Enfuse v4.2 question

[twalp]

I'm learning the command line version of Enfuse (4.2).

I've read the manual, skipping only the equations and graphs which are beyond my expertise level. 

The bracketed images I'm working with are 16-bit TIFs. Although I shot 7 images in 1EV steps I'm experimenting with using only three: +2EV 0EV -4EV

The scene is an interior with sunlit grass and sky visible through two large windows. Here's the +2EV image.

The -4EV shot is a perfect exposure of the outdoors visible through the window (the room is barely visible).

In the resulting fused image the outdoors is overexposed -- too light. Changing saturation and contrast weight has a noticeable effect on the interior but not much on the view through the window.

I tried --exposure-cutoff=0%%:95%% (%% needed for a Windows batch file) but it seems to make no difference. 

What should I try if I want to improve the view through the window?  (BTW, I initially used all 7 images and the result was about the same as with only the 3 images, thus my narrowing the input images down to one exposed for interior shadow areas, one exposed for the interior, and one exposed for the outdoors.)

Thank you.

[cspiel]

Some comments ...

  - Reducing your image set to three exposures is a Good Thing (tm).
  - Option ` --exposure-cutoff' looks promising.

... and ideas

  - Switch off saturation weighting with `--saturation-weight=0'.  This may or may not
    help, but certainly will make the effects of exposure weigthing obvious.
  - Play with the `--exposure-optimum' option, especially with values less then .5.
  - Let Enfuse write its masks with `--save-masks', then inspect the mask images
    whether they reflect _your_ intended pixel weights.

Last resort: Supply me with _small_ (~ 1500x1000 pixels) versions of your images
and I'll have a look at your problem.


HTH,
    Chris (the one who collected all the equations and graphs;)

[twalp]

Thank you!  I'll get on it tomorrow.

[twalp]

@cspiel, I've done quite a bit of experimenting since your reply but feel like I'm going in circles.

Ultimately, reducing the --exposure-optimum control from .5 progressively darkens the entire scene, which definitely brings out details visible through the window but at the expense of the room interior. I was hoping to get more detail in the window area without adversely affecting the interior. 

Here's an acceptable fusion of the three attached jpeg images. (when I reduced the 16b tif's to the size you requested I pulled jpegs instead of tif's. Let me know if that creates a problem.) 

The Enfusion parms used were: --saturation-weight=0 --exposure-cutoff=0%:95% --contrast-weight=0. I prefer the results with these settings over what I get using Enfuse's defaults.

The outdoors visible through the window is still light.  I'm wondering if there's a way to use more from the window area in the darkest image (LR01) without it impacting the room interior. 

On Wednesday, June 1, 2016 at 4:02:06 AM UTC-5, cspiel wrote:

[cspiel]

Twalp -
        THX for supplying the actual images!

Your scene has an _enormous_ range of contrast
that you want to squeeze into an image to be
displayed on a computer screen (or even worse a
print on paper).  These "inside-looking-outside"
images are amongst the most demanding not only
for Enfuse.  Enfuse tends to render scenes quite
"naturally" in contrary to a distinctive
HDR-look and thus has to sacrifice other things,
which, in your case, means local contrast in the
highlights.

I have fused the images with your parameters and
the result looks convincing (to me).  You could
immediately gain more contrast in "the outdoors"
by _reducing_ the number of pyramid levels.  Add
for example
        --levels=-1
or
        --levels=-2
to your best command line.  The result will look
more like a typical HDR image including all the
unnatural contrast in the floor and plafond.
However, the contrast in the window and clouds
will increase.

Other ideas (besides taking your images again):
 * Load the Enfuse output image into your
   favorite editor and tweak with e.g. "Curves"
   or a similar tool.
 * Play with different exposure weighting curves
   (while having `--contrast-weight=0'), this
   is, `--exposure-weight-function' in
   conjunction with `--exposure-optimum'.  YMMV,
   though.
 * Fall back to a plain HDR work-flow.  See also
   Appendix A.4 of the Enfuse Manual.


Thanks,
        Chris

[twalp]

Chris, your reply has helped me calibrate my expectations, which is probably more important than anything. 

I assume by your statement, "I have fused the images with your parameters and the result looks convincing (to me)." you got the same result as the fused sample I posted previously.  I am quite happy with it, especially if the alternative is increased contrast in the floor and popcorn ceiling. 

I will experiment with your suggestions so I can learn more about the available tools. 

Thank you VERY much for helping.

[Carlos Eduardo G. Carvalho (Cartola)]

I would also suggest that you fuse with enfuse generating a 16 bits image, like a TIF, and then process it in Rawtherapee, that is also free and can deal separately with dark and light areas. It is really powerful.

 Nowadays I have practically stopped doing image fusion (also mainly because of the use of a philopod instead of a tripod) and have been treating the RAW directly with Rawtherapee. I usually shot a little darker to then recover the details of dark areas on it.

Cheers,

Carlos E G Carvalho (Cartola)
http://cartola.org/360
http://www.panoforum.com.br/

--
A list of frequently asked questions is available at: http://wiki.panotools.org/Hugin_FAQ
---
You received this message because you are subscribed to the Google Groups "hugin and other free panoramic software" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hugin-ptx+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/hugin-ptx/5feca2a3-a4ad-4c6e-9fe6-7109fd9afc98%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[twalp]

I'm replying again to post the results of "reducing the number of pyramid levels" -- in case a future reader is curious.  It should be noted that my using Enfuse options like this are akin to me randomly flipping switches in a jet cockpit, but there definitely is a visible pattern.

Baseline Enfuse using --saturation-weight=0 --exposure-cutoff=0%%:95%% --contrast-weight=0

add --level=-1

change value to --level=-2

I assume there are situations where reducing the number of pyramid levels helps an image but not in this case. The gray cast that replaces highlights is downright nasty (though less obvious here than when viewed on my computer).

Thank you for the insights!

On Monday, June 6, 2016 at 12:08:01 PM UTC-5, cspiel wrote:

Twalp -

You could
immediately gain more contrast in "the outdoors"
by _reducing_ the number of pyramid levels.  Add
for example
        --levels=-1
or
        --levels=-2
to your best command line.  The result will look
more like a typical HDR image including all the
unnatural contrast in the floor and plafond.
However, the contrast in the window and clouds
will increase.

Thanks,
        Chris

[bugbear]

twalp wrote:
> I'm replying again to post the results of "reducing the number of pyramid levels" -- in case a future reader is curious. It should be noted that my using Enfuse options like this are akin to me randomly flipping switches in a jet cockpit, but there definitely is a visible pattern.
>
> Baseline Enfuse using --saturation-weight=0 --exposure-cutoff=0%%:95%% --contrast-weight=0

Didn't there used to be a semi interactive GUI for enfuse, which would make
such experimentation, if not simpler, quicker and easier?

BugBear

[Frederic Da Vitoria]

2016-06-08 10:03 UTC+02:00, bugbear <bug...@papermule.co.uk>:

> Didn't there used to be a semi interactive GUI for enfuse, which would make
> such experimentation, if not simpler, quicker and easier?
>
> BugBear

Do you mean EnfuseGUI?

--
Frederic Da Vitoria
(davitof)

Membre de l'April - « promouvoir et défendre le logiciel libre » -
http://www.april.org

[bugbear]

Frederic Da Vitoria wrote:
> 2016-06-08 10:03 UTC+02:00, bugbear <bug...@papermule.co.uk>:
>> Didn't there used to be a semi interactive GUI for enfuse, which would make
>> such experimentation, if not simpler, quicker and easier?
>>
>> BugBear
>
>
> Do you mean EnfuseGUI?

A quick google says "no", since the enfuseGui site says:

✓Multi platform (Windows and Mac)

And I'm on Linux.

It looks out of date too - the site speak of Windows 7.

http://software.bergmark.com/enfusegui/Main.html

BugBear

[Andreas Metzler]

bugbear <bug...@papermule.co.uk> wrote:
> Frederic Da Vitoria wrote:
>> 2016-06-08 10:03 UTC+02:00, bugbear <bug...@papermule.co.uk>:
>>> Didn't there used to be a semi interactive GUI for enfuse, which
>>> would make such experimentation, if not simpler, quicker and
>>> easier?

>> Do you mean EnfuseGUI?

> A quick google says "no", since the enfuseGui site says:

> ✓Multi platform (Windows and Mac)

> And I'm on Linux.

[...]

Perhaps had LuminanceHDR in mind which is not a enfuse frontend but
can be used to tonemap a HDR image generated by hugin.

cu Andreas
--
`What a good friend you are to him, Dr. Maturin. His other friends are
so grateful to you.'
`I sew his ears on from time to time, sure'

[Alister Ling]

Hi, I had a very similar issue with my situation of a full Moon (right at the top end of white) over a foreground city (very dark) 12 EV difference (6 frames X 2EV diff)

Here are my enfsue settings:

exposure 1; saturation 0, contrast 0, mu 0.2, sigma 0.2. Although this is a tad dark,  in post I gamma 1.3, contrast 5, and light 3 to render nicely the dark images, and barely affects the light images (of which 4 of the 6 images in the bracket are overexposed).

Ideally, I am thinking it would be nice to have some form of gamma weight built directly to enfuse, to minimize the whites blowing out while bringing up the darks.

By the way, the gray on the floor, if I have it right, is actually a reflection of the sky, much bluer than anything else in the image, because it is not blown out (clipped). One of those things that the brain just ignores or accepts but stands out in a bad way in an image. 

Regards,

Alister.