I am developing a moving xy-stage that will move the object underneath a
vertical camera. It's like areal photography on centimetre scale. Since
I move the stage, I know the coordinates of each exposure relative to
the whole. I think it should be possible to use these coordinates to
jumpstart finding the control points. But where / with what could I
start such a workflow?
You might want to use a telecentric lens for that operation. Especially if the subject is not perfectly flat with no depth.
Stitching SW is pretty good at aligning so unless your subject doesn't have unique details for the SW to find it's probably easier to let the SW find the control points than to devise a control point generator from XY movements. Seems it could be done though.
The object is a large format negative film, so perfectly flat and level. The mechanical challenge is to keep the object distance constant during the scan. Since the scan may eventually contain 30 exposures in one dimension, areas with very litle detail are just to be expected. Unless film grain will serve as sufficient detail.
Am 08.04.2014 14:22, schrieb iconoclastica:
> I am developing a moving xy-stage that will move the object underneath a
> vertical camera. It's like areal photography on centimetre scale. Since
> I move the stage, I know the coordinates of each exposure relative to
> the whole. I think it should be possible to use these coordinates to
> jumpstart finding the control points. But where / with what could I
> start such a workflow?
If you using PTGui you will be better off to use "Align to grid" feature
(under Project tab). This would require to shoot in a regular pattern.
More info on http://tinyurl.com/c78qx8y
To input coordinates directly the PapyWizard XML format would be good.
See same page further down.
I agree with Eric that your best bet is to use the "Align to grid" feature.
I have found that I get better stitching results by fine tuning the
horizontal and vertical overlap percent first while using the "Detail
Viewer" to look at the boundaries between images (make sure "Show Seams" is
selected). You can get pretty good visual alignment that way.
You would want to use the column and row overlap method. One caveat is that
the calculator is generating pitch and yaw angles for a pano head based on
the lens focal length and camera sensor size. If your camera sensor/lens
combination generates a field of view greater than 12 degrees, you'll
generate a panorama that's greater than 360 degrees in the horizontal
won't know the difference but PTGui will give you an error message. Telling
the calculator you used a telephoto lens (e.g., 400mm with a full frame
sensor) would solve that problem.
Why, thanks, these replies gave me the confidence to continue. The angle of view is 18+ degrees, the number of exposures in one particular dimension dependend on negative size and magnification. If the calculator doesn't handle that well I can write my own. The proof will be the trying but first I'll have to finish the object table and it's driver. Wouldn't have been worth the effort if the results couldn't be processed, but you convinced me it's worth a try.