No control points found automatically in panoramas with a somewhat low contrast
Tom
panoramas. PTGUI is really a great software tool for stitching
Gigapixel panoramas.
In my current project (an alpine panorama made of 640 single photos)
PTGUI managaged one half of the panorama very well, in the other half
(with a somewhat lower contrast due to backlight) PTGUI didn't find
many control points automatically. The images in which no control
points were found have a lower contrast, showing mountains at far
distance and forest: the panorama is very similar to that one - taken
from a neighboring mountain: http://gigapan.org/gigapans/59678/ , but
has a much higher resolution. Basically the image quality is not so
bad, there a many features I could use for putting control points
manually.
I tried to solve the problem by selecting overlapping images manually
and applied "Generate control points for selected images". Mostly
PTGUI doesn't find any control points, even though the image quality
is not really bad. Does anybody know a solution for making the
automatic control point selection working better?
Best regards
Tom
Hans
On Nov 17, 8:02 pm, Tom <thomas.nai...@googlemail.com> wrote:
> I am using PTGUI (currently 9.0 beta 5) for stitching Gigapixel
> panoramas. PTGUI is really a great software tool for stitching
> Gigapixel panoramas.
> In my current project (an alpine panorama made of 640 single photos)
> PTGUI managaged one half of the panorama very well, in the other half
> (with a somewhat lower contrast due to backlight) PTGUI didn't find
> many control points automatically. The images in which no control
> points were found have a lower contrast, showing mountains at far
> distance and forest: the panorama is very similar to that one - taken
> bad, there a many features I could use for putting control points
> manually.
>
> I tried to solve the problem by selecting overlapping images manually
> and applied "Generate control points for selected images". Mostly
> PTGUI doesn't find any control points, even though the image quality
> is not really bad. Does anybody know a solution for making the
> automatic control point selection working better?
images with good controlpoints first.
After you have a perfect correction for them run the Generate
controlpoints for all images again.
Hans
Tom
what you suggest is generally a good idea. I just tried to stitch a
small section (just 20 photos) of the panorama, the problem is still
the same: in most of the images PTGUI doesn't find control points
automatically. To me it seems that PTGUI requires a very good contrast
for automated control point detection.
Erik Krause
> in most of the images PTGUI doesn't find control points
> automatically. To me it seems that PTGUI requires a very good contrast
> for automated control point detection.
If nothing else helps (and you don't want to pick CPs by hand) you can
replace those images temporarily with a higher contrast version.
--
Erik Krause
http://www.erik-krause.de
Hans
On Nov 17, 9:23 pm, Tom <thomas.nai...@googlemail.com> wrote:
> Hans,
> what you suggest is generally a good idea. I just tried to stitch a
> small section (just 20 photos) of the panorama, the problem is still
> the same: in most of the images PTGUI doesn't find control points
> automatically. To me it seems that PTGUI requires a very good contrast
> for automated control point detection.
Using Raw in Ptgui is for many reasons a very bad idea.
I just made a test and posted this at panoguide.
http://www.panoguide.com/forums/qna/8699/
Hans
Tom
with Silkypix) and I had the problems described above. Then I tested
with RAW - even worse. Finally I converted the raw images to jpg - not
so bad. In the small section that I tested (20 images) all but one
image (aside from the ones showing sky only) have been stitched
correctly. At the moment I am converting the 640 images to jpg. I'll
report results here tomorrow.
Tom
On Nov 17, 11:10 pm, Hans <hans...@gmail.com> wrote:
> On Nov 17, 9:23 pm, Tom <thomas.nai...@googlemail.com> wrote:
>
> > Hans,
> > what you suggest is generally a good idea. I just tried to stitch a
> > small section (just 20 photos) of the panorama, the problem is still
> > the same: in most of the images PTGUI doesn't find control points
> > automatically. To me it seems that PTGUI requires a very good contrast
> > for automated control point detection.
>
> You do not by any chance use Raw as source?
> Using Raw in Ptgui is for many reasons a very bad idea.
>
Ken Warner
of image information in the form of increased dynamic range.
However --
With the three compact cameras I've had (Fuji e550; Fuji e900; LX3)
I have been unable to convert the raw images and get an image better
(sharper, better balance of color and contrast and less bloom and noise)
than the jpg image the camera(s) itself makes.
This probably is not the case with a good DSLR and/or ACR -- which
I don't have. But I find it much easier to shoot jpg and convert
to tiff as the first operation.
But this is just me...
Roger D. Williams
wrote:
> I like raw capture because it does make available a greater amount
> of image information in the form of increased dynamic range.
>
> However --
>
> With the three compact cameras I've had (Fuji e550; Fuji e900; LX3)
> I have been unable to convert the raw images and get an image better
> (sharper, better balance of color and contrast and less bloom and noise)
> than the jpg image the camera(s) itself makes.
>
> This probably is not the case with a good DSLR and/or ACR -- which
> I don't have. But I find it much easier to shoot jpg and convert
> to tiff as the first operation.
>
> But this is just me...
The manufacturers do an extraordinarily good job in creating JPEGs
from the raw images their sensors produce... as well they should,
since it will always only be a minority that wants to get involved
in handling the details personally. I am just pleased that in many
compact cameras they are still prepared to give access to their
RAW files. This means I can help my wife get the best results from
her first hesitant steps in photography.
Personally, I find DX0 to have the best defaults for producing good
TIFF images from RAW files, and to offer the most comprehensive,
detailed and subtle controls for "tweaking" the default settings.
The snag is that DX0 does not do the same great job for camera and
lens combinations for which specific DX0 modules have not been
produced. So although I use DX0 for many of my photos, panorama and
non-panorama, I am forced to turn to CS5's built-in RAW developer
for a minority of them. The difference is like chalk and cheese.
<sigh> It's not that CS5 is so awful. It is not. But there is really
no comparison with the speed and ease of using DX0! And unlike PS,
which I still find far too complex and baffling after several years
of using it, DX0 quickly repays any experimentation with less
familiar parameters, providing real-time, or at least very rapid,
previews of every possible effect. Although not cheap, I still
regard it as one of the best software purchases I have ever made,
right up there alongside PTgui. High praise indeed, I am sure you
will agree!
But this is just me... <grin>
Roger W.
--
Business: www.adex-japan.com
Pleasure: www.usefilm.com/member/roger
Tom
With jpgs more control points are found, however there are still a lot
of gaps and a lot of manual work to do. Control point detection seems
to work better if I work with a small section of the panorama only.
I'll try what Hans suggested: Optimization with a small group of
images and then "Generate Control points for all images".
Tom
On Nov 18, 4:23 am, "Roger D. Williams" <ro...@adex-japan.com> wrote:
> On Thu, 18 Nov 2010 08:27:17 +0900, Ken Warner <kwarner...@verizon.net>