Milkyway multi image Pano problem

[johneta]

Hi
I'm hoping someone can give me advice on milkyway multi image panos.
I'm trying to do a few multi image panos for high resolution.
Most panos I try always end up being narrower at the top-
As shown in the image.
I used Hugin in this image and i tried all the different projections and drag it around to try and straighten the image.
When it does start to straighten, then the milkyway at the top of the image, gets very stretched and too wide.
While the bottom part of it looks the right width.
To try and get the milkyway the correct width at the top, I always end up with a tapered image as show.
I can mess around in photoshop to warp it etc. but I would like to understand why it is happening.
This image is about 6 images wide x 6 images high. (around 36 total) All shot with a 6d in landscape orientation with a 58mm lens.
Camera was about horizontal on bottom row and pointing up by about 50 degrees on last row

[maw lod qan]

Even with the taper,

Wow!

[almcl]

The effect is sometimes referred to as the keystone effect and is often noticed when tilting a camera to get tall buildings in frame (lots of web articles on the subject). 

Not sure of the best way to counter it in Photoshop as I don't use it, but there should be 'straighten and crop' function somewhere?

Rather than Hugin, if you can find a copy, Micosoft's ICE does a better (and quicker) job, particularly with straightening things.

Here's what my image editing software did to your image:

The tree on the left still isn't quite vertical though.

Edited March 15, 2022 by almcl

[johneta]

Thanks almcl. I'll have a look at some of those articles.
I've got an old copy of ICE on an old laptop. I'll try it again. I remember thinking it wasn't that good, but that was a few years ago, so I'll have another look at it with fresh eyes.
Thanks for having a go at straightening, it looks good.

I also thought a bit more about taking the images, and realized as you point close to vertically up with the camera and pivot on the pano base, then you are really just pivoting round that point and get a small circle of view straight up.
Whereas when cam is horizontal it does a big sweep left and right as it pivots on the pano base.
Therefore you are recording less "width" at the top of the image. Thus its tapered at the top of a 'square' image.

I might experiment with a small angled wedge under the pano rotating bit.
This means I would have to effectively point the camera 'down' by a small angle relative to the pano base in order to get the horizontal land images. And then the overhead images would be at less up angle relative to the pano base, and therefore have more 'sweep' on the top row of images and therefore will capture more width, which will make the overall image closer to square.
I'll post how I go with this

[almcl]

One thing I found with ICE is that if it won't link all the images at the first attempt, linking just two or three and, after re-naming them so that they're in the right order, joining the new images together often worked.

My 6 panel mosaic of the Rosette nebula was constructed by this method.

Edited March 15, 2022 by almcl

[johneta]

Ok thanks.  Nice, that one done with your SW200P in your gear list?
I use an Orion 8" F3.9 and find the Rosette just fits awkwardly in the frame but I would like a bit more space around it like you have there.
Its a bit hard to get down here in NZ -  lowish in the sky in summer when my sky's are cloudy for months

[Elp]

If you transform > shear the bottom two corners toward the middle (keeping them at the bottom of the image vertically) so the left and right edges are vertical does it look as you've described?

[almcl]

11 hours ago, johneta said:

... that one done with your SW200P in your gear list?

Yes, indeed, with the Canon 700d, luckily (and unusually) I got more than one clear night. 

I guess things are similar on South Island although perhaps the Otago peninsula (if you're near there) doesn't suffer as much light pollution?

[johneta]

36 minutes ago, almcl said:

Yes, indeed, with the Canon 700d, luckily (and unusually) I got more than one clear night. 

I guess things are similar on South Island although perhaps the Otago peninsula (if you're near there) doesn't suffer as much light pollution?

Yes I'm lucky light pollution is low, as seen from the milkyway image above . Thats basically out my back door

[johneta]

9 hours ago, Elp said:

If you transform > shear the bottom two corners toward the middle (keeping them at the bottom of the image vertically) so the left and right edges are vertical does it look as you've described?

Hi, if I'm understanding you correctly,  I haven't done exactly as you mention, but I have tried various transform tools and others in PS, and I can get it to look OKish but never quite right.
As I allude to above, I think I have figured that I just haven't got enough images at a high angle. This is to do with how the camera sweeps on a standard 3way camera tripod head. when aiming up at a high angle (around 50degs).
I basically need a couple more shots of width up there for there to be enough information to match the bottom of the image which is swept at basically a horizontal angle (0degs).
When I study the stitched pano along the top row particularly, I can see the stars are getting stretched as the software tries to add width to the top of the image.

Ive decided to mount the pano rotating bit on the camera plate of the tripod so that the pano sweep goes in big arcs as the head is tilted upwards for the top rows.
That's the theory, and I will post how I go.

have a look at this if your interested in the Panorama Milkyway concept. Some handy tips. https://www.lonelyspeck.com/how-to-shoot-large-format-astrophotography-panoramas/comment-page-5/#comment-321114

The main difference with his image and mine I think is that he hasnt gone very high in upwards angle (Alt) and therefore not as much narrowing at the top as I'm getting

Edited March 17, 2022 by johneta