HDR stitching when one exposure is incorrect
lucaslockie
I didn't manage to find a solution to my problem here and in a
meanwhile I solved it, so I'll write about it in my first post to this
forum, maybe somebody will find it useful.
I did a series of 75 photos to be merged into HDR in PTGui Pro using 3
exposure values (-2,0,+2 EV). Strangely, one photo turned out to be
incorrectly exposed, by 2 stops, although the Exif data was correct
(Canon 5D). Therefore the corresponding blend plane had a hole in the
middle, and the resulting panorama had a hard edged darker area there.
So I replaced the faulty photo with the brightest from the group, with
exposure offset setting. The optimum offset value was somewhere
between -0.8 and 0.0 (I don't understand why not -2?) The hole in the
blend plane was still there, but the photo contributed to the final
panorama anyway. However, my attempts were not accurate and the border
was still visible, so I decreased the priority of the whole group to
0.001 - and problem solved!
I still have pitch-black 1 pixel wide, sharp, lines on the edge of
light sources, I suspect minimal tripod movements, so I'll try to
unlink the photos or do some photoshop trick.
Lucas
Joergen Geerds
I do assume that you shot in RAW and process them to 16 bit tiffs? I
would suggest you look at your bracket set in bridge or ACR, or a
similar program. does the "bad" set look similar to the good sets? if
your bad set looks like 0, -2, 0, instead of 0, -2, +2, you can simply
adjust the bad one in ACR (or similar) with a +2 exposure. if the bad
one is totally corrupt, I would duplicate the 0, and use the same name
as the bad one, and adjust it accordingly.
generally, I would process bracket sets as flat as anyhow possible. no
sharpening, 0 brightness/contrast, flat curve, and only a minimum or
highlight recovery on the -2.
once you re-open your .pts, you should check the image parameters...
all your shots should read the same f-stop and iso, exposure
compensation 0, and your 3 different exposure times... based on that,
ptgui will give you the estimated EV values, which should be close to
+-2 and 0 (probably some wonky number like 1.9523456889). in case of
your one bad one, you should adjust only the exposure time to make it
fit into the set.
I don;'t know if you are using true HDR or exposure fusion, but I
would dial back the ptgui internal highlight recovery (never more than
1 ot of 4), because it tends to introduce strange artifacts around
specular highlights, like your 1px black halo, or a strange grey cast.
those steps should remove the artifact you mentioned. in case it
really doesn't want to go away, i would try to render the 3 individual
exposures in question to a separate psd file, and copy them only your
master psd/psb, and gently edit the glitch away with masks.
joergen
lucaslockie
Thank you so much for reply.
I'm sorry for being not clear. I converted RAWs into DNGs with Adobe
converter and fed them into PTGui. The nadir shots were replaced with
16bit TIFs (with ProPhoto color profile), with alpha channel, to mask
the tripod and other unwanted elements.
Other sets look alright but the faulty looks different. All of them
were bracketed bracketed with 3 times: 1/2s, 1/8s and 2s. In the
faulty set the last photo is actually underexposed by 4 stops, it
looks almost identical as 1/8s one. So I replaced it (just this one)
in PTGui with the copy of 1/2s one and adjusted the exposure offset in
PTGui with the value of -0.8EV. I found this by experimenting, but a
value slightly closer to 0 would yield a better matching result.
I enabled HDR Stitching with True HDR method and default parameters.
The blended panorama (JPG + HDR: EXR) had a small, triangle area of
slightly different exposure and sharp edge around. One of the blend
planes (number 0, the brightest) had a balck hole in that area, or
just a hole when alpha channel enabled in EXR.
Then I tried to blur the border by reducing the priority of the whole
set of 3 exposures, and the value of 0.001 left no apparent trace of
the problematic area.
By the way I started the whole process by automatic linking of the
images within the sets, but I found that some photos were misaligned,
despite of cable release, time lag and mirror lockup (???), so then I
unlinked some of them, but not the above mentioned.
> if your bad set looks like 0, -2, 0, instead of 0, -2, +2, you can simply
> adjust the bad one in ACR (or similar) with a +2 exposure. if the bad
> one is totally corrupt, I would duplicate the 0, and use the same name
> as the bad one, and adjust it accordingly.
I've just tried this method and change the exposure in ACR, but I used
the 1/2s photo, as the best exposed. In the resulting panorama I got
brighter, sharp edged triangle area (priority of the set was normal
again), and a hole in one of the blend planes. Seems strange to me...
> once you re-open your .pts, you should check the image parameters...
> all your shots should read the same f-stop and iso, exposure
> compensation 0, and your 3 different exposure times... based on that,
> ptgui will give you the estimated EV values, which should be close to
> +-2 and 0 (probably some wonky number like 1.9523456889). in case of
> your one bad one, you should adjust only the exposure time to make it
> fit into the set.
I have there only "round" values - I have no "Perform automatic
exposure and color adjustment" option checked" yet.
But now I changed the exposure time to match the pattern, as you
suggested - and it worked! :) There were still some artefacts, like
deeper then elsewhere dark areas around lights (not the 1px ones), but
adjusting the priority removed them, almost completely. Before I was
only adjusting the offset, not the time.
> I don;'t know if you are using true HDR or exposure fusion, but I
> would dial back the ptgui internal highlight recovery (never more than
> 1 ot of 4), because it tends to introduce strange artifacts around
> specular highlights, like your 1px black halo, or a strange grey cast.
I'm using true HDR. I couldn't find "internal highlight recovery"
anywhere... Oh, it is in Exposure Fusion dialog. But choosing Exposure
Fusion eliminates HDR as a result, but it I prefere to have it.
In tonemapping panel I have the following settings: Compression,
Saturation, Brightness, Contrast: 0, Radius: 0.5 + Dynamic radius
adjustment checked. In advanced panel White Point: 20 and various
values for R, G and B.
I'm sorry for being chaotic: I've just tried the Exposure Fusion and
saved the result as 16bit TIF, and it worked pretty well! The black
1px halo disappeared. The first approach was a bit too contrasty, but
I managed to correct it in PS.
Joergen, thank you sooooo much for your great help! I've been
struggling with this panorma for a quite long time.
All the best,
Lucas