Bill Pierce
Well-known
For a long time one of the advantages of b&w and color negatives was their ability to capture a greater brightness range than digital cameras. While blown out digital highlights are always a possibility, something close to an impossibility with b&w negative film, a lot of modern digital cameras can now capture a density range of 14 to 15 stops. So, how do you go about doing this?
You are going to have to use your camera’s lowest real ISO. Be aware that some cameras that offer extended low ISOx below the sensor’s actual native ISO can lose range. On top of that you will have to “expose to the right,” ramping up the exposure as much as possible without overexposing the highlights. Some cameras offer viewfinder warnings, blinking or colorizing overexposed highlights. These are usually pegged to the jpg images, and the raw images will often hold detail at exposures that have just started the blinking in the viewfinder. And, of course, there is one other way to guarantee an exposure that provides as much shadow detail as possible without blowing out the highlights - bracket the exposure and use the most generous exposure that doesn’t blow out the highlights.
Why would you go to all this trouble when just relying on auto exposure and pushing the button has given perfectly acceptable results? Because in the world of digital photography it provides a relatively unique look that may please you and actually benefit some pictures. Warning, you will definitely want to play with the tone controls - contrast, highlight, shadow, even curves. The linear nature of the sensor file, unlike the curves inherent in film images, will definitely benefit from a little artistic twiddling.
Again, why do this? Even film users don’t whine about the limited range of Kodak Tmax 3200 or Kodachrome 200. Lowest ISO? That sounds like a pain. It is. But there are some images, some shooting situations that really benefit from it. Anyone willing to come to my defense?
You are going to have to use your camera’s lowest real ISO. Be aware that some cameras that offer extended low ISOx below the sensor’s actual native ISO can lose range. On top of that you will have to “expose to the right,” ramping up the exposure as much as possible without overexposing the highlights. Some cameras offer viewfinder warnings, blinking or colorizing overexposed highlights. These are usually pegged to the jpg images, and the raw images will often hold detail at exposures that have just started the blinking in the viewfinder. And, of course, there is one other way to guarantee an exposure that provides as much shadow detail as possible without blowing out the highlights - bracket the exposure and use the most generous exposure that doesn’t blow out the highlights.
Why would you go to all this trouble when just relying on auto exposure and pushing the button has given perfectly acceptable results? Because in the world of digital photography it provides a relatively unique look that may please you and actually benefit some pictures. Warning, you will definitely want to play with the tone controls - contrast, highlight, shadow, even curves. The linear nature of the sensor file, unlike the curves inherent in film images, will definitely benefit from a little artistic twiddling.
Again, why do this? Even film users don’t whine about the limited range of Kodak Tmax 3200 or Kodachrome 200. Lowest ISO? That sounds like a pain. It is. But there are some images, some shooting situations that really benefit from it. Anyone willing to come to my defense?
KM-25
Well-known
Digital has been, is and will remain like slide film, don't clip the highlights cause' you ain't getting it back. Outside of that and being very subjective it is up to the image maker to evaluate the potential in the scene in regards to light and contrast and potential negative or recorded image with the same criteria.
Even if the DR on paper is the same, not all sensors treat said 14 stops equally. For example, when expanding tonal sliders in post, my CFV II 50C back keeps the tonal range much more natural and gradual compared to my Z7II or Z9.
In terms of your third paragraph and coming to your defense? Your audience is far more experienced and knowledgable than that...;-)
Even if the DR on paper is the same, not all sensors treat said 14 stops equally. For example, when expanding tonal sliders in post, my CFV II 50C back keeps the tonal range much more natural and gradual compared to my Z7II or Z9.
In terms of your third paragraph and coming to your defense? Your audience is far more experienced and knowledgable than that...;-)
Ko.Fe.
Lenses 35/21 Gears 46/20
Meh. Any film still have unique DR comparing to any digital. Just be smart enough to realize what TMAX 3200 is not good idea for daylight and you could still push even cheapo Arista 400 @3200 and get better results than overpriced, overhyped TMAX3200. 
Jamie Pillers
Skeptic
I don't know about you, Bill, but I had plenty of film exposures with blown out highlights. I remember many frustrating hours in the darkroom trying to no avail to recover them. With digital, if I'm concerned, I'll bring up the histogram on the LCD and make sure I'm not overdoing it with the highlights. Not perfect, but good enough.
Bill Pierce
Well-known
Jamie, believe it or not, David Vestal once did a range of overexposures on Tri-X, which I believe he normally exposed at 200, and got down to E.I. 10 and was still able to print highlight detail. Needless to say, it was a bit of a joke and the development time was significantly reduced. But that's the trick - reduced development is a necessity.
580/Q
200 frames a day
"I am under no obligation to reveal the details in the shadows." I think it was Sensei Jay who said that.
And it's a paraphrase. Apologies. He shoots low, not all the time, but a lifetime of Kodachrome, with its saturation characteristics, naturally drives him to that god-knows-how-many-fractions-or-whole-stops-under spot that is required based on the subject matter - lumens and aesthetics.
So I guess, there are at least two ways of doing this. 1.) Change the dynamic range of the subject matter so it better fits in to the recording medium's DR, or;
2.) Squeeze the camera for all it is worth. That's what you're asking, I think.
What do you we have for basic subject matter? This might force me to using grads. I can't stand fiddly BS appurtenances to cameras -- most of the time. But, occasionally, when time and tide permit, I can slow down and do a set piece battle. If that's the deal, then maybe I'll mask windows, strap grads, fill, etc... don't make the camera(sensor/film) work that hard. All recording mediums -- audio, visual -- are compressed compared to real life. So by compressing the input dynamics (diffusing window light, filling shadow) we can make it look like we've got dynamic range. Cine does it all the time. Audio the same. The principles are identical.
But, If I gotta squeeze every bit out of the camera, and I'm not shooting RAW, then yeah for a while, if the subject was static and I was using a tripod, I would bracket, then in post, mask and stack. But that's still not exactly what you are talking about. By the way, I'm pretty sure that my iPhone is doing a variation of that as soon as I open the camera app: buffering and stacking and dumping--rinse and repeat until the "shutter" is "released." If Apple and some manufacturer of decent-sized sensored cameras would get together... Please?"
But if I'm shooting JPG and printing the file, I'll use a histogram or the flashing highlight warning and shoot a third of a stop under the warning. RAW is, for me, much the same, and, while it results ultimately in a JPG or PNG or whatever (compression of the DR at some other stage), one can go a little hotter and still get what you need.
But (always with the buts, this kid), if you're JPG or RAW, why not bracket by 1/3 stops and do fives? Or nines? Pick the one that is aesthetically the best. Film is cheap in the digital age.
One small observation. In some of the Fuji cameras, and I suspect this may apply to other sensor/JPG engine combinations, one doesn't get the full DR until you hit 800 ISO or thereabouts. Apparently a little gain-up is worth the trade off. Somebody who has been doing digital imaging work in television for decades explained to me, in front of several of his peers, why that happens but I couldn't keep it in my head for very long.
This post has gone on too long in proportion to how often I need the full dynamic gamut, if you will. I usually expose for the highlights and even then they are often not very close to clipping. That's a personal aesthetic and has nothing to do with the technical exercise.
XXOO
S
And it's a paraphrase. Apologies. He shoots low, not all the time, but a lifetime of Kodachrome, with its saturation characteristics, naturally drives him to that god-knows-how-many-fractions-or-whole-stops-under spot that is required based on the subject matter - lumens and aesthetics.
So I guess, there are at least two ways of doing this. 1.) Change the dynamic range of the subject matter so it better fits in to the recording medium's DR, or;
2.) Squeeze the camera for all it is worth. That's what you're asking, I think.
What do you we have for basic subject matter? This might force me to using grads. I can't stand fiddly BS appurtenances to cameras -- most of the time. But, occasionally, when time and tide permit, I can slow down and do a set piece battle. If that's the deal, then maybe I'll mask windows, strap grads, fill, etc... don't make the camera(sensor/film) work that hard. All recording mediums -- audio, visual -- are compressed compared to real life. So by compressing the input dynamics (diffusing window light, filling shadow) we can make it look like we've got dynamic range. Cine does it all the time. Audio the same. The principles are identical.
But, If I gotta squeeze every bit out of the camera, and I'm not shooting RAW, then yeah for a while, if the subject was static and I was using a tripod, I would bracket, then in post, mask and stack. But that's still not exactly what you are talking about. By the way, I'm pretty sure that my iPhone is doing a variation of that as soon as I open the camera app: buffering and stacking and dumping--rinse and repeat until the "shutter" is "released." If Apple and some manufacturer of decent-sized sensored cameras would get together... Please?"
But if I'm shooting JPG and printing the file, I'll use a histogram or the flashing highlight warning and shoot a third of a stop under the warning. RAW is, for me, much the same, and, while it results ultimately in a JPG or PNG or whatever (compression of the DR at some other stage), one can go a little hotter and still get what you need.
But (always with the buts, this kid), if you're JPG or RAW, why not bracket by 1/3 stops and do fives? Or nines? Pick the one that is aesthetically the best. Film is cheap in the digital age.
One small observation. In some of the Fuji cameras, and I suspect this may apply to other sensor/JPG engine combinations, one doesn't get the full DR until you hit 800 ISO or thereabouts. Apparently a little gain-up is worth the trade off. Somebody who has been doing digital imaging work in television for decades explained to me, in front of several of his peers, why that happens but I couldn't keep it in my head for very long.
This post has gone on too long in proportion to how often I need the full dynamic gamut, if you will. I usually expose for the highlights and even then they are often not very close to clipping. That's a personal aesthetic and has nothing to do with the technical exercise.
XXOO
S
willie_901
Mentor
In digital photography, the only way to maximize analog dynamic range for a single exposure is to maximize exposure (i.e. analog signal levels when the shutter is open). This occurs at the sensor assembly's base camera ISO setting.
Empirical Dynamic Range Estimates For Several FUJIFILM X Cameras [1]
These results are estimates from statistical analyses of unrendered raw file data. This eliminates undesirable dependence on variables such as differences in demosaicking algorithms and image-rendering parameter selections.
1/ Photographic dynamic range is different than engineering dynamic range. Photographic dynamic range is normalized to the sensor circle of confusion. Engineering dynamic range assumes the DR noise floor is defined by a 1:1 signal-to-noise ratio. Photographic dynamic range is approximately lower 2 EV lower than engineering dynamic range.
580/Q
200 frames a day
Thanks for that. It is an interesting plot and makes intuitive sense. Who authored the reduction? Are there citations? Given this, then, the trigger for the full DR in camera is obviously something written into the firmware. Possibly to accommodate some other parameter.
Ultimately it doesn't matter as, after working with a camera for a while (these days working with a digital camera is like working with a particular film or, if B&W, a stock/developer combination), one develops an awareness of how the sensor behaves under various conditions and works conscious of that restraint whether or not they are bound by it.
Edit: I just re-read the "undesirable variables" part. Yes, absolutely. But I tend to work faster than RAW will allow me and so yes, I am invoking the "undesirable" variable of jpg engine rendering. It's like shooting Kodachrome. You sent your film to the lab and for the first 200 rolls you bit your nails. After five hundred rolls, you knew, pretty much, what you were going to get.
Thanks again,
Shane
Last edited:
ranger9
Well-known
What makes me laugh is when the online film-photography pundits talk about all the dynamic range they're getting, and then they knock their photos down to 8 bits per channel and post them on Instagram...
pluton
Well-known
For some shots, not blowing out the brighter areas---clouds over a landscape, maybe---matters a lot. For some shots---documentary interior with small daylight windows-- the small blown out spots aren't a concern.
It's the shots that contain specular highlights or extreme bright spots, like sun peeking out from behind the clouds---that bother me, because the one small 'uber highlight' takes out a certain amount of the surrounding area, which looks like a mistake when printed or displayed. Only digital solution: tripod and HDR combo exposure. Too bad I only shoot with a tripod about 1% of the time!
It's the shots that contain specular highlights or extreme bright spots, like sun peeking out from behind the clouds---that bother me, because the one small 'uber highlight' takes out a certain amount of the surrounding area, which looks like a mistake when printed or displayed. Only digital solution: tripod and HDR combo exposure. Too bad I only shoot with a tripod about 1% of the time!
580/Q
200 frames a day
Well, yeah. Process - from the camera to the viewer's eyes constitutes the medium.
And for those of us who print, not one of the gamuts line up with more than a range of fidelity. It's damn close in many cases, better than a lot of colour emulsions, certainly, but it is part of a full time job keeping paper, printer, monitor, all profiled and in line with each other.
I've been on the road pretty steadily for the last 20 years or so and I do trial layouts/slideshows with a little dye-sub printer. I hate the colour sometimes and prefer looking at the colour images on a good screen - it's like light table and transparency combination, but it's "like", not "the same as", except in process. But it gives me, like a contact sheet, an idea of how things might be. When I get back to real-printer-land I can get the real print. But in the meantime...
Everything is a compromise of the ideal, a concession to the limitations of the entire medium. And, process is an inextricable part of the medium.
Shane
Attachments
willie_901
Mentor
I don't understand what you wrote. What do you mean by reduction? The author can be found by clicking on the link to the chart. This web site contains all the details. What do you mean by trigger? The firmware is irrelevant. What matters is maximizes sensor exposure maximizes analog dynamic range.
These data clearly show the only way to maximize dynamic range is to use the longest practical shutter time and widest practical lens aperture. This means using the lowest possible camera ISO setting. This is the only way to maximize bright regions' photo-sites full-well capacity. This determines the maximum possible signal level. The camera's data stream determine the read noise characteristics. Photon noise and read noise determine the noise floor. The dynamic range is the difference between the maximum signal level recored and the noise floor.
Again, I don't understand. Of course it matters. Intuition is not reliable. Using ISO 800 on any of these FUJIFILM cameras significantly reduces the maximum possible analog dynamic range. How can that fact not matter? These data are objective. When dynamic range is important use the lowest practical camera ISO setting.
What's undesirable are dynamic range measurements from rendered images. Differences in demosaicking algorithms and image processing parameters to compute the rendered image are undesirable. In the past, when I've shown data from this source sometimes readers criticize the results by assuming in-camera differences in JPEG rendering meant direct comparisons invalid. So now I always mention the results can not be influenced by rendering.
Last edited:
Oren Grad
Well-known
He took it all the way down to EI 0.37 - see Chapter 9 of The Craft of Photography. But I think the results David reported were mostly about the gently-shouldered curve of Tri-X. If you're going to print in the darkroom, it's much easier to get in trouble with T-Max 400.
olakiril
Well-known
I would say in the digital photography it makes much less of a difference to the look of the image than under/over exposing film, as long as you don't clip the highlights.
The relative linearity of a digital sensor provides quite similar contrast/color at either end, but film does not. You want pastel-like colors with lower contrast: overexpose (and bring down levels in post), you want a high contrast image: underexpose (and bring levels up in post).
With digital you just try to maximize the dynamic range by not clipping the highlights while keeping the shadows out of the noise. Given that modern sensors have more dynamic range than many good photographs actually need, the picture in post can really benefit from reducing the captured dynamic range. So I agree with what you said, in post you have to play with the tone controls to make the image look good (as you have to do with film), but very rarely I come across with a scene that I have to ETTR (exposure to the right).
580/Q
200 frames a day
Willie, before I try and clarify my comments I need to make one thing absolutely clear: I am comfortable with abstraction and conceptualization but I sincerely believe that Bill's question was based in the desire to produce a viewable image which, as well as meeting a host of other aesthetic criteria, had the best dynamic range possible while meeting those other criteria.
A viewable image. Not as an empirical test solely of a sensor's maximum dynamic range. All of my comments should be read in that light, if you'll forgive the pun. If I have misunderstood Bill's question, that mistake is mine and mine alone and I sincerely apologize.
Reduction: any time I see a plotted graph, I think that a reduction has been made. It's a habit and I may have used the term somewhat loosely. However, reading through the terms of reference and the sensor evaluation "primer" one can see that a statistical reduction has been performed. This is irrelevant, however. I'll get back to that. Mr. Claff outlines his assumptions and methodology clearly and it's informative. Thank you again for the link. I enjoyed reading it.
Trigger: I was referring to the in-camera availability of a Dynamic Range setting. For the selections and set up I use, the maximum selectable DR (400) is not available until I've hit an ISO of 800. Sorry for the confusion. That's what I meant by "trigger" - the camera allowed a higher DR setting with a higher ISO.
You are absolutely right: if we are measuring only sensor dynamic range the firmware is irrelevant.
But here's the thing: I can't see the analog dump. I can't even see the RAW file, really. Ultimately, you can't see the file as an image until it's rendered.Whether one uses SOOC JPGs or the most exotic RAW converter, it still needs to be rendered for it to be seen as an image. I need to use that sensor's dynamic range in a photograph. That means, in my world, firmware is relevant because nobody has ever shot a photograph simply, to the exclusion of all other values, to produce the maximum dynamic range -- outside of testing. Or to put it differently, no one has ever shot a photograph that could be seen without rendering.
Dynamic range is a factor in the production of a photo that meets some criteria set. Whether that criteria set is forensic, documentary, fine art, editorial, or whatever, dynamic range is just one of the factors which, amalgamated with other factors, produces the image one selects in the editing process.
Totally. Without question. The important word here is "practical". The fact that even if you sacrifice everything else to best dynamic range, the dynamic range of any sensor pales next to the human eye -- which is what will be used to look at the print/projected slide etc...
By "intuitive" I meant that the graph was easy to parse and its applicability, until I had read all the literature on the website, was intuitively sensible. It made sense.
Yes, I did say it didn't matter. Guilty as charged. And, I didn't provide sufficient context to qualify that statement. Apologies. Let me try again. If pure sensor dynamic range is the sole objective then, yes, go to native ISO. But I often want good colour, apparent sharpness, resolution, composition, etc, etc, etc, as well as dynamic range and that means I need to work the camera, respecting that some of these criteria are in direct competition (sometimes) with each other for priority.
The same way technique is more important to apparent sharpness than lens or camera, technique, or the use of the entire camera/lens/subject/light unity, will produce an image with apparent dynamic range to one's liking. Let me give you an example.
If I use fill light to illuminate the shadows and bring them closer to the highlights, I am reducing the dynamic range outside of the camera so that the camera can capture it.
But what if I don't want to use lighting? If I use ISO 800 I can expose for the highlights less a third of a stop, jack up the DR setting to max (really an in camera curve change) and the SOOC JPG will be very usable in terms of shadow detail. (Up to a point.) That's reducing the dynamic range inside the camera to produce an apparent dynamic range that is greater than it actually is.
Or, I can use the ISO that works for the practical shutterspeed and the practical aperature given the subject matter, shoot RAW, expose for the highlights less a third of a stop and drag the details out of the shadows in post. Guess what? I use all three depending on a whole lot of factors.
All of these techniques, and many more, compress the dynamic range at some point so that it "fits" within the camera's limitations and then pseudo-expands the range at some other point in the process. We're faking it. Even in B&W film - dodging and burning is an expansion of the apparent dynamic range.
What's desirable are good photographs. And some require more dynamic range than others. I'm not mounting technical experiments, I am mounting viewable images in the final analysis. I need a photograph to be seen and to look good. That is incredibly subjective. Selfish, even. Dynamic range is just one aspect. So to maximize DR in an effort to produce a good photograph means that all the other aspects still have to be respected and juggled.
Once again, the dynamic range of a sensor and any associated mechanism for preserving the data collected is vastly inferior to the human eye. This is true of virtually every recording technology. Audio is even more demanding (in some ways). But, as in audio, avoiding the clip at the top and the loss of data that falls below the noise floor is a constant struggle. Making compressed content seem less or even un-compressed is a challenge we meet in different ways.
Thanks for your time in your replies and for your perspective.
Respectfully,
Shane
A viewable image. Not as an empirical test solely of a sensor's maximum dynamic range. All of my comments should be read in that light, if you'll forgive the pun. If I have misunderstood Bill's question, that mistake is mine and mine alone and I sincerely apologize.
First, let me apologize to you for not writing more clearly. I will attempt to do better. And secondly, I would like to apologize to Bill for side-tracking the thread somewhat.
Reduction: any time I see a plotted graph, I think that a reduction has been made. It's a habit and I may have used the term somewhat loosely. However, reading through the terms of reference and the sensor evaluation "primer" one can see that a statistical reduction has been performed. This is irrelevant, however. I'll get back to that. Mr. Claff outlines his assumptions and methodology clearly and it's informative. Thank you again for the link. I enjoyed reading it.
Trigger: I was referring to the in-camera availability of a Dynamic Range setting. For the selections and set up I use, the maximum selectable DR (400) is not available until I've hit an ISO of 800. Sorry for the confusion. That's what I meant by "trigger" - the camera allowed a higher DR setting with a higher ISO.
You are absolutely right: if we are measuring only sensor dynamic range the firmware is irrelevant.
But here's the thing: I can't see the analog dump. I can't even see the RAW file, really. Ultimately, you can't see the file as an image until it's rendered.Whether one uses SOOC JPGs or the most exotic RAW converter, it still needs to be rendered for it to be seen as an image. I need to use that sensor's dynamic range in a photograph. That means, in my world, firmware is relevant because nobody has ever shot a photograph simply, to the exclusion of all other values, to produce the maximum dynamic range -- outside of testing. Or to put it differently, no one has ever shot a photograph that could be seen without rendering.
Dynamic range is a factor in the production of a photo that meets some criteria set. Whether that criteria set is forensic, documentary, fine art, editorial, or whatever, dynamic range is just one of the factors which, amalgamated with other factors, produces the image one selects in the editing process.
Totally. Without question. The important word here is "practical". The fact that even if you sacrifice everything else to best dynamic range, the dynamic range of any sensor pales next to the human eye -- which is what will be used to look at the print/projected slide etc...
By "intuitive" I meant that the graph was easy to parse and its applicability, until I had read all the literature on the website, was intuitively sensible. It made sense.
Yes, I did say it didn't matter. Guilty as charged. And, I didn't provide sufficient context to qualify that statement. Apologies. Let me try again. If pure sensor dynamic range is the sole objective then, yes, go to native ISO. But I often want good colour, apparent sharpness, resolution, composition, etc, etc, etc, as well as dynamic range and that means I need to work the camera, respecting that some of these criteria are in direct competition (sometimes) with each other for priority.
The same way technique is more important to apparent sharpness than lens or camera, technique, or the use of the entire camera/lens/subject/light unity, will produce an image with apparent dynamic range to one's liking. Let me give you an example.
If I use fill light to illuminate the shadows and bring them closer to the highlights, I am reducing the dynamic range outside of the camera so that the camera can capture it.
But what if I don't want to use lighting? If I use ISO 800 I can expose for the highlights less a third of a stop, jack up the DR setting to max (really an in camera curve change) and the SOOC JPG will be very usable in terms of shadow detail. (Up to a point.) That's reducing the dynamic range inside the camera to produce an apparent dynamic range that is greater than it actually is.
Or, I can use the ISO that works for the practical shutterspeed and the practical aperature given the subject matter, shoot RAW, expose for the highlights less a third of a stop and drag the details out of the shadows in post. Guess what? I use all three depending on a whole lot of factors.
All of these techniques, and many more, compress the dynamic range at some point so that it "fits" within the camera's limitations and then pseudo-expands the range at some other point in the process. We're faking it. Even in B&W film - dodging and burning is an expansion of the apparent dynamic range.
What's desirable are good photographs. And some require more dynamic range than others. I'm not mounting technical experiments, I am mounting viewable images in the final analysis. I need a photograph to be seen and to look good. That is incredibly subjective. Selfish, even. Dynamic range is just one aspect. So to maximize DR in an effort to produce a good photograph means that all the other aspects still have to be respected and juggled.
Once again, the dynamic range of a sensor and any associated mechanism for preserving the data collected is vastly inferior to the human eye. This is true of virtually every recording technology. Audio is even more demanding (in some ways). But, as in audio, avoiding the clip at the top and the loss of data that falls below the noise floor is a constant struggle. Making compressed content seem less or even un-compressed is a challenge we meet in different ways.
Thanks for your time in your replies and for your perspective.
Respectfully,
Shane
580/Q
200 frames a day
Bill,
Digging detail out of shadows is not that hard anymore. And, as we move the file to a print or even an online/screen-displayed final, if we don't make cellophane at the top end we can usually get a lot of details out of shadows at the bottom end and fake a broad brightness range. I shoot RAW and JPG and use the JPG 90% of the time. Hell, 99% of the time. Where the subject dynamic range is extreme I'll try reflectors and, when I can get it together, fill lighting. But I don't like lights for a host of irrational and rational reasons and I travel very light so if there is a modifier used, it tends to be a "natural" reflector, walls, open books, dinner plates, etc... The other factor is: I tend to shoot ISO 800 outdoors (unless I need a shutter speed lower than 1/500-1/1000) and 6400 indoors. By the time it gets to a print - even 20"x30" it's no factor.
As in another post: expose for the highlights -carefully- and raise the shadows by, 1. lighting at base ISO; 2. flexible ISO increase (expansion in post required). This is a broad division; many variations on the technique are available. The thing is, we just need to squish the DR so it fits in the camera and expand it in post to the print. Until sensors have variable ISOs for different areas of the sensor, we are kind of stuck with what recording engineers call companding -- compressing & expanding. Fitting the actual subject DR into the recording device, the camera, and then expanding it when we print. Dodge and burn or manipulate sliders with masks -- it's all the same. The levers are slightly different is all.
Again (ugh -- pun: A-gain), gain optimization is critical. All recording consoles have input gain controls. This is (very roughly) analogous to ISO. Getting the most robust signal (amount of light) onto the sensor so you can have the ISO as low (not below native) as possible is best. So sure. Set your drive mode for bracketing by 1/3 stops and expose as far to the right as you dare.
Because we are trying to do something that makes visual sense to us. That's why we go to the trouble. Perfectly acceptable is perfectly acceptable in certain situations. In others a more considered approach provides greater returns.
They may not be whining but you can be sure that they are tinkering with the soup. Those that have found the right developer (for them and their desired typical negative density -- another huge variable) are lucky. But all of them are working within the constraints of their chosen medium. Digital is growing in ways that film did (perhaps still does, I'm no longer au courant...) and the capabilities are expanding like they were for film in the previous century. We, the digitalistas/os, are still finding the constraints as quickly as they are being extended/expanded.
The reality that there are some images/shooting situations that really benefit...of course, but there is no policy that could ever be written because every picture is different. Buddha said: "Every shooting situation, every potential photograph, must be evaluated on its own merits." No really. He said that. I'm sure of it.
As to your defense... well, others have said it best, but Hell, I got your back.
Digging detail out of shadows is not that hard anymore. And, as we move the file to a print or even an online/screen-displayed final, if we don't make cellophane at the top end we can usually get a lot of details out of shadows at the bottom end and fake a broad brightness range. I shoot RAW and JPG and use the JPG 90% of the time. Hell, 99% of the time. Where the subject dynamic range is extreme I'll try reflectors and, when I can get it together, fill lighting. But I don't like lights for a host of irrational and rational reasons and I travel very light so if there is a modifier used, it tends to be a "natural" reflector, walls, open books, dinner plates, etc... The other factor is: I tend to shoot ISO 800 outdoors (unless I need a shutter speed lower than 1/500-1/1000) and 6400 indoors. By the time it gets to a print - even 20"x30" it's no factor.
As in another post: expose for the highlights -carefully- and raise the shadows by, 1. lighting at base ISO; 2. flexible ISO increase (expansion in post required). This is a broad division; many variations on the technique are available. The thing is, we just need to squish the DR so it fits in the camera and expand it in post to the print. Until sensors have variable ISOs for different areas of the sensor, we are kind of stuck with what recording engineers call companding -- compressing & expanding. Fitting the actual subject DR into the recording device, the camera, and then expanding it when we print. Dodge and burn or manipulate sliders with masks -- it's all the same. The levers are slightly different is all.
Again (ugh -- pun: A-gain), gain optimization is critical. All recording consoles have input gain controls. This is (very roughly) analogous to ISO. Getting the most robust signal (amount of light) onto the sensor so you can have the ISO as low (not below native) as possible is best. So sure. Set your drive mode for bracketing by 1/3 stops and expose as far to the right as you dare.
Because we are trying to do something that makes visual sense to us. That's why we go to the trouble. Perfectly acceptable is perfectly acceptable in certain situations. In others a more considered approach provides greater returns.
They may not be whining but you can be sure that they are tinkering with the soup. Those that have found the right developer (for them and their desired typical negative density -- another huge variable) are lucky. But all of them are working within the constraints of their chosen medium. Digital is growing in ways that film did (perhaps still does, I'm no longer au courant...) and the capabilities are expanding like they were for film in the previous century. We, the digitalistas/os, are still finding the constraints as quickly as they are being extended/expanded.
The reality that there are some images/shooting situations that really benefit...of course, but there is no policy that could ever be written because every picture is different. Buddha said: "Every shooting situation, every potential photograph, must be evaluated on its own merits." No really. He said that. I'm sure of it.
As to your defense... well, others have said it best, but Hell, I got your back.
Harry Lime
Practitioner
In my experience a lot of it is about the gentle roll off in the highlights that's at the root of the film magic. Throw it in a two bath developer and it's almost impossible to screw up.
The other thing about film is that when it does fail, it fails gracefully and often in an artistically pleasing way. Digital just chops it off once the receptor reaches capacity and without a lot of software voodoo it looks like, well... video.
I have found that the highlight priority exposure mode on some Nikon cameras makes a real difference. It give full attention to the highlights and relies on the low noise performance of the sensor to dig out the shadows. But with some curve magic you can work miracles with white dresses etc. If anyone needed a reason to buy a Nikon, that metering mode right there is enough.
I know of only only two digital camera that produce the smooth highlight roll off of film.
One of them is the old Fuji Film S5 DSLR that used their Super CCD. This CCD had an additional set of receptors that made a simultaneous bracketed exposure just for the highlights. The two exposures were combined on the fly and written out as a RAW file. Very, very impressive for 2010. The CCD had two sets of 6mp receptors and produced 6MP files.
http://shuttersounds.thedailynathan.com/2010/01/12/an-explanation-of-fujifilms-super-ccd-exr-sensor/
The other camera is the Arri Alexa digital cinema camera. Something like 85-90% of all movies and TV shows are shot on Alexa. The Alexa uses a custom sensor that does a simultaneous dual gain read on the receptor. One for normal range and the other for the highlights. This happens simultaneously and is not to be confused with a dual gain / ISO sensor that has two base ISO. Alexa files have a useable 15 stops of exposure and the highlights roll off like film. Nobody knows exactly how much range is captured and rolled down into the 15 stops contained in the file, but let's put it like this. I have never seen cut off blown highlights in an Alexa file. And when you do reach reciprocity failure, it fails gracefully like film (but without the black sun). I own and Alexa EV / Classic and it's basically a digital incarnation of Kodak Vision3 negative movie filmstock. Pretty impressive and you will be totally spoiled, because you'll never want to shoot another digital camera.
https://www.arri.com/en/learn-help/technology/alev-sensors
I really wish Leica would come out with a Leica M DR monster based on the sensor technology used in the Alexa. 24-36MP would be plenty but with +15 stops and that smooth rolloff at the top. 16bit or higher files for ultra smooth gradations.
Recently Canon came out with a compact cine camera that uses this approach, so it's not like no one else can do this...
The other thing about film is that when it does fail, it fails gracefully and often in an artistically pleasing way. Digital just chops it off once the receptor reaches capacity and without a lot of software voodoo it looks like, well... video.
I have found that the highlight priority exposure mode on some Nikon cameras makes a real difference. It give full attention to the highlights and relies on the low noise performance of the sensor to dig out the shadows. But with some curve magic you can work miracles with white dresses etc. If anyone needed a reason to buy a Nikon, that metering mode right there is enough.
I know of only only two digital camera that produce the smooth highlight roll off of film.
One of them is the old Fuji Film S5 DSLR that used their Super CCD. This CCD had an additional set of receptors that made a simultaneous bracketed exposure just for the highlights. The two exposures were combined on the fly and written out as a RAW file. Very, very impressive for 2010. The CCD had two sets of 6mp receptors and produced 6MP files.
http://shuttersounds.thedailynathan.com/2010/01/12/an-explanation-of-fujifilms-super-ccd-exr-sensor/
The other camera is the Arri Alexa digital cinema camera. Something like 85-90% of all movies and TV shows are shot on Alexa. The Alexa uses a custom sensor that does a simultaneous dual gain read on the receptor. One for normal range and the other for the highlights. This happens simultaneously and is not to be confused with a dual gain / ISO sensor that has two base ISO. Alexa files have a useable 15 stops of exposure and the highlights roll off like film. Nobody knows exactly how much range is captured and rolled down into the 15 stops contained in the file, but let's put it like this. I have never seen cut off blown highlights in an Alexa file. And when you do reach reciprocity failure, it fails gracefully like film (but without the black sun). I own and Alexa EV / Classic and it's basically a digital incarnation of Kodak Vision3 negative movie filmstock. Pretty impressive and you will be totally spoiled, because you'll never want to shoot another digital camera.
https://www.arri.com/en/learn-help/technology/alev-sensors
I really wish Leica would come out with a Leica M DR monster based on the sensor technology used in the Alexa. 24-36MP would be plenty but with +15 stops and that smooth rolloff at the top. 16bit or higher files for ultra smooth gradations.
Recently Canon came out with a compact cine camera that uses this approach, so it's not like no one else can do this...
580/Q
200 frames a day
Yes, this. You can sum it up by saying, Don't. Clip. Highlights. And you're so right: in digital, the transition to the "chop off point" is brutal. Ugly. If you want that for an effect, it is usually better to do it with film. Usually.
Respectfully,
Shane
Oren Grad
Well-known
I had an S5. It could certainly record a longer brightness range than other DSLRs of its day, and that was well worth having. But when you did hit its limits, it clipped every bit as abruptly as any other of the many different digital cameras I've used over the years.
EDIT: In hindsight, I think a lot of confusion arose from the fact that Fuji used a shouldered jpg curve in their DR expansion modes. But the raws behaved like everyone else's raws so far as I could see.