[ptx] longstanding & new problems/questions with 0.6

[Pablo d'Angelo]

Hi Rich,

> here some problems & questions that have arised while using hugin. most
> of them have already been reported here - unfortunately, most of them
> also received no response...

Sorry for not responding to all messages, but I'm currently quite busy
and don't have a much of time for hugin development. Naturally, given
the little time I tend to implement things that I'd like to use myself.

Especially for "easier" stuff, like checking for additional programs,
please feel free to contibute patches.

> all of these are valid as of hugin 0.6.

>1. tiff compression level field in sticher tab is disabled both for lzw
> & deflate compressions. seems strange.

Yes, the compression level for these is currently hardcoded somewhere in
the vigra library. Could be changed, but no time..

> 2. preview :
> enable auto. disable all images, click on one of them twice. button
> shows that image is inactive, but actually it is displayed. might be
> related to next one.
>
> 3. it is possible to crash hugin by enabling auto in preview & quickly
> clikcing on image buttons. i have reported this one a lot of times, i
> even tried producing a backtrace, which i sent to this list on 2006.06.30.

I'll try to look into the problem.

> 4. building libpano & hugin on slackware still requires manually
> exporting -lm. i've whined about this countless times :)

Any idea how to detect in the configure script if it is required?

> 5. in some case still deletes two keypoints when selected from the image
> window - reported here on 2005.12.15, even with a testcase. which is
> unaccessible by now already...

> 6. it is possible to open vignetting dialog several times - probably was
> not intended to do that.

Yep, not very useful...

> 8. the new vignetting removal functionality is really great, but it
> seems that automatic estimation always slightly overestimates for me. am
> i doing something wrong ? what would be the basic guidelines to achieve
> the best results ?

Currently there is no easy way to correct only for 90%. You could try to
lower the values of the r^6 parameter a little bit, if it overcorrects.

Since the correction by division is only valid for linear data, it might
help to set the gamma (in the pano tab) to 2.2, or try with linear raw images.

ciao
 Pablo