Re: [PanoToolsNG] pseudo "scanning" a large map
- paul womack wrote:
> Bruno Postle wrote:Just a followup; I did a "test run" yesterday afternoon;
>>> and making a simple "tiled" set of photos.
>> If you use a pano-head and don't move the camera, then this is just
>> a 'normal' panorama, just set your output to 'rectilinear' and use
>> horizontal and vertical control points to straighten it up.
> Sounds like my life is going to be easier than I thought.
made a 4x6 images pano, at half res (so 1224x1632 per image)
on my home laptop (*).
I made some mistakes.
I used my wooden pano head at the end of the
horizontal column of my Benbo trecker tripod,
with the map on floor; the camera was around 4 feet
in the air.
Since the situation was so well controlled
(so I thought) I didn't allow much overlap.
In the first stitch I didn't do nearly enough control
points; somehow (mistakenly) I thought panotools
would magically know that this was a simple job;
but panotools only (of course) knows what it's
been told via control points.
Since the background of the map is so
uniform, minor changes in room lighting
have given me stitching errors. The map
was under 4 tungsten lights, but the variation
in light from window caused differences.
After "cramming in" enough control points,
the stitch went fine, as well it might using a 100mm
(35 mm equivalent) lens. Not much distortion...
I hope this is of interest or assistance to someone.
(*) 333Mhz pentium, 350 Mb RAM
- Sacha Griffin wrote:
> Yup! :-)Hmm. Then I lose the benefit of easy stitching.
> Been there. Lighting is key for mosaic scanning, even more so for very
> reflective materials.
> It's MUCH better to move the artwork, than the camera, keeping the lights
> static. Albeit maybe more of a pain?
> Anyway, it's probably close enough to be happy with. Congrats.
If I simply do this at night, when ALL the light
is artificial (and thus constant) I'll avoid
I then have a secondary task of either starting
acheiving UNIFORM light over the map, or post-touching
to achieve it.