SEEING MONITOR multiple area of ​​interest

[aquariusCZ]

I have been photographing the moon for 20 years. During that time, the photography technology changed three times.

1 / scanning single frames in the infrared range
2 / transition from single frames to shooting video sequences - REGISTAX, AS3
3 / SEEING MONITOR - SharpCap. Selection of the best images.

The combination of 1-3 allows you to get nice pictures even with poor seeing.
For the price of long-term monitoring of the lunar landscape. Sometimes I take one place for 20 minutes and MONITOR often throws 50,000 shots.

When using SEEING MONITOR, it is necessary to ensure the highest possible FPS cameras.
That's why I choose a smaller area of ​​interest. (5% of total pixels)

seeing_m_1.jpg (87.35 KiB) Viewed 2063 times

However, with some seeing, MONITOR saves excellently sharp images of the crater, but the surroundings are out of focus.
Unfortunately, as the area of ​​interest increases, the FPS decreases and I lose potentially nice images.

seeing_m_2.jpg (91.79 KiB) Viewed 2063 times

The solution would be to deploy a multiple area of ​​interest (2-5 small areas distributed over the image area.)

seeing_m_3.jpg (93.45 KiB) Viewed 2063 times

The image would be saved if the seeing quality was better than the set trigger level in all areas at the same time. The high FPS is maintained and the sharpness of the entire image area is improved.

[admin]

Hi,

What camera you using an how much does the FPS drop as you make the seeing Monitor area larger? Are you using seeing triggered or seeing filtered capture? For the triggered capture I would have expected the capture FPS to be unaffected (or at least largely unaffected). Overall, if there's a way to fix the performance so that multiple areas are unnecessary, I prefer to do that.

Cheers, Robin

[aquariusCZ]

Hi,
I use the QHY5III290M camera at RAW8 and adjust the gain and exposure to get around 80 fps.
When scanning the Moon, I use the FILTER CAPTURE mode.

You are right that physical FPS cameras do not decrease with increasing area of interest. I'm sorry, I described the idea inaccurately.
The FPS SAMPLE RATE decreases. This means that the SEEING MONITOR also stores unchecked images, which can be damaged by fast seeing. FPS SAMPLE RATE drops to 25% compared to FPS cameras with a large area of interest.

Cheers, Tomas

[admin]

Hi,

Okay, that makes more sense now. For reference, what CPU are you using? Obviously there will be a difference here between using one of the latest core i9 processors compared to using a low-power CPU.

Cheers, Robin

[aquariusCZ]

Hi,
I am using an Intel i5-3230M CPU @ 2.60GHz processor, cores: 2.
The optimal shape of the area of interest will probably be a long narrow rectangle for quick seeing.
This shape ensures SAMPLE RATE = FPS cameras for less powerful PCs and at the same time covers more distant parts of the image.

seeing_m_4.jpg (84.58 KiB) Viewed 2006 times

Cheers, Tomas

[admin]

Hi,

Thanks for the additional info. They look to be some scope for improving performance in the code related to this. With the versions of SharpCap that you have now you could try turning the noise reduction down to zero which may give you a performance boost of the order of 10 to 20%.

One thing I'm wondering about is whether it would be valid to only consider every fourth pixel (for instance only pixels with even X coordinates and even Y coordinates) when evaluating the seeing. This would probably increase speed by a factor of very nearly four. In most cases I would expect people doing this sort of imaging to be over sampling (many pixels per arc second), so I would expect any great loss of accuracy in the monitoring functionality. This is particularly useful for colour cameras because if you take only the values of every fourth pixel in this way you can ignore the requirement to debayer the image before processing.

Cheers, Robin

[aquariusCZ]

Hi,
checking for only 1/4 pixel is a good idea.
To take advantage of the telescope's resolution, I capture the moon at 5 pixels / km.
Plato = 1/4 image width for QHY5III290.

Heretical idea:
Tighten the SEEING MONITOR to the end, like a LIVE STACK.
Divide the image into 100-300 areas and make a selection in memory by seeing, perform ALIGN and STACK and fold into one image.
There would be no need to work with annoying large videos files.
In addition, REGISTAX and AUTOSTAKKERT have not been updated in years.

Cheers, Tomas

[admin]

Hi,

I've had people try to persuade me to add video stacking capabilities before to be honest I'm not doing that for two reasons – one is that I don't have any time to put into it right now and the second is that I know it's a very complicated field and I think it takes a long time to catch up with those applications that you mentioned even if they haven't been updated in a while.

I will certainly be looking into ways I can improve the performance of this functionality in SharpCap 3.3

Cheers, Robin

[aquariusCZ]

Hi,
thanks, as I wrote, a heretical idea.

Will SAMPLE RATE increase in the current version?
Cheers, Tomas

[admin]

Hi,

I'm afraid I'm no longer adding that sort of improvement or any new features to SharpCap 3.2. The only things going into SharpCap 3.2 now are SDK updates for manufacturers and important bug fixes. If I don't stick to that rule then SharpCap 3.3 gets nowhere!

Cheers, Robin