2012/12/19 JohnPW <johnpwatk...@gmail.com> > Exactly! > > > On Wednesday, December 19, 2012 5:13:57 AM UTC-6, Frederic Da Vitoria > wrote: >> >> I expect the faux-bracketing to keep the lightest parts of the >> darkest exposure and the darkest parts of the lightest exposure. If >> this were true, there should not be any loss in the highlights. >> Obviously enfuse decided otherwise, but why did he do so and is there >> any way to make him behave as I expect? >> > > I'm expecting this script to increase the amount of information in the > image (but really, that seems counter to basic laws of thermodynamics.) It > does seem to do that generally, but clearly information is lost in the > highlights. Although I don't perceive it, I imagine that information is > also lost at the dark extremes as well. My suspicion is that it has to do > with the fact that the tips of the sigmoidal curve are locked down on the > white and black points and somehow the curve is not acting the way we > expect at these extremes. > > With a real bracketed shot (and with a brackets made from a RAW image) the > "least exposed" image will give you the most usable detail in the > highlights. In both cases the curve still has a long tail with a gradual > slope that spreads the highlights out. With this script, the end points > remain at exactly the same point on every intermediate image (there's > nowhere else for them to go.) This forces the extreme tip of the curve to > compress severely and the points on the curve nearest the endpoint must > necessarily change radically. I'm not sure what the shape of the curve at > these extremes would be, but I'm thinking it must have the effect of > clipping the highlights instead of spreading them out. If so the same thing > happens at the low end. So my guess is, contrary to what I expected, that > the extremes get slightly clipped, rather than enhanced, to the benefit of > the main part of the image. > > That's not to say it's bad though. Obviously it the fat part of the image > benefits. The same is true when you adjust the contrast curves in the > normal way. Effectively you sacrifice less important information at the > extremes (or wherever) to emphasize more important information in the > middle (or wherever.) >
I played a little with o1 in Gimp. Here is what I found. First I thought about using the Shadows and Highlights plugin http://registry.gimp.org/node/116 I thought that it should achieve the same kind of results. To my surprise, I could not, actually, I was getting inverse results, I mean inverse when compared to o2. So I decided to use the good old curves filter. I applied a sigmoid curve. I don't know how to describe this best, but I'll try: I dragged a point near the bottom left of the curve slightly up and left, and a point near the top right of the curve slightly down and right. The result should be to lighten the shadows and darken the highlights, thus revealing details in those parts. This should be something like what is achieved with HDR. I did reach those results, but at the same time, I lost contrast, which after all was to be expected. Then I applied the reverse sigmoid on the original picture, and reached an enhanced contrast, which seems closer to what I see in o2. But I noticed that I saw more details in the shadows of à2. So I believe that the faux-bracketing results are closer to a "D" shaped curve, with a point near the bottom left of the curve slightly up and left and a point near the top right of the curve slightly up and left too. This means that enfuse does not apply a sigmoid curve, but it seems he gives the preference to the lightest point. -- Frederic Da Vitoria (davitof) Membre de l'April - « promouvoir et défendre le logiciel libre » - http://www.april.org -- 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 hugin-ptx@googlegroups.com To unsubscribe from this group, send email to hugin-ptx+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/hugin-ptx
