"fake" HDR vs layered "HDR" vs shadow/highlight

[goliath]

I'm starting to explore the path of HDR images and read on quite a few forums and tutorial.

I decided to try one using PS CS3 from a single RAW image from which I save a 0, a +2 and a -2 EV in 3 jpeg. When merging to HDR, the highlights were totally blown out so I save another version at -4 EV and run through the process again. The highlights were still awful so I uncheked the file with +2 EV, leaving PS to use the 0, -2 and -4 to process the HDR. Then I convert the 32 bit HDR to 8bit image, adjusting result using local adaptation curve. The result is still quite poor. Some tutorial talk about importing a layer of the original image and use mask in some area to correct the highlight (which I did not do for the picture below). I found the process quite time consuming ...

#1 - HDR from PS CS3 from a single RAW file

Then I came across a tutorial to create a "layered HDR" image. I tried the tutorial with the same jpeg I had created with different EV. The initial result was really not convincing so I played a little bit with layers/mask/blending mode to improve the output but it still look washed out (look at the trees to the right). Faster process than #1 but still time consuming and results is not so good.

link to tutorial http://photoshoptutorials.ws/photoshop-tutorials/photo-manipulation/layered-hdr-tone-mapping.html

#2 - "HDR" from layers/mask

Then I decided to try the simple image/adjustment/shadow-highlights on the original jpeg. 30 second process and look at the result ... I find it much better than #1 and #2 and it took less than 1/10 of the time.

#3 - simple shadow/highlight adjustment in PS CS3

From what I understand, an "HDR image" is an image that has a greater dynamic range than "normal image" because the bit depth is more important (and thus allow for more variation in intensity for each color channel). But when it comes to saving back to jpeg format (8 bit), the information available in shadows and highlights is bring back to the 0-255 scale +/- in the "midtones", which is about what the shadow/highlight process does when you look at the resulting histogram.

I know the original image may not be the ideal candidate for HDR image as the lighting conditions are not that extreme but still, can someone help understand why the "fake" HDR and "layered HDR" are so bad compared to the simple shadow/highlight?

#4 - original image with 0 EV adjustment in jpeg format

Thanks

 

[epp_b]

I don't see how any of those are HDR. They all look pretty overexposed to me.

The point of HDR for landscapes is to have a correctly-exposed, detailed foreground with a sky that isn't blown out. This usually can't be achieved without bracketing, although a graduated ND filter should work for HDR landscape shots.

This isn't, perhaps, the best example, but it shows what I'm trying to say:



The sky is moody with detail, with only a little bit of overexposure in the highlights near the sun. The trees are a bit dark, but you can still tell that they're trees and not silhouettes. Yes, the sky is a bit blurred because I need some more practice on my bracketing

 

[Arch]

Well... first off creating a fake HDR (using 1 raw) is really not the best way to use the HDR technique. There is no more information in these 'HDR's than is in the original RAW file.
That is why with a little work (shadow highlights / tone mapping / curves adjustments) the results don't look much different, and in some cases look better because you have more control over the image.

You would however see a difference using a program like Photomatix... it again is not going to give you information that isn't there, but, it will do a much better job at tone mapping the image than PS does.
Aside from using other software, in future if you really want to experiment with HDR i would recommend using different exposures rather than just one, and also shooting a scene which really demands it

 

[goliath]

Thanks for quick reply.

I understand that the picture was not a very good sample for trying to experiment with HDR (overexposed and light conditions not extreme enough) but I still don't get it why using PS merge to HDR with EV 0, -2 and -4 I end up with an image overexposed that much. The original 0 EV is not that overexposed ....

I tried very quickly with another sample I made with 4 pictures with different exposures and results are much better ... and in no way the shadow/highlight can bring back the details... much of the shadows but nothing in the highlights and if I take the underexposed image, it's the reverse...

Let me know what you think

#1 "correct" exposure original jpeg

#2 HDR - still not perfect but much better than my 1st try

#3 layered "HDR" - need additional masking work to get rid of highlights

#4 shadow/highligh adjustment from picture #1

 

[epp_b]

Um...mmm....those don't really lend themselves well for HDR shots. Try something like a sky with nice moody clouds with something dark in the foreground.

 

[Garbz]

This may be approached from the wrong way. Firstly each of the techniques can achieve stunning results, and each can look like it was taken with a phone camera. (welcome to HDR).

Secondly that tutorial you linked on HDR. I couldn't make it past page one. Pretty much the first thing he mentions is wrong. Tone mapping is a very different process than what he says and he's confusing it with the HDR merge. Secondly I have seen a tutorial showing Tonemapping in Photoshop can produce a result nearly identical to that of photomatix, for a wide variety of images, which leads me to believe he doesn't know how to use the software properly to begin with.

Pick one process and learn the ins and outs of it before you start comparing it with something else. "still not right" is not the way I would classify the HDR you just did above. "a long way to go" may be a better option.

Compare fully processed well thought out end results of each technique. Not a quick copy of what the internet tutorials do, and you are likely to get much nicer results.