Mark -
> I have some questions regarding enfuse. Although it is a bit too late
> for my book, I'd like to be able to better use the tool in the
> future. Any advice would be appreciated!
I wrote an Enfuse "User- and Reference Manual" (30+ pages).
It summarizes my experience of using Enfuse for mainly focus-stacking.
The document explains options, weighting functions, masks, memory
usage, and shows many applications of Enfuse, too.
I planned to submit it to Andrew, but his input pipeline seemed to be
clogged. Meanwhile, my focus has shifted. (Pun intended.)
> - GrayProjector (enfuse version 3.2) doesn't seem to work. My enfused
> images tended to have highlights that were brighter than I expected,
> and was hoping this would have solved some of the problem (using l-
> star), but a series of test images with various GrayProjector settings
> seemed to be identical.
manual> `--GrayProjector=PROJ'
manual> Use gray projector PROJ for conversion of RGB images to
grayscale
manual> masks. Enfuse uses grayscale projections of color images
in
manual> exposure weighting and contrast weighting.
Enfuse works with scalar pixel weights when exposure or contrast
weighting. For color input images this requires to project them from
color-space into a luminance-like representation. Option
'--GrayProjector' selects the projector to be used in this
transformation.
The reason for the projector being optional is that the projectors
produce different contrast and usually only the photographer can
decide what "best contrast" means for a particular set of images.
Option '--debug' is helpful to judge the masks' quality. Note that
mask contrast is less important when exposure weighting, but crucial
when contrast weighting.
> - EdgeScale doesn't allow me to input any value; I always get 'illegal
> numeric format'. Searching through the source code didn't seem to be
> helpful: perhaps it has been too many years since I've done coding!
Hmm, for me the following example works:
./enfuse -v -o out.tif \
--wExposure=0 --wSaturation=0 --wContrast=1 \
--EdgeScale=0.1 pinus-mugo-<01-15>.tif
I also tried '--EdgeScale=0.3:3:75%' with success. Could it be that
you have activated a fancy LOCALE on your system?
> - Assuming these expert options do work, and I'm doing something
> wrong, what is the intended purpose and suggested use of these
> options?
Dunno, whether you do something wrong. Your actual command-line could
be helpful though.
The purpose of this options. Well the purpose. Ugh. The purpus here
is an intricate thing...
If you fuse by contrast e.g. to produce a focus stack, Enfuse uses the
standard deviation of the pixels (after grayscale projection as
described above) of a moving window (edge length:
--ContrastWindowSize) to compute the weight. Now if you have
high-contrast out-of-focus edges this technique fails: the local
contrast is high, but the edge in fact is smooth no matter what size
the window has. Lenses with good bokeh amplify this problem. Andrew
was the first to point out this issue about a year ago.
To remedy this problem we came up with an alternative algorithm:
Compute the weight with a Laplacian-of-Gaussian. The
Laplacian-of-Gaussian is more or less immune against the
"high-contrast out-of-focus edge" problem, however it does not pick up
faint edges as diligently as the "standard deviation" algorithm.
Therefore, todays Enfuse features _both_ algorithms and the user can
select each of them or even use both of them in combination
(--MinCurvature).
The optimum choice of all parameters is difficult, but once achieved
Enfuse is capable of producing focus stacks of outstanding quality.
HTH,
Chris
--~--~---------~--~----~------------~-------~--~----~
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
-~----------~----~----~----~------~----~------~--~---
