HDR High Dynamic Range Examples

[Jack Flesher]

FWIW, I was a beta tester for CS2 and the HDR Merge tool got a lot of discussion. Bottom line is it is a tool in the works for Adobe; they simply added it in its current form as a basic functionality in CS2 because they had what you see already developed.

As I understood it, they did this for two reasons, 1) it is a need in certain forms of digital video editing and 2) since they had the basics available they felt it might also whet the appetites of traditional digital imagists -- ostensibly a more functional version of HDR will give them a reason to upgrade to CS3...

Cheers,

[Jack Flesher]

Oh, and to answer Frank's original question: though I have seen impressive representations of incredible DR, I have never seen what I'd term a "really good" image made using the tool.

All FWIW,

[tim atherton]

"This noise
is there due to a lack of bit depth in the shadows hence shooting to the
right on a histogram"

I trust you aren't using the in camera histogram as an accuarate representation of where to place the highlights and shadows when actually shooting a scene?

[David Luttmann]

Matthew,

I'm afraid you are the one who isn't informed. I've already pointed out tests that prove that the dynamic range of DSLRs today attain approx 10 stops....which is equal to that of color negative film. If you can't obtain those 10 stops, it has to do with your processing technique and not limitations of the sensor. The Imatest software package is quite accurate in this respect and is used extensively in bench tests between DSLR & different films. In fact, it's the industry standard in this regard.

I have already produced dynamic range tests using Imatest to show the dynamic range I said a DSLR a capable of. Color neg film tests in between 9 & 10 stops just like a DSLR. I've tested neg film, slide film, and DSLRs. You on the other hand, have produced nothing other than an opinion. Until I see some comparison tests from you, I would appreciate you not blathering on about something you apparently know nothing about.

The fact that the toe extends highlight detail somewhat does not equate to extra stops of dynamic range. You apparently need to learn a little bit about how this range is measured. Produce your test results confirming that color neg film has a greater dynamic range than some modern DSLRs like the 20D or 5D from Canon, or keep to yourself!

[Jack Flesher]

Matthew:

Dave is 100% correct on this point.

Modern DSLR's do regularly achieve between 9 and 10 stops of DR, and some of the higher-end models approach 11. BUT! They have to be shot in their raw mode to deliver that range.

The biggest disservice -- really an obfuscation -- to the DSLR-DR subject was a recent online test done by one of the prominent websites... Unfortunately they chose to do their comparison by shooting their selection of DSLR's in JPEG mode and avoid any RAW file processing! Of course JPEG mode seriously compresses all images, and most DSLRs generate less than 7 stops from their in-camera JPEGs...

[David Luttmann]

First off, I didn't "pluck" from other sites. I don't put down film all the time....I just defend misinformation from people like you who have never done any testing and just parrot back the "film is better because I say it is" nonsense without ever properly comparing. You'll notice that others here are pointing this out to you as well.

My tests were done with the following:

Fuji Reala 100, Fuji NPH 400, Fuji Velvia 50, Fuji Astia 100, Fuji Provia 400 & Kodak Portra160. (Yes, I like Fuji film)

All scans were done on my following scanners:

Imacon 343 in 16 bit, and Minolta Scan Dual 4 in 16 bit , 16x multisampling mode.

I would say that these two scanners are representative of what most of the public can afford (the Minolta) and what most working pros can afford (the Imacon). I'm not concerned with $200 scans from drum scanners as I highly doubt you would have all your work drum scanned. And if you could afford the drum scanner to get an extra stop.....calling it 11 stops.....then you could afford a 16 bit digital back with 11-12 stops of dynamic range.

Really, this topic is a non-issue. The dynamic range measured in stops has been tested to death on various sites and by numerous pros....including myself. Even if there is a 1/2 stop or 1 stop difference between the two, be it in digital's or film's favor, that is not enough to have a real meaningful difference in the quality of photographic imaging. It is really only fodder for measurebators who have nothing better to do than try and put down digital capture.....and will grasp at every straw to do so.

[David Luttmann]

Oh, and by the way....the drum scans I have had done show little better than an extra 1/2 stop.

[David Luttmann]

Matthew,

I'm not going to redo all these tests to prove something to you that others have already acknowledged. This is old news. I did these tests quite a while back and see no reason to do them again. That & the fact that I can't find where I saved the Imatest results.

My suggestion is to go & purchase Imatest and try these tests yourself to see how they work in your workflow. Once you find that the results echo what has already been said, then you too can move on. I have no problem with film as a medium....I still use it a fair bit because I like to.....but I also understand its strengths & limitations. Dynamic range is no longer one of the strengths.

Feel free to share your test samples. Until then.....

[Jack Flesher]

Matthew:

Okay, if I understand you, what you are saying is that a direct digital file does not look at its base like scanned C41. If that is it, no argument here.

And I don't think Dave is bashing film either...

The bottom line is this: IF we have similar DR to begin with, and IF enough color information in the original digital file (say anything greater than 10 bits per channel) then we can almost certainly emulate (though not necessarily duplicate) the "look" (in digital parlance we call it a profile) of the C41 emulsion. This is fact and not conjecture.

Cheers,

[bglick]

Half all the tonal range of a digital SLR is in the top stop, half the remanining tonal ranges are in the "next to top" stop. So, 2 stops uses up 75% of all the available tonal range, 3 stops, we used up 87.5% of the total tonal range available. Now, regardless whether its 12 bit, or 14 bit capture, after maybe 4 stops of digital capture, there is so little tonal range remaining, I wonder how valuable the remaining stops are. I think this point is being overlooked, and its one area neg film still is superior as the tonal ranges remain higher throughout the same image exposure lattitude. How much? I don't know, but I would love to see this analzyed by someone :-)

So the ability to see detail / texture in more then 4 or 5 stops is very possible with a RAW file, as its remarkable how much hidden detail exist, but as for tonal range, well, its just not there. For some images with solid colors in the underexposed stops, this is not an issue, but where there is larger tonal range needed to be seen, well, it often results in everything from posterization to muddy appearance. How bad its displayed is also at the mercy of digital capture, such as if the color is red or blue, the sensor has only half the pixels to record with, since Bayer records in RGGB mode.

I feel this is still one major shortcoming of digital capture. It's my guess, its this shortcoming that Canon and other makers are focussing on in the R&D labs, not more MP. I am curious if the new 39 MP chips are recording tonal range in a fashion different then what digital SLR's record in, as I just described. Scanning backs record much better in this regard, but I am not sure of the method used.

As one poster mentioned, when you can control lighting, such as in studio, digital capture is remarkable...but one area it seems film still shines, specially negative film is recording large amounts of tonal ranges throughout large exposure lattitude. Of course, negative film has much less resolution, so often you have to go up in format size to maintain the sharpness you had in the next smaller format when using chrome films.

So, I think there is some confusion here regarding recording texture / detail, which RAW files come close to that of negative film, vs. recording the same amount of tonal range in each of the stops where texture is recorded. This inheritent weakness in digital capture is what HDR is all about, but as someone mentioned, its a very immature feature of PS CS2, but the premise is good. My guess is, Adoble won't develop it if they feel the weakness will be overcome by hardware technology in the sensor / electronics. I guess we will have to wait till CS3 to find out their positon.