New to Huglin looking for some help.

[nor s]

Hi,
  A little about me. I've been doing astrophotography for many years and occasionally want to image an object that  is larger than the field of view of my equipment. The obvious way of doing that  is to image the object in sections and then stitch them together to form one complete image. In the past I have been using  photoshop to manually match the segments and back grounds. A difficult problem at times. An acquaintance put me on to Hugin so I loaded up version 2012 and gave it a try on the, Pleiades. To my pleasant surprise it did a very good job of combining the 2x3 image grid into a single picture.

Later When I tried to open Hugin again, it requested the lens info twice and then quit. I tried it several more time even after rebooting the PC but still no luck. I decided to un install2012 and get the  latest from the website 2019.2 for 64bit WIndows10 PC.

When I first started the new version it complained about open GL not available. After rerunning it I got no more errors and loaded my 8 images 2x4 grid mosaic. Each image is 3020x1985 with an overlap of about 10%.  I selected the points automatically  but noticed there were no points is some of the images on the top and bottom  edge so I added them. 

Finally when I click the STITCH button it fails.


The  log is attached.

Any help would be appreciated.  If I'm using the wrong program to stitch these mosaics together perhaps someone can recommend one that would work better.

Thanks
 Nor

[dkloi]

Welcome to the group. It would help if you could provide the images. You should be able to stitch astronomical images together. I note on the log file:

Number of active images: 3

Output exposure value: 0.0

Canvas size: 6040x7940

ROI: (0, 0) - (6040, 7940)

FOV: 360x473

Projection: Equirectangular(2)  

The FOV is a bit big (473 deg vertically), this is probably causing the problem. The FOV of you images are probably too large, this can be manually set (by specifying the focal length and crop factor) but this may have been optimised to the wrong values, leading to the problem.

Once we can see the photos and your Hugin project file, we can tell you more.

[astro]

HI
   I've been playing around with Hugin 2019.2 and finally got it to work today. Yes the finished size is rather large since each segment is 3020x1985 which roughly translates to 6040x7940 as the completed 2x4 mosaic.  I've attached a finished mosaic and the screen capture of the stitcher tab that  I finally used.

Not really being familiar with this program I was just poking and hoping that  something would work. ONce I have it all figured out I'll put together a work flow.

Thanks for the reply.
Cheers
Nor

--
A list of frequently asked questions is available at: http://wiki.panotools.org/Hugin_FAQ
---
You received this message because you are subscribed to a topic in the Google Groups "hugin and other free panoramic software" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/hugin-ptx/UU4YWvD3O_M/unsubscribe.
To unsubscribe from this group and all its topics, send an email to hugin-ptx+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/hugin-ptx/988f780b-118f-4ebc-b28b-00730b637d3cn%40googlegroups.com.

[dkloi]

Stitch looks good. You can see in the stitch tab that the FOV is now a more reasonable 4 deg x 5 deg. The optimisation process seems to have come up with a more proper value for the lens HFOV.

I have made available example Hugin pano projects at:

http://www.dkloi.co.uk/?p=1490

http://www.dkloi.co.uk/?p=1501

http://www.dkloi.co.uk/?p=1518

They are for complete panospheres but many of the the basic concepts are unchanged for a partial pano and may be useful for gaining familiarity with Hugin.

[astro]

Hi Daniel,
   Wow you sure have that  process down pat. I will have a closer look at your steps and see how they will work on my stuff.

Cheers
 Nor

You received this message because you are subscribed to the Google Groups "hugin and other free panoramic software" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hugin-ptx+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/hugin-ptx/087d8f06-6b66-4d5a-b1f7-30156e69336cn%40googlegroups.com.

[nor s]

Hi Daniel,

     I've been trying to use HUGIN to stitch mosaics of the Rosette nebula and ended up with  a distorted image. Part of the nebula was missing and kind of squeezed together new the middle. The preview image showed that  some of the segments were turned slightly.   I have come to the conclusion that my problem is the fact that HUGIN corrects for all kind of effects such as curvature or perspective, etc which are common in terrestrial imaging using different lenses. However images shot with a telescope of objects in the night sky are for all practical purpose FLAT. 

It sure seem like HUGIN is a very powerful program, but does it have a simple flat stitch function that  will allow the stitching of images without turning or skewing the segments? I see that  there are all kinds of different options to use for stitching unfortunately I not familiar with any of them.

Your suggestion would be greatly appreciated.

Thanks

 Nor  

[Gunter Königsmann]

There are several tutorials for stitching murals or scanned images. I guess these should be near to your use case. Basically the first step is telling hugin that the camera has a very small field of view and that each image was made using a different lense (which tells the program it might have been made from a different location, too). Then you allow hugin to optimize the distortion to zero and the x and y positions the photos were shot from, but *not* the other parameters (scale of the images,...).

...but a specific mural/scanned image mode would be nice, admittedly...

--
Sent from my Android device with K-9 Mail. Please excuse my brevity.

[dkloi]

You should enable the advanced or expert modes where you have the option to control what parameters are optimised. You may want to optimised individual parameters, or certain subsets. If you know your effective horizontal field of view for each image, you can enter that value and disable its optimisation whilst allowing other lens parameters to be optimised. It would help if you could post examples or the source images so that we can see what is going on.

[khi...@umich.edu]

I've used Hugin to stitch a variety of astro photos; overlapping star fields are excellent for determining your lens parameters, as there are guaranteed to be no (detectable) parallax errors.  The challenge is to select enough star pairs in overlapping images - the automatic control point detection can't always tell which star is which so I generally end up adding and fine-tuning control points manually (current Hugin versions might be better, it's been a few years since I last tried this).  But with 20 or so control points between each image pair, and optimizing everything except translation (I use the Expert interface) I've routinely gotten the errors down to one pixel or less.

But it does help to start with something close to the correct FoV - telescopes in general don't tell the camera their focal length (or anything else) so the EXIF data doesn't give Hugin enough information...

[Klaus Foehl]

I want to second the recommendation to manually provide a good guess for the field-of-view. And at least initially to not include it as an optimisation parameter. In Photos > Optimize > Geometric select "Custom parameters", then an additional "Optimizer" tab appears, and and there with lens parameters you check b, d, e.

If things are ok, you may check v as well. But given that telescopes are strongly tele, this can lead to unwanted funny behaviour. The v parameter may wrongly deputise for higher distortion parameters that the hugin lens model lacks.

Better not use the parameters a and c. They are mathematically flawed. If there were more than one good distortion parameter in hugin, b that is, alignment errors could be even in the 1/10th to 1/100th pixel range.

Caveat: noise-filtering cameras might spoil your star positions (blackfield correction is ok). On a good image (I have not tested with stars), the hugin "Finetune" should be good to 1/5 to 1/10 pixels per individual Control Point.

To view this discussion on the web visit https://groups.google.com/d/msgid/hugin-ptx/c30dac0a-7691-4f05-b277-5b8bde62eee2n%40googlegroups.com.