing in some other way. A traditional brightness control lifts the brightness of each pixel by the same amount. This instead scales up pixel values by a constant multiplication factor, so if you choose 1.2 it will make all pixels 20% brighter than they were. It's handy if your camera images are a bit under-exposed, allowing you to brighten the output image without affecting the stack. I suspect that I may have to change this in a later update to be one of a brightness/contrast pair.

cheers,

Robin
Borodog
Posts: 324
Joined: Fri Jan 01, 2021 7:25 pm

Re: New Feature : Planetary Live Stacking

#59

Post by Borodog »

Great update, Robin! Thank you so much.

However, I am going to push back on the not applying the Display Stretch to the time lapse, though. It really is important to getting a good looking time lapse without having to post-process. The most critical part is that it allows you to clip out the ringing, and the second most critical part is it allows you to increase the contrast by moving the midpoint to the right. Brightening the image by moving the white point left is nice but it's the least of the 3.

Would it be possible to have the time lapse routine simply grab the display stretch settings and apply them to the image as a last step before it is written? I imagine that the transform based on the black, mid, and white points must be fairly simple, no?

Personally, if you don't implement this, it's not the end of the world for me; I can write out PNGs, batch process the frames in Wavesharp to apply the histogram adjustments, and then either assemble the final animation as an animated PNG in Wavesharp or as an AVI or animated GIF in PIPP. But I'm a bit of a power user who already has some background with planetary imaging. I imagine many people wanting to use this will just want to write an AVI that looks like what they see on the screen and share it, and will be disappointed if what they see is not what they get.

Edit to add: That Show Unprocessed Frames checkbox is genius!
User avatar
admin
Site Admin
Posts: 12984
Joined: Sat Feb 11, 2017 3:52 pm
Location: Vale of the White Horse, UK
Contact:

Re: New Feature : Planetary Live Stacking

#60

Post by admin »

Hi,

thanks for the feedback - will try to work out a good way to handle all the different use cases here.

cheers,

Robin
Post Reply