ASIair Plus plate solving accuracy..?

[Flappa30]

I got first light with my ASIair Plus (AaP) last week. How do others find the accuracy of the plate solving..? I captured images of the same target on two consecutive nights, but found that the framing was very different on the second night. Nothing on my setup was moved or changed between sessions. The plate solving only ever seems to do a single correction, so the accuracy must be pretty loose. Cuiv The lazy geek has mentioned this in one of his recent YouTube videos. I haven't tried using a previous image (say from the night before) to see if this is any better. Has anyone else tried this, does it work..?

Now that I know about this I will give it a try myself.

Does it just use the existing images coordinates instead of plate solving..?

Like others I have been using NINA for a while, and find the simplicity of the AaP appealing, but if the repeatability isn't there for multi-session imaging I'm not sure of its use beyond beginner level imaging.

[Richard_]

I believe that the AAP uses a different coordinate system to the likes of Stellarium and Telescopius. One uses Jnow and the other uses J2000, I forget which way around. Simply put, if you use the RA/Dec coordinates directly from the aforementioned software/website and plug into AAP, you're not going to be directly on target. 

In AAP, if you go into the plan mode you can import coordinates from telescopius which then be transposed into the other coordinate type for AAP. This approach mostly work fine but as you can imagine, it's a bit of a faff to set up for every target. I dither my images so I tend to always use the first image so that all subsequent images have a better chance of overlapping. 

On the other hand, using images from previous sessions work great for me. Open up the image on AAP, plate solve and "go to" from there. It doesn't use the coordinates in the FIT header, but rather plate solves the image. 

Recently, I made a multi night, multi panel image of the Veil Nebula. The overlap of the same panel night to night wasn't bad at all, and because I didn't rotate my camera, the framing was very close each time. When I get home, I'll try and dig out the master file for each panel which should hopefully illustrate this! 

[Richard_]

OK, I actually remembered to take a look at this! I have some examples of using GoTo by using the "Plan" mode and also from plate solving a previous image and slewing to the coordinates. I also have examples from shorter focal length (250mm) and longer focal length (780mm). I'd be really interested to hear your thoughts on this (sorry for the long read!).

Short Version

For the shorter focal length, I used "Plan" mode for my Veil Nebula mosaic and it's safe to say that over three separate nights spanning a few weeks, there is no noitceable shift in image position. It's safe to say that the ASI Air Pro (AAP) is very repeatable for this focal length.

For the longer focal length, I framed my target one night (Heart Nebula, specifically Melotte 15) and performed an "Autorun" sequence. On the next night, I viewed the first image from the previous night, hit "GoTo" which plate solved the image and then slewed to the target. The AAP managed to slew to the same spot as the previous night, but I suspect I made an error with the amount of dither so it looks like there is more variation in overlapping of images. That being said, my master light doesn't seem to indicate two differently framed objects (i.e. inaccuracy in plate solving), so I think it's safe to say that plate solving and imaging this way is also repeatable at this focal length.

 

Veil Nebula

Captured using my Redcat 51 (250mm focal length) and ASI533.

I've uploaded a screenshot from PixInsight which shows the following:

• Left side shows the master light image for each of the four panels of my mosaic
• Right side shows the master light created after registering and merging the four panels

Each panel was imaged over multiple nights, which I've summarised below. I created a 2x2 mosaic plan in Telescopius with ~50% overlap and imported the RA/Dec into the ASI Air Pro "Plan" mode in the app. I repeated this plan for each of the nights stated below. The camera rotation was not changed over the whole duration, so there are no rotation errors in the images.

• Pane 1
  • 11th October 2021
  • 01st November 2021
• Pane 2
  • 11th October 2021
  • 21st October 2021
  • 02nd November 2021
• Pane 3
  • 12th October 2021
  • 01st November 2021
  • 02nd November 2021
• Pane 4
  • 21st October 2021
  • 01st November 2021
  • 02nd November 2021

As you can see, there is a dark border around some of the edges for each panel. When PixInsight registers and overlaps the individual subs, there are regions where there is poor overlap, which is denoted by the darker borders. These borders could be caused by, for example, incorrect camera rotation night-to-night, dithering between subframes or different RA/DEC co-ordinates between imaging sessions. Pane 2 appears to have the thickest looking border on the right hand side. For my 3,008 x 3,008px image, the dark border on the right hand side accounts for 78 pixels, or approximately 2.6% of the width of my image. I find this perfectly fine!

The merged mosaic on the right hand side now shows misalignment between all four panes caused by differences in camera rotation (e.g. my camera was not perfectly square between panels). Again, the regions where there is little overlap is very small and I was really happy with the end result considering it was my first mosaic.

 

Heart Nebula

Captured using my WO FLT 120 (780mm focal length) and ASI533.

I've uploaded a screenshot of my master light for the Heart nebula which was imaged over the following nights:

• 21st November 2021
• 22nd November 2021

For the first night, in the AAP app I searched for Heart Nebula (SH2-190) and hit "GoTo". I then performed an Autorun on this target as Melotte 15 was dead centre of the frame so I didn't need to make any adjustments. On the second night, I opened up the first sub from the previous night, hit "GoTo" which prompted a plate solve followed by slewing to the target. Melotte 15 was dead centre again, so I repeated the Autorun program. Looking at the registered image, the darker borders are much more prounounced compared to the images shown in the Veil nebula but there doesn't seem to be an obvious "Night 1 versus Night 2" shift.

The darker border now measures 212 pixels, ~7% of the sensor width. While I was "Blinking" through my subs I noticed that the stars were shifting much more than they were compared to when I blink through subs from the Redcat. I suspect this may be due to the amount of dithering I set. For the Redcat, I set it to either 15 or 30 pixels and I did not change this for the FLT120, so I wonder if the dither is way too high for the longer focal length of the FLT120. If this is the case, I reckon I can reduce the dark border to that shown in the Veil nebula by reducing the amount of dither for the FLT120.

Either way, for first light with the FLT120, I was really happy with alignment of my subs and repeatability of the GoTo.

[wimvb]

Can't you set the accuracy you want? In Ekos, I have set it to 30 arcseconds, or about 32 pixels. I'd imagine that the ASIAir allows something similar

[Flappa30]

Richard, thank you your reply on this. I found just using the search and then go to on two consecutive nights meant that I had a position shift of about 20% of the image.  My scope has a FL of 430mm. I’ll try the using a previous image technique, which should hopefully improve this. I’m used to using NINA, where the repeat accuracy is great time after time. It may do 2 or sometimes 3 adjustments to achieve this. AAP seems to be limited to one adjustment only.

Wim, no you can’t set the required accuracy, that’s the problem. The pre-set value (that you can’t see) would seem to be very loose, probably several arc minutes. It works fine, and is good enough to roughly centre the target as a one off, but not for accurate repeatability over several sessions.

I hope that this gets an upgrade as the basic functionality is clearly already there. This is pretty fundamental for deep sky imaging targets. Come on ZWO, this falls short of what we need/expect. Allow us to set the accuracy and the number of tries to get there.

Edited November 30, 2021 by Flappa30

[wimvb]

26 minutes ago, Flappa30 said:

Wim, no you can’t set the required accuracy, that’s the problem. The pre-set value (that you can’t see) would seem to be very loose, probably several arc minutes. It works fine, and is good enough to roughly centre the target as a one off, but not for accurate repeatability over several sessions.

That is a very big minus for the ASIair. I guess that the app tries to be smart and set the tolerance according to sensor and focal length used. Otherwise long focal lengths wouldn't be possible, and very short focal lengths would run into mount accuracy problems.

[Scooot]

2 hours ago, Flappa30 said:

Richard, thank you your reply on this. I found just using the search and then go to on two consecutive nights meant that I had a position shift of about 20% of the image.  My scope has a FL of 430mm. I’ll try the using a previous image technique, which should hopefully improve this. I’m used to using NINA, where the repeat accuracy is great time after time. It may do 2 or sometimes 3 adjustments to achieve this. AAP seems to be limited to one adjustment only.

Wim, no you can’t set the required accuracy, that’s the problem. The pre-set value (that you can’t see) would seem to be very loose, probably several arc minutes. It works fine, and is good enough to roughly centre the target as a one off, but not for accurate repeatability over several sessions.

I hope that this gets an upgrade as the basic functionality is clearly already there. This is pretty fundamental for deep sky imaging targets. Come on ZWO, this falls short of what we need/expect. Allow us to set the accuracy and the number of tries to get there.

Do you know that if you press goto again it will attempt to recenter the target. I’ve found that by doing this I can tweak it so its very close to dead centre if I want to after pressing it 2 or 3 times. That’s with my 375mm focal length. 

[Flappa30]

Scooot, thank you that’s something else to try, pressing go to again forces it to do another plate solve. If this actually works, why would it refine the position if the first plate solve got it to within the pre-set accuracy..?

It’s almost like it wants to do more plate solving, but isn’t allowed to…

[Scooot]

11 minutes ago, Flappa30 said:

Scooot, thank you that’s something else to try, pressing go to again forces it to do another plate solve. If this actually works, why would it refine the position if the first plate solve got it to within the pre-set accuracy..?

It’s almost like it wants to do more plate solving, but isn’t allowed to…

I can’t answer that. What happens is my mount moves a minuscule amount as if you were to nudge it.

I discovered it did this when I was trying to align the camera along an axis. I’d move the star off centre a bit with the mount and then rotate the camera slightly, take another shot, to see if the star had moved nearer to the axis I was aiming at. I’d then press goto to get the star centred again and repeat if necessary. 

[Richard_]

I think this is the inaccuracy issue being discussed, that the first plate solve says "good enough" and carries on, instead of continual plate solving until it hits a certain criteria like wimvb says. What might be acceptable for 200mm wide field could be terrible for 1,000mm+ deep space! 

I'll see how I fair with my FLT120 if I plan to do multiple nights on the same object. I reckon repeat plate solving for an Autorun routine on 1 object is fine, but not so good for multi object plans or if you perform a meridian flip during the imaging sequence. 

Edited December 1, 2021 by Richard_
Correct my bad phone typing grammar and add in comment on meridian flip