On 22 Sep., 23:22, Bruno Postle <[email protected]> wrote:

> As I understand it, Hugin only uses the stack numbers for linking
> positions during optimisation.

I am not using the stacks to position the images - I uncheck the
positional link. What I use them for (apart from the obvious, which
doesn't work) is to set the CPG on the right track. I use cpfind for
both detectors in the 'stacked panorama' scenario. This way, I only
get CPs to images in another stack via the middle exposures, and the
CPs to fine-adjust the stacks inside the image, but no interstack CPs
for the other exposures.

> However stitching ignores these stack numbers.  With 'exposure
> fusion' enabled it just looks at which photos overlap at least 70%
> (by area) to determine which to fuse as stacks before blending.

This, then, is the source of my problem

> This does mean that small photos that overlap with large photos may
> get fused even if you would prefer them to be blended.

Thanks for your analysis. I now managed to circumvent the problem by
stitching first the 8mm exposures and then the 18mm ones to finally
blend the two subpanoramas into one. This is a cumbersome process,
though, and it would be much easier if the stitching process could
honour stacks if they are present and only do it's undocumented
hardcoded 70% trick when run in the 'exposure fusion from stacks from
arbitrary arrangement' mode. The current behaviour is misleading,
since a user who has assigned stacks will expect that an 'exposure
fusion from stacks' will in fact fuse these stacks, and not some
stacks the program makes up itself. In fact, the 'direct' route should
be simpler - the stacks are already set up, so there's no need to do
any trickery to make them up from thin air.

> Hugin currently doesn't give any indication of these calculations in
> the GUI, you have to stitch to find out what will happen.  It would
> be nice to show it somehow in the Preview, or just list the stacks
> in the Stitcher tab, or maybe a python plugin could list them.

I was working on stacks with an hsi/hpi plugin. The first plugin I
wrote was one to automatically assign stacks of three images to a all
images in a project - usually my brackets are of three images, and the
plugin (which works fine) helps me often :)

The problem with it is twofold: firstly, the 'three images' shouldn't
be hardcoded. I can use wxPython on my system to let it ask for the
desired number, but passing parameters to hsi/hpi plugins has not been
implemented for other platforms, even though it's the single most
urgent feature lacking to help hsi/hpi to become a useful tool.

Secondly, I tried to penetrate the logic which is used inside hugin to
assign/change/revoke stack numbers. In particular I failed to cleanly
implement behaviour like unstacking of all images and restacking with
a different images-per-stack number, not to mention assigning a
different scheme to a subset of images. As obvious as the stacking is
inside the pto file, as obscure is the handling inside the C++ code. I
couldn't fathom it, so I asked for help here:

http://groups.google.com/group/hugin-ptx/browse_thread/thread/2bb6f69866c56ad5#

as you can see, noone answered my question. Since this was all the
help I got I kept my plugin to myself. Every user who posts a 'hugin
cannot stitch my holiday snaps' message tends to get at least an
answer pointing them to the documentation, so I was pissed off by this
non-echo, even after asking a second time.

I think there should be a button in hugin saying 'autodetect stacks'
which triggers the 70% trick and reflects it by assigning the
autodetected stacks to the images. the 'exposure fusion from stacks'
should use the extant stacks, if present, and if not present it might
ask if it should try to autodetect or plain abort. And the 'exposure
fusion from arbitrary arrangement' should autodetect stacks if no
stacks are present and might complain if stacks have been previously
assigned and conflict with it's autoamtically generated ones.

Kay

-- 
You received this message because you are subscribed to the Google Groups 
"Hugin and other free panoramic software" group.
A list of frequently asked questions is available at: 
http://wiki.panotools.org/Hugin_FAQ
To post to this group, send email to [email protected]
To unsubscribe from this group, send email to 
[email protected]
For more options, visit this group at http://groups.google.com/group/hugin-ptx

Reply via email to