I do not think that you do the lovely lady justice with the HD she seems to have aquired a five o'clock shadow. I must admit I would never have thought to apply HD to human faces since it is most successful with high contrast images. I can understand why the b/w would be more appealing in this case since that level of contrast would probably work with a b/w image, particularly with some 'grain' added. Sorry if this sounds a bit harsh but I do admire you for trying the experiment.
No no - that's fine - I agree. But then the B&W does not really look HDRish. I've seen some outstanding human HDR - just have not been able to recreate it.
Last edited by McQ; 25th January 2010 at 08:45 PM. Reason: removed broken link
This is an example of why I haven't tried to generate an HDR image from a single RAW file. There clearly isn't any information in the highlight on her nose, cheek and chin.
If it were possible to have her hold as still as a statue, you could increase the shutter speed 2 stops to capture that highlight detail.
The best solution I can come up with is auto-bracket at high shutter-speed and hope for the best.
Is there a better way?
Have you tried Topaz? For single images it can do some amazing things. Trying to tonemap a single image is like trying to squeeze blood out of a stone, if it ain't there, it aint there and the program is just guessing, IMHO.
Colin - what is the point you are trying to make anyway besides correcting semantics.
A single RAW image has between 12 and 14 bits if data per colour channel, so somewhere between 4096 and 16384 shades of red, green and blue. When you take the photo either:
1. The photo is under-exposed, so you capture a lot of highlight detail, but the colour values do not hit the 4096 or 16384 values;
2. The photo is perfectly exposed with blacks and whites captured.
3. The photo is over-exposed, so you have blown high lights.
In each of these three cases you can do some processing and get something that looks okay, with (2) perhaps being the easiest to work with.
However, with a single image, you have captured only the dynamic range that your sensor allows. You have a regular dynamic range.
The point of HDR is to capture multiple images, and use some processing tool to extend the range of the result. Normally HDR software uses 32-bits per colour channel when combining the images together. HDR techniques produce images that cannot be generated from a single raw image and post-processed with curves. Multiple images are used to piece together the result from perhaps 3 images, one providing the shadows, one the mid-tones and one the highlights. Each of these 3 images uses as much of the dynamic range of the camera ccd to provide as much detail as possible in these areas.
None of my answer should be controversial, I've explained what HDR is. For sure, the term HDR is applied to "a look" or "a style". I'm definitely referring to the technique.
Grinder - no one is saying you can't use the term HDR to apply to your image. All that I am saying, and Colin too, is that a single raw image cannot generate an HDR image - you cannot extend the dynamic range of a single image. No one wants a war!
No resultant image has a high dynamic range; measure any image you can fine and you'll find that the minimum value is still pure black (0) and the maximum is still white (255). What makes it HDR is the SOURCE that went in to making the final image, not the tone-mapping process that (may or may not) give it a certain look.
To say that an image has an "HDR Look" when they're really referring to a psudo ultra-tone-mapped look is wrong wrong wrong. In fact, an ideally processed HDR image would look like any other normal dynamic range image. It's important that people get out of this bad habit of misassociating a particular artistic processing choice with what HDR is all about, as the two aren't related. The map is NOT the territory.
Good post Graham,
I too do not wish to be contentious, but I do now believe there is an element of semantics in the common mis-understandings of the term.
I know personally I mistook the "High" of HDR as referring to the 8 bit Dynamic Range of the jpg image. If you accept that it refers to the Dynamic Range of the camera's sensor, whether that be 12 or 14 bits, then "High" means more than the camera can capture in a single exposure, thus satisfying Colin's definition. For me, this was the bit I didn't grasp at first, hence spelling it out here, to aid the clarification.
EDIT: This was written before Colin's most recent post above, but since I've limited myself to the bit range aspect, I don't think I have disagreed.
Last edited by Dave Humphries; 15th June 2009 at 10:24 PM. Reason: Added "For me.."
for me personally it's not about semantics. Now nitpicking and saying it's incorrect to label a ldr image produced from a hdr that took 3 captures to encompass the whole scene range is about semantics since we all know in that case it is a hdr image that's been tonemapped to make an ldr image.
However there is nothing "hdr" technically, practically etc etc from a single capture since it's by very definition (of all levels right down to simple essence of it) a low dynamic range capture and you cannot get out more than you put in (synergy doesn't enter into it with hdr, 2+2 will never be greater than 4 in this case). Personally I think people can call whatever they want hdr but it doesn't mean it's correct to do so and I side with colin on this since I feel the point he makes is valid and beneficial and isn't some obscure technical point but an accepted (increasingly misunderstood) fact.
Secondly understanding what hdr actually is and isn't will help everyone to get the style look they are after the best way too. There are more than 1 ways to skin a cat and more than 1 way to give a stylistic tonal compression or remapped colour effect.
"Any image where special techniques have been employed (Bracketing, GND filters etc) to overcome the cameras inability to capture the full dynamic range of the scene with a single exposure"
It's a "fine point", but none-the-less, I do think it's important to keep in mind that a true HDR representation of an image can only exist in the memory of our computers; there's no way to display an HDR image on screen* or paper, thus what we're really looking at is the resultant of a HDR source tone mapped to a normal dynamic range result (ie the tonal range of the result is the same as every other normal image). The HDR bit only refers to the dynamic range range of the original scene, not the final image.
* Most monitors have a dynamic Range of only around 5 (to 6) stops there is however at least one monitor (I forget the name) that used locally positioned LEDs for backlinghting ... the brightness of individual LEDs can be controlled giving a total DR of 11 stops.
So in reality there is NEVER an HDR image shown on this forum and thus every post is incorrect?thus what we're really looking at is the resultant of a HDR source tone mapped to a normal dynamic range result
The forum SHOULD be "Tone Mapping" and not "High Dynamic Range". . . . just for the sake of a friendly argument!
I just had a long conversation with a computer science student who works in my lab. He researches real-time HDR rendering. A couple weeks or so ago I sent out an email with a link to some "HDR" photographs that I had taken just to share with the lab.
He believes that we shouldn't refer to these images as HDR as the final results are not, in fact, HDR. They (HDR researchers) seem to take offense to us (photographers and artists) using the term HDR as it is something that they have been working on as computer science researchers for a long time and they feel that it misrepresents their work.
I tend to also agree, though find it a bit clunky to say "Check out my LDR images, which were created from a HDR composite!" At the end of the day, the resulting image looks very different from a traditional photograph.
I'm not sure that "tonemapped" encompasses it either. Hmmm.
Those are quite stunning and look like HDR composites to my eye. However, I understand your point.