p.1 #2 · An HDR question - Is bracketing really necessary?
Aaron D wrote:
Is bracketing really necessary? Since a single RAW contains 8 EV's worth of exposure latitude, what's the point in bracketing? Is there a point?
No, it isn't necessary, but there limitations to single shot HDR images.
For example: The signal to noise ratio is much lower in the shadow areas, pushing that in a single shot HDR will give you much less clean shaddows than a multiple shot tone mapped merge. In a scene with a wide dynamic range, you may have to serverely underexpose to avoid highlight clipping, then pushing those shaddow areas will really give rise to mud.
p.1 #4 · An HDR question - Is bracketing really necessary?
It is necessary. There is no such thing as a single shot HDR (check Wiki). Manipulation of a single shot is just a different way of processing a non HDR image.
The definition of HDR is the capture of more information than is possible from a single frame (therefore bracketing). the current top end sensors can capture about 14 EV (film has 7-11), but there are still scenes that can exceed that. The classic example is interior architecture shots, where you want the interior details but do not want the windows blown out. Also, even if you can fit the entire range in a single image, there may be advantages to processing with HDR in some cases.
What Brian described is completely true, but a single shot is by definition not HDR
p.1 #5 · An HDR question - Is bracketing really necessary?
hmmmm. Then I expect that the definition will need changing as future single shots will hold more DR than some of todays HDR shots. Besides, the usual way to utilise all of that single-shot DR is to turn it into multiple shots that are processed individually and then combine them as you would with the usual HDR approach.
p.1 #6 · An HDR question - Is bracketing really necessary?
We all know when a photo looks "nominal" or "normal" when we see it. What baseline is being used for the comparsion? Usually the memory of how that or similar situations was perceived by eye. Because of the way the eye pupils adjust brain filters eye input and "normalizes" color like AWB we perceived a full range of detail everywhere except specular highlight and deep voids like a cave and that what usually defines "looks normal" in a photo reproduction.
A practical way to test the range of the camera relative to that "normal" baseline and shoot in very strong light (sun and shade) with sun in face and at back. What I find doing that is the shots with the sun at my back look more "normal" SOOC than backlit ones. It's the same light, exposure for highlights, and camera sensor but the light direction changes the contrast of the content I think should look "normal".
Outdoor light over the shoulder is never eye-level "flat" like flash on camera but it casts shadows behind everything and highlights everything the camera sees from the front. Most of the time the shadows are mostly noise, but it's not noticed because the sensor records 95% of what is seen an more important "normally"
In cross light like a backlight face exposed for detail where the sun hits the shoulder of a white shirt the face will not look normal Sunny 16. Shifting to SHADY 5.6 will make the face "normal" but nuke the shirt and create a glowing halo around the backlit hair. Seeing that in a photo really doesn't seem odd because that's been the norm since color film was introduced. It can't handle contrast of a crosslit sunny scene either.
Confusing? Yes. That's why it's difficult to quantify when exposure isn't "right" or "normal" subjectively. Technically "nominal" exposure is seen by eye detail everywhere. That was the gaol with the B&W zone system when I used it.
But a digital SHADY 5.6 shot with a normal face and blow out white sky and shirt is subjectively considered "Par" for outdoor ambient only shooting in backlight. So based on subjective evaluation if you expose what the viewer will think important and don't make the stuff that isn't rendered accurately big and distracting in the photo it will look "normal"
The way I've always shot since having instant feedback is first expose for highlights under clippint, second look at the results and decide if it passes the "looks normal" test subjectively (playback, left side of histogram), and add flash or bias the exposure to get faces and other important stuff normal. Then I evaluate that result and consider if the blown highlights or clipped shadows will distract too much. If so I change the cropping to get them out of the frame, or change my POV, which changes the angle of the light relative to the camera (e.g., put the sun over my shoulder). Then I throw every Photoshop trick at the file to make it look as normal as possible...
I just used this in another flash related thread to illustrate how I approach a backlit "candid" shot:
I find most HDR I see looks fake because the tonal range less contrasty than perceived by eye because of all the detail pulled out of shadows in areas I'd just scan over by eye on the way to look at the faces and other lighter focal points. On the rare occasions I shoot scenics with tripod for HDR I first blend for the full range / exaggerated shadow detail, THEN burn the unimportant areas back down so in the photo the viewer skips over the less important stuff and hangs out longer on my intended focal points. In the example above it's the contrast of the white in the map (the lightest tone in frame) that pulls the eye there to find the person. I had to "burn in" the foreground because the flash fallout make it unnaturally bright. I "dodged" the trees and other details to pull out more detail but burned in the distracting people in the background on the right.
p.1 #7 · An HDR question - Is bracketing really necessary?
mhayes5254 wrote:
It is necessary. There is no such thing as a single shot HDR (check Wiki). Manipulation of a single shot is just a different way of processing a non HDR image.
The definition of HDR is the capture of more information than is possible from a single frame (therefore bracketing). the current top end sensors can capture about 14 EV (film has 7-11), but there are still scenes that can exceed that. The classic example is interior architecture shots, where you want the interior details but do not want the windows blown out. Also, even if you can fit the entire range in a single image, there may be advantages to processing with HDR in some cases.
What Brian described is completely true, but a single shot is by definition not HDR...Show more →
I equate it a bit like assembling a string ensemble that has instruments dedicated for the higher and lower tones instead of just using a single instrument with a wide range in the middle. While the mid-range instrument may have a range that can overlap and produces some of the same tones as the others, it can't produce the fullness/finesse of those tones as well at it's extremes as the others can in their respective ranges. And of course, the trio, quartet, quintet can produce a harmony/fullness that a single instrument can't quite achieve.
Not a perfect analogy ... but a solo vs. an ensemble ... each have different capabilities, and a different sound/vibe/look/feel. Some pieces can be performed by either, and for others the soloist has to choose which range will please the audience best.
A lot can be done with a single image in PP. My .02 is that if you're gonna bracket, there is no need to bracket anything less than 2 stops. PP can readily handle 2 stops or so without much trouble, so I typically bracket with 2 1/2 stop increments. This also happens to be close to the amount of diff between sunny side/shadow side of a given subject @ Sunny 16 (incident metering) @ capture (linear gamma).
p.1 #8 · An HDR question - Is bracketing really necessary?
I also prefer to bracket in 2 stop stops but unfortunately many Nikon bodies allow a max of 1 stop in the bracket function. You can do it manually of course, but the bracket function is convenient so I typically will bracket 5 stops and throw away the in between ones.
I do not do much HDR but have recently been trying the merge to HDR Pro in PS, which seems to work well.
Alan - no need to change the definition. We will just do it less. I also have used the multiple conversion approach for processing but just think of it as a different way to do PP. My first Digital SLR was a d70 with 10 EV of DR and 12 bits. My D800 has 14.3 and 14 bits. With the older cameras, I would more often need HDR but can often get it in one shot now. You could say I have an HDR sensor, but it makes the nomenclature confusing and there are plenty of grey areas. For now, it seems less confusing to apply the term "HDR" to the case of processing multiple frames to get more DR. I suppose this will evolve.
p.1 #9 · An HDR question - Is bracketing really necessary?
Yeah, HDR is actually a file format and not an effect. But people refer to "Tonemapping" as the HDR effect or just mistakenly call them HDR like the OP here. The definition of HDR will not change - until the file format changes of course. It is sometimes confusing when people mix up their terminology (in this case tone mapping with HDR) but I knew what Aaron meant by the context he used. It shouldn't be a big deal.
p.1 #10 · An HDR question - Is bracketing really necessary?
HDR is very popular with people who are new to photography because it seems like a magic solution -- but those who have worked with photography for a while often realize that manual blending is a far better solution. Save yourself from falling into the "HDR pit" by skipping this automated process, and learning how to do manual blending. Shooting HDR is much like leaving your camera in green mode and letting it make all the decisions for you.
p.1 #11 · An HDR question - Is bracketing really necessary?
Tone Mapping is extremely useful and often can work where manually masking and blending would be a nightmare. If you have multiple exposures all the better. One just has to use it sensibly is all. I'm not personally very interested in the genre some people refer to as the HDR Effect tho - that's kind of a different thing.
BTW cgardner,
A slightly more general automated approach to your example method there is just to use the three sliders in ACR: Fill Light, Blacks, and Recovery. Most of the time their default thresholds come out being the same as the blending masks I would have to create by hand.
p.1 #12 · An HDR question - Is bracketing really necessary?
Bifurcator wrote:
BTW cgardner,
A slightly more general automated approach to your example method there is just to use the three sliders in ACR: Fill Light, Blacks, and Recovery. Most of the time their default thresholds come out being the same as the blending masks I would have to create by hand.
Thanks for the suggestion. I do actually that and more in ACR globally on the RAW for every image before doing selective local adjustments with masks in PS. I prefer the separate masks so I can tweek by eye with sliders between 0-100% to find what looks best after opening the masks.
In most outdoor situation where scene>sensor and highlights are exposed for detail mid-tones get rendered a zone or so darker than they actually are and I pull them back up with brightness in ACR.
I use flash whenever possible in situations like that to get normal full range in foreground. Less attention gets paid to under exposed shadows in the background.
. I also have used the multiple conversion approach for processing but just think of it as a different way to do PP. My first Digital SLR was a d70 with 10 EV of DR and 12 bits. My D800 has 14.3 and 14 bits. With the older cameras, I would more often need HDR but can often get it in one shot now. You could say I have an HDR sensor, but it makes the nomenclature confusing and there are plenty of grey areas. For now, it seems less confusing to apply the term "HDR" to the case of processing multiple frames to get more DR. I suppose this will evolve.
