Help required - SGPro Framing & Mosaic Wizard

[alcol620]

Hi folks

Desperately in need of some advice/help.

I have been trying to get a 2 part mosaic of IC1848 Heart & Soul nebula, but the end result when pulling the 2 images together is two frames well out of alignment. Not sure why, I think I have followed the SGPro process using the Framing and Mosaic Wizard.

My process was as follows:

Opened the Wizard

  Typed in the Object as IC1848, did a fetch. Image arrived and I drew a rectangle around it having set rotate selection as 180 degrees to get the image where I wanted it, set camera tiles as 2 x 1. See the image attached with appropriate overlap.

   Clicked "Create Sequence" ( ticked "auto rotate or validate rotation on mosaic start" and Precision centering with plate solve")

The Sequence was produced with 2 parts IC1848-1 and IC1848-2 The coordinates for the 2 images shown in the sequence were (IC1848-1) 02 29 12.73s  60 06 19.70s   and (IC1848-2) 02 58 04.11s   60 57 52.41s

I started the Sequence and SGPro duly ran the Plate solver and asked me to manually rotate the camera until it aligned with the camera angle selection. A number of IC1848-1 images were taken.

SGPro then moved on to the second image in the sequence IC1848-2 and did a plate solve no requirement to rotate the camera as it was found to be in the correct position. SGPro then took images of IC1848-2. The 2 images obtained are attached for ingormation

However, when I combined these 2 images using Pixinsight's creating a rough mosaic with Star Alignment, they did not line up and indicated rotation.  See the attached image (It is a very rough image as on 1x300s subs have been used to quickly test whether the 2 images aligned).

I have repeated this process twice with the same result.

Very grateful for any feedback if anyone can see where I am going wrong.

Thanks

alcol

[gnomus]

Were the images of the Soul nebula taken on the same side of the meridian as the Heart?

I think that a meridian flip does all sorts of things unless the camera is orthogonal to the scope.  Personally, I would take great care to ensure the camera was as orthogonal as I could get it.  Then I would click off the button in the mosaic wizard that asks to manually rotate.

 

[alcol620]

Yes all on the same side of the meridian, images within minutes of each other. Can't get my head round where the rotation comes from!

[symmetal]

This confused me at first on my first mosaic, but it's because you have had to move the scope in RA to get the second frame. The composite image shown on the mosaic wizard is as it would appear in a telescope/camera combination having a wide enough field of view to have all your objects in the frame at once. On your two mosaic frames after taking the first frame's images you had to move the scope in RA to frame the second object. If you stand behind the scope looking along its length at your target and move the scope in RA back and forth between the two frames of your mosaic, the scope rotates clockwise and anticlockwise relative to you and the object you're looking at. Hence the second frame images are rotated relative to the first frame. Exaggerate the movement you have to swing the scope to show the effect. 

Moving in DEC to image your mosaic frames there is no rotation. You need a larger overlap between frames if you move in RA for your mosaic because of this rotation. You could rotate the camera as you re-frame after your RA movement to counteract the scope rotation but it's probably easier to rotate the images when you assemble your mosaic. 

[symmetal]

Sorry, I only half explained it above. The degree of rotation is dependent on the declination of the object you're imaging. If you're imaging at the celestial equator there is no rotation of your image if you move in RA for your mosaic. The further north (or south) of the equator the more pronounced the rotation is. The Heart and Soul is at a high declination so the rotation is quite high. If you took a mosaic near the pole the image could rotate 90 degrees as you framed your next panel by moving RA.

[alcol620]

5 hours ago, symmetal said:

Sorry, I only half explained it above. The degree of rotation is dependent on the declination of the object you're imaging. If you're imaging at the celestial equator there is no rotation of your image if you move in RA for your mosaic. The further north (or south) of the equator the more pronounced the rotation is. The Heart and Soul is at a high declination so the rotation is quite high. If you took a mosaic near the pole the image could rotate 90 degrees as you framed your next panel by moving RA.

Thanks for the explanation. makes a bit more sense now. However, I thought SGPro's Framing and Mosaic Wizard took account of all this to allow mosaic's to be a simple process. It never suggests that you need anything much more than 10 - 15% overlap to be able to combine images. With the sort of rotation seen here that % would have to be significantly higher and then when cropped large area of the image are lost.

[symmetal]

2 hours ago, alcol620 said:

Thanks for the explanation. makes a bit more sense now. However, I thought SGPro's Framing and Mosaic Wizard took account of all this to allow mosaic's to be a simple process. It never suggests that you need anything much more than 10 - 15% overlap to be able to combine images. With the sort of rotation seen here that % would have to be significantly higher and then when cropped large area of the image are lost.

The mosaic help file does say that the camera rotation angle is very important making mosaics and that if you don't have a mechanical camera rotator then the SGP Manual Rotator Tool can help you frame your mosaic panels. It doesn't say that an EQ mount will case mosaic panel image rotation without a camera rotator which I think it should. If you're imaging near the poles a couple of test images should be taken and assembled to see if more overlap is required (if you don't have a rotator), but for the majority of the sky the 15% default should suffice.

[alcol620]

4 hours ago, symmetal said:

The mosaic help file does say that the camera rotation angle is very important making mosaics and that if you don't have a mechanical camera rotator then the SGP Manual Rotator Tool can help you frame your mosaic panels. It doesn't say that an EQ mount will case mosaic panel image rotation without a camera rotator which I think it should. If you're imaging near the poles a couple of test images should be taken and assembled to see if more overlap is required (if you don't have a rotator), but for the majority of the sky the 15% default should suffice.

Thanks.  As you can see I used the manual rotator and was using a 15% overlap on the images and that wasn't enough. I thought that the Manual rotator did the same job as the mechanical rotator? Presumably using the TSAPO 60mm scope which gives  a wide FOV means that the RA rotation will be significantly greater than if using a scope/camera combination with smaller FOV? I was under the impression that SGPro Mosaic Wizard would take care of the rotation as when it plate solves it tells me how much to rotate the camera to, before starting to take the images?

[symmetal]

I presume you have ticked the 'Auto rotate or validate rotation on mosaic start' box on the Mosaic Sequence window of the Framing & Mosaic Wizard and selected 'Precision centering with plate solve' in the same window, and also selected 'Manual Rotator' in the equipment section of the 'Sequence' window (click the down arrow below Telescope to get the additional equipment options). 

When it performs a 'Centre on Target' before it starts imaging the panel, it will automatically loop by exposing and platesolveing the camera view to centre on the panel location obtained from the Mosaic wizard. If the rotator options are selected as in the previous paragraph it will also give you a camera rotation angle to perform and you have to manually rotate the camera this amount. It will keep doing this loop until the centre location and rotation angle are below a set threshold or you tell it to ignore the rotation error. In this case the rotation angle between your panels when you've imaged should be what the rotation angle was displayed, before you started imaging (1 degree for example).

Your two panels don't appear to have had the camera rotated between the two panels when you imaged them. The above 'Centre on Target' loop should have given you a rotation angle required before you began imaging the second panel.

A mechanical rotator will automatically rotate the camera to the correct angle during the 'Centre on Target' phase but if a manual rotator is selected as you have, you have to physically rotate the camera before each panel of your mosaic.

As you say a scope/camera combination with a wide field of view will require greater rotation between panels compared to a small field of view combination.

I hope this makes sense. 

[alcol620]

Thanks Symmetal

Very helpful, all much clearer now. I thought I had these settings set (although with all the settings required in SGPro it is possible that they were not). When I ran the sequence for the first image it ran plate solve and requested manual rotation of the camera, which I did, and it took the images. When it went on automatically to move to the second image in the sequence it plate solved again and it came up with a rotation error of less than 1 degree (within the tolerance I set so no manual rotation required) and went on to completed the solve and take the 2nd image. Hope to get out tonight and check all this more carefully.

I will report back - thanks again

[alcol620]

Hi folks

Just did a quick run off the 2 part IC1848, Set the sequence running, plate solve started and asked that the camera be rotated by 90 degrees, I had just completed a run on a different image, I manually rotated the camera, (several times until plate solve was happy that it was within the 0.5 tolerance that I had set. The sequence then took the requested images of IC1848-1 and slewed to the next part of the mosaic IC1848-2, plate solve started up and checked the position and reported that the camera was at 0.2" within the 0.5 tolerance set and required no rotation of the camera before it went on to complete the plate solve and SGPro then went on to take the images of IC1848-2

As far as I could see I had set all the required settings for plate solve to set the camera rotation position before taking images. 

Presumably for manual rotation there is no requirement to make entries in the SGPro Control Panel Manual Rotator setting?

Any ideas why the plate solver is not picking up the rotation in the second image (see the images posted earlier in this thread)?

[alcol620]

7 hours ago, alcol620 said:

Hi folks

Just did a quick run off the 2 part IC1848, Set the sequence running, plate solve started and asked that the camera be rotated by 90 degrees, I had just completed a run on a different image, I manually rotated the camera, (several times until plate solve was happy that it was within the 0.5 tolerance that I had set. The sequence then took the requested images of IC1848-1 and slewed to the next part of the mosaic IC1848-2, plate solve started up and checked the position and reported that the camera was at 0.2" within the 0.5 tolerance set and required no rotation of the camera before it went on to complete the plate solve and SGPro then went on to take the images of IC1848-2

As far as I could see I had set all the required settings for plate solve to set the camera rotation position before taking images. 

Presumably for manual rotation there is no requirement to make entries in the SGPro Control Panel Manual Rotator setting?

Any ideas why the plate solver is not picking up the rotation in the second image (see the images posted earlier in this thread)?

Checked the images this morning and the same result, with about 6 degrees of rotation not picked up by the plate solver. Presumably something I am missing?

[symmetal]

I put up with the rotation as my FOV is a lot less than yours so I never bothered to have it checked on my mosaics so my explanation above was what I thought it did on camera angle. It seems the mosaic wizard is not as sophisticated as I assumed it was. All it is doing when it checks the camera rotation angle is that the camera is aligned to the same angle relative to the equatorial grid as was selected when you defined the mosaic. On the Framing and Mosaic Wizard opening screen there is the rotate selection slider, which it seems determines the angle of the camera relative to the equatorial grid for each panel of the mosaic and not the mosaic as a whole. Which is why after it had asked you to rotate the camera on the first panel, it you don't touch the camera after that it will say all the other panels are correctly rotated. So the closer you are to the celestial pole the more rotation you get on each panel relative to its neighbour. If you took a 4 panel mosaic (2x2) centred on Polaris the panels would be rotated 90 degrees with respect to each other. It would take a bit more maths to work out the camera rotation angle relative to the adjoining panel as it depends on FOV of the panel and its declination but I wouldn't think it would be too difficult for the software to achieve. All you can do is increase the panel overlap for mosaics closer to the pole.  Sorry for giving you incorrect information on the above posts. 

I don't know what other software packages do about the rotation for mosaics, perhaps they are more sophisticated. 

[alcol620]

Hi

Many thanks for taking the time to explain in detail what is going on. I now have a much better understanding of the mechanics. I have posted a link to this discussion on the SGPro forum in the hope that the developers will have a view on how the Wizard functions and whether there is a need to modify it to take account of RA rotation between panels.

Thanks again for your detailed feedback - very helpful

Alcol

[symmetal]

No problem. I did check before on the SGPro forum and there were discussions about mosaics but the rotation issue didn't seem to be mentioned. It'll be interesting to see what the reply is. 

[daz]

Have been reading this with interest as I will be branching out into mosaics this year..

I read the posts on the SGP forum and followed the references to the various scripts.. the CatalogStarGenerator seems pretty useful, and if the MosaicPlanner does indeed calculate the rotation needed per panel, this would be i) useful and ii) annoying because it would mean manually adjusting the camera position...

I suppose you could image the panels that don't need rotation on one night and then the others on a subsequent night....

[symmetal]

The SGPro forum hasn't responded to Alcol's query it seems. 

Here's the rotation I get on what I've done so far on a 20 panel Heart & Soul mosaic imaging at about 1 degree FOV. This is a rough mosaic (quick stretch and rotation) of the Ha (16 subs of 240s at 2x2 binning, per panel, 15% overlap). Each panel ends up rotated about 2.4 degrees to its neighbour in RA. This is at dec 60 degrees. For a large mosaic, if you want to crop to a rectangular frame when finished you have to include a fair amount of sky above and below your target to avoid black holes in your image.

[alcol620]

On 20/08/2017 at 02:00, symmetal said:

The SGPro forum hasn't responded to Alcol's query it seems. 

Here's the rotation I get on what I've done so far on a 20 panel Heart & Soul mosaic imaging at about 1 degree FOV. This is a rough mosaic (quick stretch and rotation) of the Ha (16 subs of 240s at 2x2 binning, per panel, 15% overlap). Each panel ends up rotated about 2.4 degrees to its neighbour in RA. This is at dec 60 degrees. For a large mosaic, if you want to crop to a rectangular frame when finished you have to include a fair amount of sky above and below your target to avoid black holes in your image.

Interesting Symmetal thanks. No response from SGPro forum yet. However, I found this reference that suggests that SGPro does not allow rotational adjustment between mosaic sections, but some one at Pixinsight has written a script that will allow this to happen. (At least I think that is what it means?)

See 

Bhwolf

Oct '15

 

I saw Andres in the PixInsight forum has a really cool script for PI to do mosaic planning. One such feature is adjusting camera rotation to compensate for sky curvature, as you can see in his post here:

http://pixinsight.com/forum/index.php?topic=9057.msg58578#msg5857819

While it would require a rotator (in most cases) and may only be needed in widefield, it seemed like a good idea so thought I'd mention it here as a possible enhancement to the F&M wizard...

 

[alcol620]

On 20/08/2017 at 02:00, symmetal said:

The SGPro forum hasn't responded to Alcol's query it seems. 

Here's the rotation I get on what I've done so far on a 20 panel Heart & Soul mosaic imaging at about 1 degree FOV. This is a rough mosaic (quick stretch and rotation) of the Ha (16 subs of 240s at 2x2 binning, per panel, 15% overlap). Each panel ends up rotated about 2.4 degrees to its neighbour in RA. This is at dec 60 degrees. For a large mosaic, if you want to crop to a rectangular frame when finished you have to include a fair amount of sky above and below your target to avoid black holes in your image.

Interesting Symmetal thanks. No response from SGPro forum yet. However, I found this reference that suggests that SGPro does not allow rotational adjustment between mosaic sections, but some one at Pixinsight has written a script that will allow this to happen. (At least I think that is what it means?)

See 

Bhwolf

Oct '15

 

I saw Andres in the PixInsight forum has a really cool script for PI to do mosaic planning. One such feature is adjusting camera rotation to compensate for sky curvature, as you can see in his post here:

http://pixinsight.com/forum/index.php?topic=9057.msg58578#msg5857819

While it would require a rotator (in most cases) and may only be needed in widefield, it seemed like a good idea so thought I'd mention it here as a possible enhancement to the 

 

[alcol620]

Hi Symmetal and others.

No replies/feedback as yet on SGPro Forum on adjusting camera rotation to adjust for sky curvature despite posting in a couple of places. 

[symmetal]

Hi alcol,

Yes, odd that nobody replied. I think that the pixinsight forum entry you referred to is talking about the same problem that you have, but I wouldn't call it sky curvature. I would say sky curvature causes the same problem cartographers have in drawing maps, that of mapping a curved surface (the Earth) to a flat sheet of paper requiring the various projections that are used. Image distortion will increase the further you are from the central reference point. Unless you are doing a really widefield mosaic like 10 degrees FOV this distortion would be very small. The sky curvature error would be the same whether you were imaging at the celestial equator or at the pole. The sky is just a very big sphere in reality.

I found on my mosaic of the Heart & Soul you can align images left and right of the centre with rotation and translation and images above and below with translation and get the stars to align as they only have to match on one side but panels towards the corners cannot be aligned with just rotation and translation as they have to match two adjacent sides. You can align the horizontal stars pretty well or the vertical stars but not both at the same time. I have to use Registar to distort the panels slightly in order to match. I assumed that this was because of the sky curvature.

The large rotation problem we have is due to having to take long exposures using eq mounts. If the Earth wasn't rotating you could just use a standard tripod and pan left, right, up and down to get your panels and there would be no rotation error between the panels whether you're imaging towards the celestial equator or the pole. There would still be the sky curvature error though.

That's my take on it anyway. If anyone has other thoughts feel free to pitch in. 

Alan 

[alcol620]

Thanks Alan, very helpful insight. I am surprised at the lack of input from others who have taken wide field mosaics. Thanks for your interest.

Regards

Alec