How to improve the generation of a mosaic panorama

[Federico Arboleda]

Hi All,

I built an imaging station to photograph drill core on a camera slider: https://vimeo.com/368520148 The camera is moved on a single axis (diagram attached), known distances. The core tray remains fixed.  I am using PTGui to create a panorama of the resulting images.  I had to use the Align to Grid function because sometimes PTGui cannot find control points on its own

I tried this https://www.ptgui.com/support.html#6_5 and it did not work.  I think there is too much parallax as illustrated in the attached document.  

Questions:

1. Is there something I can add to the scene to improve the ability of PTGui to find control points?  Colored ribbon, shapes out of cardboard?

2. Is there something I am doing wrong to generate the panorama?  How can it be improved so less stitching defects are created?

Following  is a sample dataset: https://imago.sharefile.com/d-s3e0e90895da499d9

Many thanks in advance!

[Erik Krause]

Am 05.12.2019 um 22:15 schrieb Federico Arboleda:

> 2. Is there something I am doing wrong to generate the panorama? How can
> it be improved so less stitching defects are created?

It would have been far better to shoot from a single viewpoint, with a
longer lens from as far away as possible.

However, if the positions from which your camera shoots are exactly
repeatable you can use a template from a successful stitch without any
control points and optimization. All necessary adjustments to automatize
this via Batch Stitcher can be found on Project Settings tab.

And it would be better to shoot more images and use only the central
stripe in order to minimize parallax.

--
Erik Krause
http://www.erik-krause.de

[John Houghton]

On Thursday, December 5, 2019 at 9:15:59 PM UTC, Federico Arboleda wrote:

1. Is there something I can add to the scene to improve the ability of PTGui to find control points?  Colored ribbon, shapes out of cardboard?

You could create a duplicate set of images that have been adjusted in terms of brightness etc to reveal the detail for the purposes of facilitating the creation of control points. These can be replaced in the project by the original images before creating the final output panorama.

 

2. Is there something I am doing wrong to generate the panorama?  How can it be improved so less stitching defects are created?

I used the suggested method of assigning a large value focal length, and created control points only on the core samples. (If necessary, initially create a few control points manually to get a reasonable alignment. Then use the "Generate control points here" function to generate many more points targeted on the core samples automatically. See the Help for details of this function). My result is attached.

John