×

INDI Library v2.0.6 is Released (02 Feb 2024)

Bi-monthly release with minor bug fixes and improvements

New User - A speed question

  • Posts: 20
  • Thank you received: 3
Hy

Cloudy skies, so I haven't had a chance to try it again and post my setup.

Peter
3 years 8 months ago #56808

Please Log in or Create an account to join the conversation.

  • Posts: 20
  • Thank you received: 3
hy:

It's been awhile, but I have been testing and experimenting with your suggestions.

So here's what I have seen using the Alignment module:

It takes 1 sometimes up to 2 1/2 minutes from "exposure done" to "image received" using the online solver. It typically takes from 90 to 160 sec for the solver to complete:

[2020-08-02T22:47:12.246 Eastern Daylight Time INFO ][ org.kde.kstars.ekos.align] - "Capturing image..."
[2020-08-02T22:47:12.271 Eastern Daylight Time INFO ][ org.kde.kstars.indi] - Canon DSLR EOS 77D : "[INFO] Starting 15 seconds exposure. "
[2020-08-02T22:47:34.074 Eastern Daylight Time INFO ][ org.kde.kstars.indi] - Canon DSLR EOS 77D : "[INFO] Exposure done, downloading image... "
[2020-08-02T22:48:57.199 Eastern Daylight Time INFO ][ org.kde.kstars.ekos.align] - "Image received."
[2020-08-02T22:48:58.024 Eastern Daylight Time INFO ][ org.kde.kstars.ekos.align] - "Solver iteration #1"
[2020-08-02T22:48:58.575 Eastern Daylight Time INFO ][ org.kde.kstars.ekos.align] - "Uploading file..."
[2020-08-02T22:50:10.007 Eastern Daylight Time INFO ][ org.kde.kstars.ekos.align] - "Upload complete. Waiting for astrometry.net solver to complete..."
[2020-08-02T22:51:38.925 Eastern Daylight Time INFO ][ org.kde.kstars.ekos.align] - "Solver completed in 161 seconds."
[2020-08-02T22:51:38.991 Eastern Daylight Time INFO ][ org.kde.kstars.ekos.align] - "WCS information updated. Images captured from this point forward shall have valid WCS."
[2020-08-02T22:51:38.993 Eastern Daylight Time INFO ][ org.kde.kstars.ekos.align] - "Solution coordinates: RA (21h 01m 52s) DEC ( 68° 14' 28\") Telescope Coordinates: RA (21h 01m 52s) DEC ( 68° 14' 55\")"
[2020-08-02T22:51:38.993 Eastern Daylight Time INFO ][ org.kde.kstars.ekos.align] - "Target is within 00° 00' 26\" degrees of solution coordinates."

If i use astrometry on the Pi3, takes from 40 sec to over a min from "capturing image" to "image received" and 1 1/2 min to 2 min to solve:

[2020-08-02T23:52:11.655 Eastern Daylight Time INFO ][ org.kde.kstars.ekos.align] - "Capturing image..."
[2020-08-02T23:52:11.765 Eastern Daylight Time INFO ][ org.kde.kstars.indi] - Canon DSLR EOS 77D : "[INFO] Upload settings set to client only. "
[2020-08-02T23:52:11.768 Eastern Daylight Time INFO ][ org.kde.kstars.indi] - Canon DSLR EOS 77D : "[INFO] Starting 15 seconds exposure. "
[2020-08-02T23:52:33.359 Eastern Daylight Time INFO ][ org.kde.kstars.indi] - Canon DSLR EOS 77D : "[INFO] Exposure done, downloading image... "
[2020-08-02T23:53:11.049 Eastern Daylight Time INFO ][ org.kde.kstars.ekos.align] - "Image received."
[2020-08-02T23:53:11.052 Eastern Daylight Time INFO ][ org.kde.kstars.ekos.align] - "Solver iteration #1"
[2020-08-02T23:53:59.464 Eastern Daylight Time INFO ][ org.kde.kstars.ekos.align] - "Starting remote solver..."
[2020-08-02T23:53:59.641 Eastern Daylight Time INFO ][ org.kde.kstars.indi] - Astrometry : "[INFO] Astrometry solver is enabled. "
[2020-08-02T23:54:13.031 Eastern Daylight Time INFO ][ org.kde.kstars.indi] - Astrometry : "[INFO] Solving image... "
[2020-08-02T23:55:45.702 Eastern Daylight Time INFO ][ org.kde.kstars.ekos.align] - "Solver completed in 93 seconds."
[2020-08-02T23:55:45.728 Eastern Daylight Time INFO ][ org.kde.kstars.ekos.align] - "Solution coordinates: RA (21h 37m 45s) DEC ( 57° 28' 48\") Telescope Coordinates: RA (21h 36m 37s) DEC ( 57° 29' 21\")"
[2020-08-02T23:55:45.728 Eastern Daylight Time INFO ][ org.kde.kstars.ekos.align] - "Target is within 00° 16' 53\" degrees of solution coordinates."
[2020-08-02T23:55:45.743 Eastern Daylight Time INFO ][ org.kde.kstars.ekos.align] - "Syncing to RA (21h 37m 45s) DEC ( 57° 28' 48\")"

What I find interesting is that when I am capturing images in the CCD module locally to the Pi3, it takes only about 10 sec from image captured until I can view them on Pi3.

I was thinking that the difference is that the alignment module is sending a copy to Ekos running on my laptop as well as capturing it on the Pi, thereby making the process much longer. If I'm using Astrometry for solving on the Pi, why would it send an image to EKOS on laptop?

Thanks for all your help

Peter
3 years 7 months ago #57693

Please Log in or Create an account to join the conversation.

  • Posts: 437
  • Thank you received: 31

Replied by Paul on topic New User - A speed question

Peter,

The solver works faster when you are close to where it thinks it is.

In your case I can only see an iteration #1, it would be interesting to see what happens with an iteration #2 which should be faster.

You will need to ask for more accuracy to do this.

Another thing is to ensure you have the correct focal length set.

Paul
3 years 7 months ago #57697

Please Log in or Create an account to join the conversation.

  • Posts: 1208
  • Thank you received: 559
Peter,

Try a couple things.

1- Try downsampling your image. I tend to downsample by 3 I believe, though not sure. See the "Bin" dropdown, which is the 4th to bottom line on the Align page.
In the Options (bottom right) of that page, under solver options, you probably want "Use Position" "Auto Update", and also "Downsample, 2, Auto". Also "Use Scale" and "Auto Update" that too.
Keep on using the local version. Using the online version is slow in my experience.

2- Also try to use ASTAP (see www.hnsky.org/astap.htm to install)
Either one should work faster than what you have.

If these don't work, can you send a picture of your "Solver Options" pop-up, and your Align Page?

Hy
3 years 7 months ago #57698

Please Log in or Create an account to join the conversation.

  • Posts: 20
  • Thank you received: 3
Hy:

I've attached a screenshot of the Align module and solver options.

The bin option on the Align module is grayed out. Down sample is set to 2. So I'm consisting getting about 70-90 sec for a remote solver iteration. As you can see last night I was able to solve in two iterations. I still think the issue is the downloading of the image to my laptop running KStars/Ekos. Since it is solving the image remotely on the Pi, why does it download the image to my laptop? When I take an image sequence and capture the image remotely on my Pi, it only takes about 10 sec from completing the image capture before I can view the image on the Pi.

Thanks for hanging in with me to see what improvements I can make. So far everything else seems to be working ok.
3 years 7 months ago #57778
Attachments:

Please Log in or Create an account to join the conversation.

  • Posts: 20
  • Thank you received: 3
Paul:

I think I have everything set correctly. As you suggest, it's really making sure you're close to where you think it is. I have used the iOS app and the Wifi dongle to do a two star alignment on the EQ6R before running Kstars/Ekos to choose targets. Last night having carefully done that, I was able to get align to work in 2 iterations.

Peter
3 years 7 months ago #57780

Please Log in or Create an account to join the conversation.

  • Posts: 1208
  • Thank you received: 559
My reaction to your screenshots:

1- You're using the remote astrometry.net. You should be using the local one.
2- Binning seems to be disabled for you on the left image, 2 below the exposure drop-down. I bin 3x3. However, I have a monochrome CCD, not a DSLR. Sorry, not familiar with using Canon DSLRs for Astro, and don't know if that's why you can't bin there. Perhaps you should up the binning on the right window next to Auto?
3- I only expose 5s or so, with my f/5.6 scope
4- I have removed the astrometry.net index files that are way too detailed for my needs.

See the screenshots of my setup (though I've since moved to ASTAP, I'm showing you my local astrometry.net settings.

3 years 7 months ago #57811
Attachments:

Please Log in or Create an account to join the conversation.

  • Posts: 1119
  • Thank you received: 182

Not wanting to contradict Hy, just telling you what works great for me:

1) Astrometry.net use LOCAL, NOT online, to solve. I have all but the 3 largest file sets, including the Tycho catalogue, installed, total size 5.6 GB. That is enough from wide-field to 1600 mm focal length.

2) ASTAP never worked reproducibly for me. If it worked, it was FAST, but it often failed. Astrometry.net never fails.

3) Solving NEVER takes longer than 15s with Astrometry.net when solving locally on the Pi4. I can live with that easily. I usually only image one target/night, so only about 2 min total alignment time, including meridian flip.

4) FORGET the Pi3. Unbearably slow. I used it, just like Hy, until the very moment the Pi4 arrived. Ever since then, it has become the server for my 3D printer.

5) I (still) boot from the SD card, but star catalogues are installed on an SSD and I am saving images to the SSD. I also have a swap partition on the SSD (only required for hi-thread compiling).

Jo
3 years 7 months ago #57837

Please Log in or Create an account to join the conversation.

  • Posts: 20
  • Thank you received: 3
Hy:

You're right. I attached the wrong screen grab. I have been using the copy of Astrometry on the Raspberry Pi. This speeds up the solve by about 10 sec. In looking at the timestamps in the logs, most of the time taken up in solving is uploading the image back to my laptop. Still not sure why remote solving would cause Indiserver to upload a captured image, when I'm solving on the Pi.?

Jo;

I just received my Raspberry Pi4, so I'll be upgrading so I'm sure local solving will be faster. I had not considered using an SSD. Something I need to look into. Also, as I mentioned above, most of the elapsed time for an alignment iteration is taken up with uploading the image from my camera to my Windows laptop running KStars/Ekos. Still not sure why the software does that, when solving is done entirely on the Pi. If it didn't do that it would be a lot faster.

Thanks to both of you for offering good suggestions and ideas.

Peter
3 years 7 months ago #57957

Please Log in or Create an account to join the conversation.

Time to create page: 0.933 seconds