Luminous Landscape Forum
Equipment & Techniques => Digital Cameras & Shooting Techniques => Topic started by: BJL on March 13, 2018, 08:10:08 pm
-
There has been much discussion at this site of the idea of “Expose To The Right” [ETTR], but unfortunately it is stated in terms of the “photosite histogram”, with the right end corresponding to photosites at their full well capacity, but that is not directly available. Instead, there are histograms for standard JPEG output at the selected EI, and that is always further to the right, more so as the EI setting is increased above the sensor’s saturation based speed as measured by the ISO definition of SSat (often called “base ISO speed”). It gets worse when people think that pushing that JPEG histogram to the right at elevated EI settings has the same virtues as ETTR.
I am working on a good strategy, based on what the camera’s JPEG histogram or highlight overexposure blinkies tell me. One key is that the difference between the right-hand end of the JPEG histogram and the actual “photosite histogram”, in stops, is the gap between the Exposure Index setting (“ISO setting”) on the camera and the true SSat (base ISO speed) — plus a half stop. Also, we can get that true SSat from DXOL it is what DXO wrongly calls the “true ISO” in the case where the camera is at its lowest normal EI setting.
In this post, just comments on when true ETTR is usually done: with the camera at its minimum normal EI setting. Then as a guideline, one can get the JPEg histogram at the right edge, then increase the exposure by
“EI - SSat + 1/2”.
For example, if the minimum EI setting is 100 but the SSat measure there is 70 (1/2 stop less), the JPEG histogram is one stop to the right of the photosite histogram, and so you could increase exposure one stop beyond the “JPEG Histogram To The Right” level.
Also, you can probably use the highlight blinkies to find that “Histogram To The Right” level: reduce exposure till the blinkies disappear (except maybe in specular highlights that you do not mind being blown out) than add one stop of exposure (probably meaning doubling the exposure duration).
I use the blinkies method cautiously, only adding half a stop.
Details later (at least if there is any interest) like where that half stop adjustment comes from.
-
Use the right tool for the right job:
https://www.rawdigger.com
With respect and some love to Michael who coined the term ETTR, it's got to go away. It is optimal exposure for raw data (OEFR?). ;)
-
Use the right tool for the right job:
https://www.rawdigger.com
Or FastRawViewer (https://www.fastrawviewer.com/). But while working in-camera, is there really any alternative better than an educated guess?
-
But while working in-camera, is there really any alternative better than an educated guess?
Decades before cameras had Histograms and some of us shot film, no educated guessing needed for transparency film (which requires optimal exposure).
Once you understand how your sensor behaves, with the aid of RawDigger or FastRaw Viewer, you work along the same lines.
-
Use the right tool for the right job:
https://www.rawdigger.com
With respect and some love to Michael who coined the term ETTR, it's got to go away. It is optimal exposure for raw data (OEFR?). ;)
Agree with RawDigger or the FRV variant. Also, optimal exposure works. Years ago, Emil Martinec posted where he equated ETTR with Maximizing Exposure. Either worked.
When I was on cannot I would use camera (5D3) spot meter to measure significant highlight. Now on GH5\G9, I use zebras. Prior testing with RD allowed me to then know how many stops I could punch the exposure to optimize/maximize the RAW exposure
-
Or FastRawViewer (https://www.fastrawviewer.com/). But while working in-camera, is there really any alternative better than an educated guess?
Use an incident light meter?
Kent in SD
-
Use an incident light meter?
Kent in SD
Still have to “calibrate” to resultant RAW exposure
-
Still have to “calibrate” to resultant RAW exposure
Exactly! But yeah, an incident meter can come in handy considering how 'dumb' reflective meters can be in some cases (white dog on snow, black cat on coal).
-
Exactly! But yeah, an incident meter can come in handy considering how 'dumb' reflective meters can be in some cases (white dog on snow, black cat on coal).
Reflective meters are dumb, but not in this case. By metering the significant highlight and adjusting exposure up based on prior tests (so you know how you camera RAW limits are) you have an optimized exposure which you can now push around to optimize resultant image.
Incident meters were invaluable when using “dumb” film cameras....and still useful if you want to “get it right in camera”with jpegs. They do not compare with properly using today’s digital cameras.
-
Reflective spot meters are not dumb when the photographer knows how to use them. Same with incident meters! Film or digital.
-
Use the right tool for the right job:
https://www.rawdigger.com
Yes, I am sure that a more precise calibration could be made by comparing some JPEG histograms to raw histograms at the camera's minimum E setting — which is where having highlights go beyond maximum raw level is what one needs to avoid. (But only at the lowest EI setting level, and maybe at one or two higher EI setting values in cases where the lowest EI settings are special "low" ones.)
However, I think a far simpler calibration is possible; something like:
- Test with a uniformly bright main subject, to get a histogram with a spike at the right.
- Find the exposure level that puts the camera's histogram at the right — or in the blinkie method, the highest exposure before you get massive blinkies.
- Take a series of images starting at that exposure level and moving up though the next several levels (say 1/3 stop increments in shutter speed)
- Check how much higher the exposure can go before the raw histogram goes over the top: the difference is the " allowable JPEG overexposure".
- Maybe repeat with some different sort of scenes, to see if that measure of "allowable JPEG overexposure" is consistent — if not, I would play it safe and use the lowest result.
With respect and some love to Michael who coined the term ETTR, it's got to go away. It is optimal exposure for raw data (OEFR?). ;)
I agree that "ETTR" is no longer an ideal name. I would like a name which emphasizes that this idea of optimal exposure is not relevant in "limited light" situations where the maximum allowable exposure level leaves the photosite histogram well to the left (no photosite close to FWC), and the question is then how much analog gain to apply—that is, how high it is worth pushing the EI setting.
-
https://www.fastrawviewer.com/blog/spot-meter-exposure (https://www.fastrawviewer.com/blog/spot-meter-exposure)
https://www.fastrawviewer.com/blog/in-camera-histogram-doesn%27t-represent-exposure
-
https://www.fastrawviewer.com/blog/spot-meter-exposure (https://www.fastrawviewer.com/blog/spot-meter-exposure)
https://www.fastrawviewer.com/blog/in-camera-histogram-doesn%27t-represent-exposure
Correct....Iliah Borg has many good write ups on it. At your referenced FRV and On RAWDIGGER site.
-
From
https://www.fastrawviewer.com/blog/in-camera-histogram-doesn%27t-represent-exposure
I am reassured by this observation:
while the in-camera histogram for #2640 is dangerously close to the right wall, indicating essentially no headroom (not to mention that the whole frame is one solid "flashing area", indicating overexposure), the RAW has a headroom of slightly more than 1/2 of a stop before highlight clipping;
That sounds exactly like the half stop adjustment that I mentioned in my OP, and it is no mystery once you compare the ISO 12232 specifications for (a) standard JPEG output placement, and (b) SSat. The later is explicitly described as allowing half a stop more headroom that standard JPEG output.
In brief:
- Exposing according to reflected light metering will place the metered subject brightness in standard JPEG output at level 112, just under 18% of maximum, about 2.5 stops below maximum level.
- SSat is the Exposure Index that will place that same meter subject brightness half a step lower, so about 3 stops below maximum level; about 12.5%.
P. S. There is an extra adjustment: as with every DSLR AFAIK, the minimum normal EI setting of the Nikon DF used in that test (100) is a bit higher than the SSat (75, according to DXO). That is an extra 1/3 stop, so raw placement will be 1/2 + 1/3 = 5/6 stop lower than JPEG placement.
-
Reflective spot meters are not dumb when the photographer knows how to use them. Same with incident meters! Film or digital.
As usual, you do not read...or maybe listen....or maybe understand what others are saying.
That is insulting and inappropriate. If you wish to argue, do so coherently and politely.
Jeremy
-
That is insulting and inappropriate. If you wish to argue, do so coherently and politely.
Jeremy
What? Nobody screamed yet for this section to be closed down too?
:-)
-
What? Nobody screamed yet for this section to be closed down too?
:-)
Polite is all that is required. Rude isn't.
Dave S
-
If you use a Canon camera supported by Magic Lantern firmware, then you can view a Live raw histogram on the display; the firmware also has a utility for automatically exposing to the right, or optimizing the raw exposure. It is called Auto ETTR in ML.
https://www.magiclantern.fm
Auto ETTR is a "module" that the user needs to enable within the Magic Lantern interface.
There is really no guesswork once you use a meter (your in-camera meter or handheld) and examine the raw data - spot meter highlights that you want to contain detail, shoot an exposure sequence and examine the raw files in Raw Digger to see where sensor saturation occurs. You may be able to eek out slightly more exposure if you use a raw converter that can reconstruct highlights effectively, especially if they are close to neutral.
On my Canon 5DIV, I use a handheld spot meter, meter the brightest highlights where I want to preserve detail and expose three stops higher and it is "spot" on (ha!). I use a Sekonic 758-D that permits me to set up the second ISO button with a "filter factor" - I set it to read +3 stops and I get my exposure reading for the shot (for typical daylight lighting conditions).
With the 5DIV, in dual pixel mode, you can grab one more stop of highlights if you split the dual pixel raw into the two raw exposures, using the DPRSplit utility, and merge the two resulting DNGs into a single pseudo-HDR file.
https://www.fastrawviewer.com/DPRSplit
You can merge them into a single DNG using ACR/Lightroom or Anders Torger's LumaRiverHDR, if DNG is a better format than EXR, etc.
kirk
-
On my Canon 5DIV, I use a handheld spot meter, meter the brightest highlights where I want to preserve detail and expose three stops higher and it is "spot" on (ha!).
That is a useful real-world observation. What ISO speed setting are you using on the camera, and on the light meter? Because three stops is very close to the expected gap between spot metered highlights and raw placement at base ISO speed.
-
That is a useful real-world observation. What ISO speed setting are you using on the camera, and on the light meter? Because three stops is very close to the expected gap between spot metered highlights and raw placement at base ISO speed.
It has been a few years since I left Canon for m43, but, as I remember, the fine single point on the 5D3 was ~1.5°, which work great for spot metering...and 3 stops was the gap...2.5 if you want to be safe.
I would always advise for the user to check their camera and metering with a RAW histogram such as RawDigger. Particularly if using an external light meter, which may not be calibrated to the camera light meter.
Edit: actually, Canon states the spot meter is ~1.5% of viewfinder. I found this sufficient for my purposes. The area of the meter is not the circle, but as shown below.
-
There are a couple of aspects of this concept of ETTR or 'optimal exposure' which I find very relevant for my own photography, which I'll list below.
(1) How much time does one have, before losing the shot, to employ techniques like taking a spot meter reading of the brightest part of the screen, then calculating an increase of exposure by 3 stops, or 2.75 stops, or 2.5 stops, depending on camera model, or examining the camera's histogram of a jpeg image and making a guess as to the optimal exposure?
(2) How anally retentive, or fanatical, does one need to be in order to get a satisfactory dynamic range which does not include blown highlights?
In answer to the first question, I decided years ago, when using Canon equipment, that it was much better to auto-bracket exposures for each shot, then later in Adobe Camera Raw select the shot that was best. Sometimes the best shot was not the one with the optimal exposure, but the underexposed shot which just happened to catch the best moment, when there was movement in the scene, and was the sharpest because of the faster shutter speed. That was a bonus. Also, if the scene was static, one could always merge to HDR in Photoshop.
In answer to the second question, there is less reason to be anally retentive now that recent models of DSLRs have significantly improved DR, especially if one is using Nikon cameras of course. ;)
Nevertheless, still being just a little bit fanatical myself, I do my best to avoid ruining shots with blown highlights which might contain detail relevant to the composition, such as amazing clouds in a bright sky.
The way I do this, which I've mentioned before, so I hope I'm not being too tedious, is to manually set exposure whilst viewing the camera's metering system at the bottom of the optical viewfinder.
On my D810, the exposure wheel is behind the shutter button and can be easily adjusted with my thumb without taking my eye away from the viewfinder.
I always use a single focusing square, which is also connected to the camera's metering system, and is activated with the AF-On button. The camera produces a meter reading which relates to whatever part of the scene is covered by the single focusing square, but this is not 'spot metering'.
On the D810, I don't even have to press a button to get a reading. I simply move the focusing square to any part of the scene in the composition, by swinging the camera rather than tediously moving the focusing square with the joystick. The exposure reading is shown at the bottom of the viewfinder.
The exposure reading scale in the viewfinder stretches from a minus sign on the extreme left, to a plus sign on the extreme right, and a zero in the center. If the exposure reading exceeds either extreme of plus or minus, a white arrow head appears at the extreme right or extreme left.
The issue that needs to be determined by anyone using such a method with a Nikon camera, is what reading at the foot of the viewfinder equates to an ETTR exposure, and also in relation to what RAW converter.
I use Adobe Camera Raw, and that's the only histogram I'm concerned with, at least initially. Consistency is the name of the game. Through basic, practical experiments, I have determined that an exposure reading to the far right of the D810's metering system, when the focusing square is positioned over the brightest part of the image (relevant to the composition, of course), is the correct exposure for an ETTR shot.
This is consistent for every shot, but only for my D810. When I use my Nikon D5300 with walk-around zoom lens, I have to adjust the exposure manually so that the meter reading in the viewfinder is only about 2/3rds of the way towards the extreme right. Also, the D5300 does not give a reading without pressing the AE-L/AE-F button, so one is actually refocusing whenever taking an exposure reading of the bright part of the scene.
However, I do understand that all this is not necessarily relevant for those who are not using Nikon cameras, except in so far as you do need to work out what is the best and/or quickest way of getting an optimal exposure with your particular model of camera, if you shoot in RAW mode.
The method I've outlined does cause a delay, compared with auto-focus in conjunction with auto-exposure, but a delay of only 3 or 4 seconds, and less when you've already made general adjustments for the lighting conditions of the environment.
-
Ray, you seeem to provide a second data point for a simple rule of thumb: “place the brightest relevant highlights at +3”. And also a strategy that makes a guideline like this – once adapted to particular camera – good enough in many cases: exposure bracketing, which can be automated and so done in a fraction of a second.
But let me check two things.
1) is the right end of that OVF light meter at +3? [UPDATE: yes, I checked, and the range on the D810's exposure indicator is the usual -3 to +3]
2) Is this at the D810’s minimum normal ISO dial setting, EI=64?
-
There are a couple of aspects of this concept of ETTR or 'optimal exposure' which I find very relevant for my own photography, which I'll list below.
(1) How much time does one have, before losing the shot, to employ techniques like taking a spot meter reading of the brightest part of the screen, then calculating an increase of exposure by 3 stops, or 2.75 stops, or 2.5 stops, depending on camera model, or examining the camera's histogram of a jpeg image and making a guess as to the optimal exposure?
Common sense says you always take one or more shots to capture a scene before optimizing. This would be true whether the optimization was for exposure or taking from a different viewpoint.
-
ACR and Lightroom histogram will show actually blown highlights as not being blown. That is, it will attempt to “recover” highlights that have blown channels in them. This often works, but often doesn’t.
Ray, I would suggest you use a RAW histogram to prove this for yourself.
-
There are a couple of aspects of this concept of ETTR or 'optimal exposure' which I find very relevant for my own photography, which I'll list below.
(1) How much time does one have, before losing the shot, to employ techniques like taking a spot meter reading of the brightest part of the screen, then calculating an increase of exposure by 3 stops, or 2.75 stops, or 2.5 stops, depending on camera model, or examining the camera's histogram of a jpeg image and making a guess as to the optimal exposure
The answer is of course it depends. ;)
You're shooting something that warrants the time to bracket and the subject is stationary/fixed (not a portrait etc), no reason not to bracket. Or take time to perhaps spot meter.
There are times when that's impossible and you'd lose the shot. For example sports, wildlife, street photography etc. Case in point, as someone who shot the 1984 Olympics on film (transparency), in all kinds of differing settings, no such luxury. And no, 100% of what I captured (30-50 rolls a day) where not perfectly exposed. But transparency like digital or any other capture, requires (demands?) the photographer understand how his meter(s) work in conjunction with the film/ASA just as he/she does with digital. If you know you can move, say +1/3 to 1/2 over the meter's recommendation without blowing out highlights you don't wish to blow out, (even if 'optimal' exposure from spot metering based on your understanding of how your camera/meter works suggests more), you're +1/3 to 1/2 closer to that goal of optimal exposure for the data.
When I'm shooting fast and loose, I have my 5DMII set to compensate +1/3 to +1/2 KNOWING I'm closer to the so called 'right' (of a Histogram I totally ignore) and years of doing so have not produced issues, I'm simply getting better data doing so.
What I recommend is understanding the limitations of your sensor by viewing brackets under differing situations with RawDigger. There's absolutely no reason IMHO to view a camera Histogram that isn't raw as was the case before cameras had Histograms! And the media (transparency) needed to be exposed without perhaps 1/3 to 1/4 a stop. Snip tests, push/pull processing helped to a degree but only a tiny degree. IF this could be accomplished with transparency film, I can't fathom why it can't be done with digital. The first step is to ignore the camera Histogram when shooting raw or, as suggested for Canon shooters, Magic Lantern if you really, really HAVE to view a Histogram while shooting.
-
ACR and Lightroom histogram will show actually blown highlights as not being blown. That is, it will attempt to “recover” highlights that have blown channels in them. This often works, but often doesn’t.
Absolutely! And it only tells you about ProPhoto RGB linear and rendered data along with that highlight recovery. Only a raw Histogram actually tells you about the facts of exposure on the raw data.
ACR will attempt to rebuild highlights if one or two channels have clipped but one hasn't. And yes, how well this works is often iffy. Those channels did clip and RawDigger will show you this. Also, it has a really awesome feature where say you were over exposed by 1 stop. You see this with their clipping overlay. You can set it to show you what would clip had you exposed differently! It shows you what the result would be if say you were actually over exposed 1/2 stop or under exposed by the same amount or more. Super useful.
WHY all raw converters can't do this is beyond me. I've asked Adobe for a raw Histogram for many, many years.
-
You're shooting something that warrants the time to bracket and the subject is stationary/fixed (not a portrait etc), no reason not to bracket. Or take time to perhaps spot meter.
There are times when that's impossible and you'd lose the shot. For example sports, wildlife, street photography etc.
Of course, ETTR is also usually irrelevant to such situations for the simple reason that they require an elevated EI ("ISO speed") setting on the camera, and so one is usually not going to have any photosites receiving anything close to full well capacity. Let me say again that EXPOSING to the right is only relevant when the camera's lowest [normal] EI setting can be used. It is nothing to do with even the raw histogram at higher EI settings where the signal is amplified more, sending far less than full wells to maximum raw level. Maybe I will later post my take on the importance (or not) of worrying about raw histogram placement when at higher EI settings and with unavoidably suboptimal exposure of the photosites.
When I'm shooting fast and loose, I have my 5DMII set to compensate +1/3 to +1/2 KNOWING I'm closer to the so called 'right' (of a Histogram I totally ignore) and years of doing so have not produced issues, I'm simply getting better data doing so.
Yet another experienced photographer in this thread who has come to a similar rule of thumb to what I suggested for getting roughly optimal sensor exposure level, at least when "shooting fast and loose"!
What I recommend is understanding the limitations of your sensor by viewing brackets under differing situations with RawDigger.
Yes, probably one good way to calibrate the information that one is able to get in the field when no "raw histogram" is available.
There's absolutely no reason IMHO to view a camera Histogram that isn't raw as was the case before cameras had Histograms!
Here I disagree: subject to further testing and evidence, I am rather confident that there is a simple and easily-determined approximate relationship between the camera's JPEG histogram (at default settings for contrast etc.) and what the raw histogram would tell you, so the camera's histogram can probably get you within 1/2 stop of perfection once one adds a suitable shift to what that histogram suggests. Not perfect due to issue like one color being blown out while the JPEG luminosity, but a good approximation. (And maybe even the camera's overexposure "zebras" or "blinkies" convey enough information about the right end of the JPEG histogram, if one knows how to interpret them.)
Of course, I sometimes forget that not everyone has access to either a JPEG histogram or overexposure blinkies in the viewfinder while composing; increasing the need for spot metering or such.
And the media (transparency) needed to be exposed without perhaps 1/3 to 1/4 a stop. Snip tests, push/pull processing helped to a degree but only a tiny degree. IF this could be accomplished with transparency film, I can't fathom why it can't be done with digital.
Fortunately, good modern sensors are vastly more tolerant than transparency film, with its DR as low as five stops in the case of Velvia, so I see no practical reason to worry about getting within 1/3 stop or less of "optimal", as far as visual IQ is concerned.
-
Of course, ETTR is also usually irrelevant to such situations for the simple reason that they require an elevated EI ("ISO speed") setting on the camera, and so one is usually not going to have any photosites receiving anything close to full well capacity. Let me say again that EXPOSING to the right is only relevant when the camera's lowest [normal] EI setting can be used.
The key word is usually. Depends on the camera no? Case in point, higher ISO produces less noise:
(http://digitaldog.net/files/100vs800iso.jpg)
Here I disagree: subject to further testing and evidence, I am rather confident that there is a simple and easily-determined approximate relationship between the camera's JPEG histogram (at default settings for contrast etc.) and what the raw histogram would tell you, so the camera's histogram can probably get you within 1/2 stop of perfection once one adds a suitable shift to what that histogram suggests.
That may be very true. My point is, it's not necessary to have any Histogram to produce optimal exposure. Wasn't with film which it appears some agree here was more demanding in terms of nailing exposure. Don't get me wrong, I'd LOVE a raw Histogram on the camera in some cases. I'd love one in the raw converter. I don't need one in the former. I was classically trained to expose transparency film many years before cameras had Histograms.
Fortunately, good modern sensors are vastly more tolerant than transparency film, with its DR as low as five stops in the case of Velvia, so I see no practical reason to worry about getting within 1/3 stop or less of "optimal", as far as visual IQ is concerned.
Unless there is zero visible difference between optimal exposure for raw and an exposure that is (x amount off), and I'd love to see demonstrations of this, it isn't the case with my old 5DMII (time for a Sony?), then I'll strive for optimal exposure and image quality.
-
The key word is usually. Depends on the camera no? Case in point, higher ISO produces less noise:
Ah yes, that was a significant issue with the earlier generations of CMOS sensors as in that Canon 5D Mk II. This seems to be vastly less significant with the newer on-chip column-parallel ADC technology. Briefly, because now virtually all noise enters _before_ analog gain is applied, at least once the camera is more than one or two stops above minimum EI setting, so that the noise in the analog signal is being amplified to well above the noise floor of the ADC process. Anyway, the strategy is different from EXPOSE TTR then: one is not adjusting exposure, but accepting the limits on how much exposure can be delivered to the photosites, and adjusting the ISO speed setting on the camera (adjusting the analog gain) to get a desirable raw histogram placement.
That may be very true. My point is, it's not necessary to have any Histogram to produce optimal exposure. Wasn't with film which it appears some agree here was more demanding in terms of nailing exposure. ... I was classically trained to expose transparency film many years before cameras had Histograms.
I agree on not needing a histogram! It is just one option for measuring where the highlights are, so you can then using knowledge about what that says where they will be "placed" on the sensor, relative to full well capacity. (I am happy with blinkies; Ray with spot metering on highlights and placing them at +3.)
One thing is actually easier with transparency film than with the pursuit of "maximum exposure without blown highlights": the goal with film is to meter a certain part of the scene and select the exposure that places that part of the scene where you want it in the developed film. That is more akin to choosing JPEG placement, and is dealing only with things that you or the light meter can directly see. There are no "invisible targets", as there are when people want to minimize the amount of raw file highlight headroom is being "left on the table".
-
Ray, you seeem to provide a second data point for a simple rule of thumb: “place the brightest relevant highlights at +3”. And also a strategy that makes a guideline like this – once adapted to particular camera – good enough in many cases: exposure bracketing, which can be automated and so done in a fraction of a second.
But let me check two things.
1) is the right end of that OVF light meter at +3? [UPDATE: yes, I checked, and the range on the D810's exposure indicator is the usual -3 to +3]
2) Is this at the D810’s minimum normal ISO dial setting, EI=64?
+3 from what base, BJL? Do you mean, from the zero point in the middle of the metering scale, or from an average exposure of the scene when the camera is in autoexposure mode?
Such considerations are of no practical significance with my method. My camera is set to manual exposure. I adjust the shutter speed by turning a wheel with my thumb until the camera's meter reading reaches the far right of the metering scale when the single focusing square covers (what I estimate to be) the brightest part of the scene in which I want to retain detail or color. In a landscape scene, that part of the composition which is the brightest is usually the clear blue sky or clouds. No calculations are required.
However, if the sky has many similarly bright patches of clouds, or white walls in sunlight, and I don't have the time to swing the focusing square from one bright patch to another to measure which is the brightest patch, because there are kangaroos in the foreground which might soon hop away (for example), then I'll use a shutter speed which is slightly faster than required for a far right meter reading.
If the scene does not have a high brightness range, and/or the situation calls for an appropriately fast shutter speed which is of more concern than an ETTR exposure, then I will often take advantage of the 'close to ISO-invariant' nature of the Nikon D810, and other Nikon models. Instead of raising ISO a stop or two, I'll just underexpose. The loss of DR is very minimal.
For example, if one underexposes 2 stops at ISO 100, instead of using the same exposure at ISO 400, on the D810, one loses approximately 1/4th of a stop of DR. If one underexposes by 4 stops at ISO 100, instead of raising ISO to 1600 to produce an ETTR shot with the same exposure, one loses approximately 1/3rd of a stop of DR. No big deal, except for the anally retentive. ;D
Out of curiosity, I've just tested the range of the metering scale in the viewfinder of my D810. It's always 4 stops from the extreme left to the extreme right with the zero point in the middle, whatever the ISO setting.
-
For example, if one underexposes 2 stops at ISO 100, instead of using the same exposure at ISO 400, on the D810, one loses approximately 1/4th of a stop of DR.
If one underexposes by 4 stops at ISO 100, instead of raising ISO to 1600 to produce an ETTR shot with the same exposure, one loses approximately 1/3rd of a stop of DR. No big deal, except for the anally retentive. ;D
I understand the first sentence but not the second, specifically: instead of raising ISO to 1600 to produce an ETTR shot.
-
I understand the first sentence but not the second, specifically: instead of raising ISO to 1600 to produce an ETTR shot.
The meaning of ETTR is 'Expose To The Right' of the histogram. The letter R in ETTR must relate to a histogram, not to a full well situation in the original exposure before the camera has applied analogue gain. If you want to exclude the use of the expression ETTR from high ISO situations, then a different acronym should be chosen, such as ETFWCABISO (Expose To Full Well Capacity At Base ISO). ;D
-
Common sense says you always take one or more shots to capture a scene before optimizing. This would be true whether the optimization was for exposure or taking from a different viewpoint.
The procedure I've described above, which might apply only to certain Nikon cameras, removes the need to take more than one shot and fuss around examining histograms.
As I mentioned, on the D810 one doesn't even need to press a button to get an exposure reading in the viewfinder. One can just swing the camera around until the single focusing square has covered every bright area in the composition that one has noticed. It takes just a few seconds.
If one discovers that a particular bright patch (of clouds for example) causes the meter reading to exceed the far right, one simply increases the shutter speed by another 1/3rd stop or whatever. It works every time for me.
-
The meaning of ETTR is 'Expose To The Right' of the histogram. The letter R in ETTR must relate to a histogram, not to a full well situation in the original exposure before the camera has applied analogue gain. If you want to exclude the use of the expression ETTR from high ISO situations, then a different acronym should be chosen, such as ETFWCABISO (Expose To Full Well Capacity At Base ISO). ;D
I understand what ETTR means. How was the first exposure different from the second other than 2 vs. 4 stops under? Raising ISO for ETTR?
-
I understand what ETTR means. How was the first exposure different from the second other than 2 vs. 4 stops under? Raising ISO for ETTR?
The difference is in the appearance of the histogram. An underexposure at any ISO setting results in the histogram, whether the camera's histogram or Adobe Camera Raw's histogram, being more to the left. This situation is not advisable when using Canon cameras. An ETTR at ISO 200 or 400 always has significantly better dynamic range than the same exposure used at ISO 100. This is where the Nikon cameras have an advantage.
However, at much higher ISOs, such as 1600, 3200 and 6400, the loss of DR resulting from a 2 stop underexposure at ISO 1600, as opposed to an ETTR exposure at ISO 6400, is of little significance with both Canon and Nikon cameras.
-
ACR and Lightroom histogram will show actually blown highlights as not being blown. That is, it will attempt to “recover” highlights that have blown channels in them. This often works, but often doesn’t.
Ray, I would suggest you use a RAW histogram to prove this for yourself.
The only histogram I use is the one in Adobe Camera Raw. Before adjustments are made, an ETTR shot will usually appear to have blown highlights, especially if the composition includes a bright and cloudy sky. However, moving the 'highlights' slider to the left, and/or the 'whites' slider to the left, and/or the 'exposure' slider to the left, will always fix that appearance of blown highlights, except when the right of the histogram has a significant spike or 'cliff edge', which means one has really overexposed.
My method of exposing to the right of the camera's meter, does not result in significant spikes or cliff edges at the right of Adobe's histogram.
Photographs are all about appearance. If the adjusted image in Photoshop doesn't look as though it has any blown highlights, then that's all that matters for me. The fact that some other more accurate RAW histogram reveals that one or more channels are in fact clipped, is of little practical significance for me, although I understand it might be very significant for those who design cameras and software.
-
The meaning of ETTR is 'Expose To The Right' of the histogram. The letter R
Yes, but which histogram? Not any JPEG histogram of course, but also not any raw histogram at an elevated EI setting. The primary rationale for the method is adjusting exposure in the sense of how much light is received by the sensor, and thus the histogram referred to is the one reporting the level of exposure received by the photosites, with the right being the line corresponding to a full well. That is, it is about choosing the maximum Exposure Index that avoids overfilling any photosite.
It is time to discuss the big difference between moving this "exposure histogram" to the right (by increasing exposure time, reducing aperture ratio, or adding light to the scene) versus moving the raw histogram by increasing the "ISO speed" and thus the analog gain applied before ADC. It helps to break noise sources into three main parts:
1) Photon shot noise in the light reaching the sensor: "shot noise".
2) Noise produced in the sensor before any amplification is applied, such as dark current noise: "pre-amp noise".
3) Noise introduced after analog gain is applied; I believe that these days this is mostly quantization noise in the ADC: "post-amp noise".
There might also be something intermediate:
2.5) Noise introduced during the amplification stage.
How is the local signal to noise ratio at any part of the image affected by increasing actual sensor exposure (Exposure Index) versus changing the gain, as done by increasing the ISO speed setting?
- A one stop increase in exposure doubles the signal (photo-electron count), which increases photon shot noise by sqrt(2), and does not change the other noise sources. In all but the darkest parts of the image, photon shot noise dominates noise, so SNR improves by 2/sqrt(2) = sqrt(2): a half stop gain. In very underexposed parts of the image, where the sensor illumination is say ten stops or twelve or more stops below FWC, noise produced in the camera might become dominant, so there the SNR improvements can be more, sliding up to a full stop. This is what "engineering dynamic range measures", so gains in that measure can be misleadingly good, compared to the SNR improvements in most parts of the image.
- a one stop increase in analog gain (same exposure to the sensor) doubles the numerical values of the signal, the photon shot noise, and the pre-amp noise. These sources are again dominant in all but the darkest parts of the image, so there is no improvement in SNR there. However, in the darkest parts, that post-amp noise (mostly ADC quantization noise?) might be a significant part of all noise, at least until the "ISO gain" is enough to bring the shot plus pre-amp noise enough above that post-amp noise. So there can be a deep shadow SNR gain in the first few steps up the ISO speed range. Further up the scale, modern sensors enter an "ISO-less" range of EI settings (which I have seen starting at law as 400 or less and as high as 800 or even 1600) and then further analog gain has no significant SNR benefit, even if it moves the raw histogram further to the right.
- By the way, in some cameras ISO speed increases beyond some high setting apply further "ISO gain" digitally without increasing analog gain, using bit shifting or such. This of course has no SNR benefit at all; it just simplifies some steps in producing JPEGs.
I started some experiments today with my OM-D E-M5, comparing two images of the same subject, both at f/4 and 1/2000s, one "on meter" at EI=1600, the other at EI=200, so underexposed four stops and giving an almost totally black default JPEG conversion. After equalizing levels from the raw files in Lightroom, the noise levels were visibly indistinguishable. However, I need next to test on scenes with a wider subject brightness range, to check out the shadow handling.
-
An underexposure at any ISO setting results in the histogram, whether the camera's histogram or Adobe Camera Raw's histogram, being more to the left. This situation is not advisable when using Canon cameras.
That certainly used to be true, back when you were using Canon cameras. Is it still true now that Canon has adopted column-parallel ADC?
I agree with the rest of your post, except that when say "Nikon cameras", you probably mean more or less all cameras except (some) Canon ones.
-
+3 from what base, BJL? Do you mean, from the zero point in the middle of the metering scale, or from an average exposure of the scene when the camera is in autoexposure mode?
I mean that in manual mode, the exposure indicator scale at the bottom of the viewfinder is at its right limit, so the camera is suggesting that you are overexposing by three stops because it does not know that you are metering on the highlights, not the mid-tones.
Such considerations are of no practical significance with my method.
Quite so: all you need to know is that you have determined a predictable relationship between what that exposure indicator says and your desired exposure placement. I only ask so that I can use your evidence to test and refine my understanding of how the various measurements and outcomes are related. Partly because I am refining a similar approach, based instead on using highlight blinkies or the JPEG histogram in the EVF, so that I do not have to spot meter.
-
An ETTR at ISO 200 or 400 always has significantly better dynamic range than the same exposure used at ISO 100.
Considering that exposure is Aperture and Shutter alone, what has ISO and ETTR got to do with this? Sorry, I still don't understand what you are suggesting. Amplification in the processing?
-
Considering that exposure is Aperture and Shutter alone, what has ISO and ETTR got to do with this? Sorry, I still don't understand what you are suggesting. Amplification in the processing?
Yes: Ray takes ETTR to refer to the raw histogram after amplification, rather than the (hypothetical) sensor exposure histogram, so he is practicing and advocating what I call "amplify to the right". This was definitely of value with old-style Canon CMOS sensors, but see my long post above where I for one thing try to assess if and when there is any value to ATTR in the modern era of column-parallel ADC.
-
Considering that exposure is Aperture and Shutter alone, what has ISO and ETTR got to do with this? Sorry, I still don't understand what you are suggesting. Amplification in the processing?
Crikey! Andrew. Don't you believe in DXOMark graphs? Of course it's amplification in the processing; the camera processing in relation to the choice of ISO setting.
The attached graph shows the DR performance of the Nikon D810 compared with the Canon 5DSR and 5D Mk III.
If you underexpose by one stop at ISO 100, with the Canon cameras, you'll lose almost a full stop of DR, compared with using the same exposure of the same scene, from the same position, using the same F stop, at ISO 200.
If you underexpose by 2 stops at ISO 100, instead of an ETTR exposure at ISO 400 using the same shutter speed and f/stop, you'll lose almost 2 stops of DR. Isn't that an obvious deduction, from the graph?
-
Crikey! Andrew. Don't you believe in DXOMark graphs? Of course it's amplification in the processing; the camera processing in relation to the choice of ISO setting.
No but thanks for clarifying what you're referring to (that's why I asked).Thanks to BJL for doing so first. ;)
-
I've just stepped outside onto the deck at the rear of the house, and taken an ETTR shot in relation to ROTM (Right Of The Meter). ;)
As you can see from the attached image, plus an enlargement of the histogram in Adobe Camera Raw before any adjustments have been made, it's not quite an ETTR exposure.
In this particular instance, I could have given at least 2/3rds of a stop more exposure before clipping any of the channels, so my general procedure of exposing to the far right of the camera's metering scale is a conservative approach to ensure that the brightest parts of the composition which I might have missed due to a lack of time to check every bright patch in the scene, are not blown.
In this particular shot, the brightest part of the scene is easily discernible, so I've really nailed it, as the histogram shows.
Since I'm not anally retentive, and since I use a camera with an extremely high DR rating, I'm not at all concerned about sacrificing a mere 2/3rd's of a stop of DR. ;)
-
I've just stepped outside onto the deck at the rear of the house, and taken an ETTR shot in relation to ROTM (Right Of The Meter). ;)
As you can see from the attached image, plus an enlargement of the histogram in Adobe Camera Raw before any adjustments have been made, it's not quite an ETTR exposure.
You don’t seem to recognize that indeed there is an adjustment in ACR.You don’t seem to recognize that the histogram shown in ACR tells you very little if anything about the actual exposure. You don’t seem to recognize this histogram is a rendered histogram of the current editing settings. Move exposure slider or highlight all the way to the right. Or left! How’s that histogram look? How does that Histogram correlate to your exposure of the raw data? It does not! Now try Rawdigge.
-
You don’t seem to recognize that indeed there is an adjustment in ACR.You don’t seem to recognize that the histogram shown in ACR tells you very little if anything about the actual exposure. You don’t seem to recognize this histogram is a rendered histogram of the current editing settings. Move exposure slider or highlight all the way to the right. Or left! How’s that histogram look? How does that Histogram correlate to your exposure of the raw data? It does not! Now try Rawdigge.
Just to add to what Andrew says, LR/ACR has significant highlight compression and roll off. My tests, years ago, on 5D3, showed that almost all highlights with rendered highlights above 96-97% showed at least one blown channel when looked at in RawDigger.
-
... I am rather confident that there is a simple and easily-determined approximate relationship between the camera's JPEG histogram (at default settings for contrast etc.) and what the raw histogram would tell you, so the camera's histogram can probably get you within 1/2 stop of perfection once one adds a suitable shift to what that histogram suggests. Not perfect due to issue like one color being blown out ...
BJL, - there were some excellent posts by Bill / bjanes in previous discussions on this subject.
I'm sure you are aware, however, I thought it is worthwhile to remember it here:
http://forum.luminous-landscape.com/index.php?topic=96523.msg788460#msg788460
http://forum.luminous-landscape.com/index.php?topic=96523.msg788730#msg788730
http://forum.luminous-landscape.com/index.php?topic=68334.msg587241#msg587241
Peter
--
-
You don’t seem to recognize that indeed there is an adjustment in ACR.You don’t seem to recognize that the histogram shown in ACR tells you very little if anything about the actual exposure. You don’t seem to recognize this histogram is a rendered histogram of the current editing settings. Move exposure slider or highlight all the way to the right. Or left! How’s that histogram look? How does that Histogram correlate to your exposure of the raw data? It does not! Now try Rawdigge.
Of course I realize that there are thousands of adjustments that have been made from the time I press the shutter button to the time the file is demosaiced in Adobe Camera Raw. Those adjustments are in accordance with the camera design and the software design. I have little control over them.
ACR is the software I've been using for many years. If I were dissatisfied with it, I'd switch to other software, just as I switched from Canon to Nikon many years ago. The D700 was my first Nikon camera which replaced my Canon 5D
As I've mentioned before, if nothing appears to be blown in the highlights and nothing appears to be wrong or odd, then it's of no concern to me. If it ain't broke then don't fix it.
Also, how could you possibly assume that anyone would use ACR without making adjustments with the sliders. I move various sliders from left to right every time I process a RAW image. I obviously must know how it changes the histogram. Did you miss the point that the image in my previous post was to show how it looked before I'd made any adjustments.
Sometimes I'll move the exposure slider to the right, even though the image does not look underexposed, and compensate for it by moving the highlights and whites sliders to the left, because that has the desired effect on other parts of the image.
My general procedure is to process the image first in ACR, moving the appropriate sliders, and making other adjustments till the image looks right, then fine tune the image after opening in Photoshop in 16 bit mode using the ProPhoto RGB color space. Okay?
-
Of course I realize that there are thousands of adjustments that have been made from the time I press the shutter button to the time the file is demosaiced in Adobe Camera Raw. Those adjustments are in accordance with the camera design and the software design. I have little control over them.
The discussion was exposure and Histograms, you provided some in ACR. It doesn't tell us the truth about the raw exposure.
As I've mentioned before, if nothing appears to be blown in the highlights and nothing appears to be wrong or odd, then it's of no concern to me. If it ain't broke then don't fix it.
Despite the possibility of over exposing your data.
-
Despite the possibility of over exposing your data.
Absolutely! My method of correctly exposing for the highlights, as I've described in great detail, ensures that there are no visible effects of blown highlights in ACR, which is the program I'm accustomed to.
Invisible effects of blown highlights, which might only be apparent in Rawdigger, are of little concern. Why should they not be? I'm not in the job of designing software for RAW files. If Adobe has succeeded in repairing, correcting or even obscuring any slight effects of blown highlights, then 'good on them' I say. Well done! ;D
-
Absolutely! My method of correctly exposing for the highlights, as I've described in great detail, ensures that there are no visible effects of blown highlights in ACR, which is the program I'm accustomed
If you over exposed, (you admit the possibility) You didn’t expose correctly as you view another lie of a Histogram: Going back to my first recommendation: use the right tool for the right job if you must have to view a Histogram to property expose (photography 101). That isn’t ACR! It’s Rawgigger; no lie, just the facts Man 😋
-
If you over exposed, (you admit the possibility) You didn’t expose correctly as you views another lie of a Histogram: Going back to my first recommendation: use the right tool for the right job if you just have to few a Histogram to property expose (photography 101). That isn’t ACR! It’s Rawgigger; no lie, just the facts Man 😋
Andrew,
I'm aware of the effects of overexposure, just as I'm aware of noise and banding in deep shadows, with some models of camera. Both the Canon 5D and the Nikon D7100 upset me because I could see banding in the shadows, although the banding in the D7100 shots was only visible in much deeper shadows than in the 5D shots.
I'm getting the impression, if I were to use Rawdigger instead of ACR, I would tend to use a faster shutter speed to avoid blowing highlights which were not visibly blown in ACR. Is that correct?
If I were to use a faster shutter speed, all else being the same, the noise in the shadows would increase. Is that correct?
-
You do make things difficult ::)
Getting the right exposure has not changed in 60 years. I just shot the view up my road using British Standard Exposure Tables B.S.935/1957 (incorporated in the handy calculator shown in picture below) to estimate the exposure. The first picture shows result in fast RAW viewer. It could perhaps have stood a 1/3rd stop more exposure for absolute perfection but...
-
Andrew,
I'm aware of the effects of overexposure, just as I'm aware of noise and banding in deep shadows, with some models of camera. Both the Canon 5D and the Nikon D7100 upset me because I could see banding in the shadows, although the banding in the D7100 shots was only visible in much deeper shadows than in the 5D shots.
I'm getting the impression, if I were to use Rawdigger instead of ACR, I would tend to use a faster shutter speed to avoid blowing highlights which were not visibly blown in ACR. Is that correct?
If I were to use a faster shutter speed, all else being the same, the noise in the shadows would increase. Is that correct?
Don’t you have RawDigger and can’t you like others posting here who do, run your own tests on your own sensors?????
-
BJL, - there were some excellent posts by Bill / bjanes in previous discussions on this subject.
I'm sure you are aware, however, I thought it is worthwhile to remember it here:
http://forum.luminous-landscape.com/index.php?topic=96523.msg788460#msg788460
http://forum.luminous-landscape.com/index.php?topic=96523.msg788730#msg788730
http://forum.luminous-landscape.com/index.php?topic=68334.msg587241#msg587241
Peter
--
Thanks! I do somewhat recall some of those conversations, with input from experts like Bill Janes, but I am sure that I can benefit from reading more.
For one thing, I realize that my previous description was naive about the shape of the tone curve underlying a camera's default JPEG conversions and its displayed histogram. (Is it a safe assumption that when one is shooting for the raw files, the camera's JPEG settings can be at defaults?) Even if midtones are placed close to the standards-specified JPEG level 112, about 2.5 stops below maximum, the defaults tone curves roll off a bit above that, so that the maximum JPEG level corresponds to scene brightness more than 2.5 stops above the midtones.
-
For one thing, I realize that my previous description was naive about the shape of the tone curve underlying a camera's default JPEG conversions and its displayed histogram. (Is it a safe assumption that when one is shooting for the raw files, the camera's JPEG settings can be at defaults?) Even if midtones are placed close to the standards-specified JPEG level 112, about 2.5 stops below maximum, the defaults tone curves roll off a bit above that, so that the maximum JPEG level corresponds to scene brightness more than 2.5 stops above the midtones.
Considering we can pick between sRGB which doesn't have a gamma curve but a TRC and Adobe RGB (1998) which has a gamma curve, one could assume (and I hate to do so), there would be small differences. Another reason why the JPEG Histogram, a lie, isn't all that useful:
Everything you thought you wanted to know about Histograms
Another exhaustive 40 minute video examining:
What are histograms. In Photoshop, ACR, Lightroom.
Histograms: clipping color and tones, color spaces and color gamut.
Histogram and Photoshop’s Level’s command.
Histograms don’t tell us our images are good (examples).
Misconceptions about histograms. How they lie.
Histograms and Expose To The Right (ETTR).
Are histograms useful and if so, how?
Low rez (YouTube): http://www.youtube.com/watch?v=EjPsP4HhHhE (http://www.youtube.com/watch?v=EjPsP4HhHhE)
High rez: http://digitaldog.net/files/Histogram_Video.mov (http://digitaldog.net/files/Histogram_Video.mov)
-
You do make things difficult ::)
Getting the right exposure has not changed in 60 years. I just shot the view up my road using British Standard Exposure Tables B.S.935/1957 (incorporated in the handy calculator shown in picture below) to estimate the exposure. The first picture shows result in fast RAW viewer. It could perhaps have stood a 1/3rd stop more exposure for absolute perfection but...
For what is being discussed here (ETTR or optimal raw exposure) that shot is at least 1 to 1-1/3 stops underexposed (eyeballing the histogram), according to the raw histogram. That is, the green channel's brightest highlight (probably in the non-specular snow) falls to zero frequency well below sensor clipping such that the exposure could be increased by over 1 stop and still not produce sensor clipping in areas of meaningful detail. If the deep shadows were of concern in this image, you could probably push the highlights in the snow even more and rely upon highlight compression and reconstruction to bring the neutral highlights back from clipping from the unclipped red and blue channels, giving the shadows even more signal to work with. How much more would depend upon the raw converter and the various highlight reconstruction strategies they employ. Exposing for a particular raw converter's capability is a bit of a gamble - it may give you extra exposure to work with now, but if you ever change converters you may have to deal with the consequences of pushing your luck a little too far if that other converter does not employ the same highlight reconstruction strategy. This is why the raw histogram is so useful.
During raw conversion the exposure would need to be pulled back to reestablish the tonal representation of the scene, but that is part of using the ETTR technique.
kirk
-
For what is being discussed here (ETTR or optimal raw exposure) that shot is at least 1 to 1-1/3 stops underexposed (eyeballing the histogram), according to the raw histogram. That is, the green channel's brightest highlight (probably in the non-specular snow) falls to zero frequency well below sensor clipping such that the exposure could be increased by over 1 stop and still not produce sensor clipping in areas of meaningful detail. If the deep shadows were of concern in this image, you could probably push the highlights in the snow even more and rely upon highlight compression and reconstruction to bring the neutral highlights back from clipping from the unclipped red and blue channels, giving the shadows even more signal to work with. How much more would depend upon the raw converter and the various highlight reconstruction strategies they employ. Exposing for a particular raw converter's capability is a bit of a gamble - it may give you extra exposure to work with now, but if you ever change converters you may have to deal with the consequences of pushing your luck a little too far if that other converter does not employ the same highlight reconstruction strategy. This is why the raw histogram is so useful.
During raw conversion the exposure would need to be pulled back to reestablish the tonal representation of the scene, but that is part of using the ETTR technique.
kirk
In practice I use ETTR on my X-Pro1. On the X-Pro 1 you can set the camera up to provide a soft contrast JPEG pre-view image and thus a histogram that reveals the most information about dynamic range for RAW files. The settings I use are
Dynamic Range - DR100%
Film Simulation - Provia
Highlight Tone - (-2) Soft
Shadow Tone - (-2) Soft
These internal RAW developer settings affect the histogram seen in viewfinder even though I am shooting RAW only.
I see this somewhat optimised histogram in the viewfinder and can adjust the exposure compensation dial with my right thumb to move the histogram to the right (or left if the meter is overexposing).
BTW you were right - one stop under exposed.
Here is the picture after correction in Capture One and it could have done with that extra 1 stiop to bring out detail in the shrub across the road. :-[
-
During raw conversion the exposure would need to be pulled back to reestablish the tonal representation of the scene, but that is part of using the ETTR technique.
kirk
Michael called it "Normalization" and I think it is still a good name. Some folks open a raw exposed optimally for raw, they believe there are 'no settings' applied when there always is and this default setting is utterly wrong for this kind of capture. It's like in the 'old days' when people would state that Linear encoded captures were dark. Yeah, if you viewed them incorrectly with a gamma corrected assumption, they looked awful. Assign the correct ICC profile that defines this capture, it doesn't appear dark at all (unless the photographer didn't expose correctly)! :o I recall some Kodak DCS cameras that produced this option and have such captures in archive. They needed 'normalization' too. The data was just fine.
-
Crikey! Andrew. Don't you believe in DXOMark graphs? Of course it's amplification in the processing; the camera processing in relation to the choice of ISO setting.
The attached graph shows the DR performance of the Nikon D810 compared with the Canon 5DSR and 5D Mk III.
Those are some rather old cameras, form before Canon adopted column-parallel ADC, so here are some more up-to0-date comparisons. I use the "print" graphs, rescaled to compensate for different pixel counts. (This is even after I momentarily put aside DXO's outright error in the horizontal scale (through its confounding the upper limit of raw files with Exposure Index), and the roughly counteracting adjustment needed to discount measures of highlight headroom and instead assess shadow handling.)
There is still the dubious relevance of the bottom few stops of that "engineering" dynamic range, going down to a photographically useless SNR=1. One way to dig our some information about SNR in the deep shadows is to look at DXO's "SNR 18%" graphs. Directly, they refer to SNR in the "mid-tones" at the stated EI, but also, the values at one high EI roughly reflect SNR at a level "x" stops lower when using an EI x stops lower. For example, SNR 18% at EI=25,600 is from photosites getting the same illumination as photosites 5 stops below mid-tones [about 8 below the top of the DR] in exposure at EI=800, and those 8 stops below mid-tones [about 11 below the top of the DR] at EI=100.
Then all three cameras perform very similarly, hitting the 20dB level (which corresponds to SNR = 10:1, a guideline for minimum tolerable SNR as embodied in the ISO SNR10:1 measure of shadow handling) at EI between 25,600 and 51,200. This suggests to me — but by no means proves! — that the differences in the DXO DR graphs are mostly due to different behavior in the irrelevant lower stops where SNR is too low to be "photographically useful".
-
Considering we can pick between sRGB which doesn't have a gamma curve but a TRC and Adobe RGB (1998) which has a gamma curve, one could assume (and I hate to do so), there would be small differences. Another reason why the JPEG Histogram, a lie, isn't all that useful:
Everything you thought you wanted to know about Histograms
Another exhaustive 40 minute video ...
Maybe I'll find the time for 40 minutes of exhaustion later! For now my goal is to find a JPEG histogram that lies "consistently". If I know that the JPEG histogram on a particular camera at particular settings overstates exposure at the right end by around 2/3 stop, I can exceed its exposure recommendations by a bit less — say 1/3 stop — and bracket a bit where possible.
Or maybe I just go by the histogram when enough bracketing is possible: "Trust, but verify".
-
Bracket, YES!
-
Those are some rather old cameras, form before Canon adopted column-parallel ADC, so here are some more up-to0-date comparisons. I use the "print" graphs, rescaled to compensate for different pixel counts. (This is even after I momentarily put aside DXO's outright error in the horizontal scale (through its confounding the upper limit of raw files with Exposure Index), and the roughly counteracting adjustment needed to discount measures of highlight headroom and instead assess shadow handling.)
Wow! I didn't realize that Canon had improved it's DR so much in its latest 5D Mk IV, it's now only about 1.25 EV behind the latest Nikon FX, at base ISO, and only 1/2 a stop below the D850 at ISO 400. Well done, Canon. ;D
However, I don't think I'll be buying a 5D Mk IV, despite still having a number of Canon lenses sitting on the shelf, because the pixel count of the D850 is about 50% greater, which improves the resolution of all lenses used.
There is still the dubious relevance of the bottom few stops of that "engineering" dynamic range, going down to a photographically useless SNR=1. One way to dig our some information about SNR in the deep shadows is to look at DXO's "SNR 18%" graphs.
Not at all. SNR at 18% is approximately the level of the average skin tone in a face. It doesn't tell you anything about the shadows. Most cameras of the same format size have very similar 'SNR at 18%' values. MF digital cameras tend to have slightly better 'SNR at 18%' than FF 35mm, and FF 35mm formats have better SNR than APS-C or cropped formats, whether Nikon or Canon.
However, if one compares cameras of significantly different ages, the progression of technology can result in a smaller format having as good an SNR as a larger format. For example, the cropped format Nikon D7200 has 'SNR at 18%' values at least as good, and slightly better at base ISO, than the FF Canon 5D. Of course, the DR of the smaller format D7200 is massively better than the Canon 5D, almost 3 & 1/2 stops better at base ISO.
The 'SNR at 18%' values of the Nikon D850, at base ISO, are a significant 6.5dB higher than the Olympus E-M1 MkII.
-
...
However, I don't think I'll be buying a 5D Mk IV, despite still having a number of Canon lenses sitting on the shelf, because the pixel count of the D850 is about 50% greater, which improves the resolution of all lenses used.
..
The resolving power of your lenses is fixed by their optics - compared to lower res sensors, the D850 sensor will provide a better chance of exposing the limits of those optics and the flaws in focus technique.
kirk
-
The 'SNR at 18%' values of the Nikon D850, at base ISO, are a significant 6.5dB higher than the Olympus E-M1 MkII.
SNR @ 18% is where shot noise already dominates other noises sources (those affecting deep shadows), so there sensors of similar generation & technology differs just by a die size... FF shall be 2 stops better than m43 being twice bigger (when equalized to the same mp and comparing more or less similar tech generations - which D850 and E-M1mkII are, both being Sony Semi tech)
-
For now my goal is to find a JPEG histogram that lies "consistently".
the answer to that question is known for a long time... you need to use UniWB, then only you have either consistent lie or better yet with some additional OOC JPG parameters you actually can nail clipping in raw and with zebra and/or blinkies see where it happens with <= 1/3 EV precision (which is good enough - because quite a lot of cameras might have ill effects near clipping in raw, so pushing further might not be worth the trouble).
-
Not at all. SNR at 18% is approximately the level of the average skin tone in a face. It doesn't tell you anything about the shadows.
You have completely overlooked the EI shift part of my comparison. Yes, the "SNR18" report for EI=200 is what you get at midtones when following the camera's metering set at that EI of 200. If you instead want to judge SNR in parts of the scene that are, say, four stops darker, you need to measure with photosites receiving four stops less light. This is the illumination received in the mid-tones at EI four stops higher, so you can get an approximate reading of that by looking at the SNR18 for that higher EI: 3200 in this example. It is not an exact match, because although the photosites are receiving the right amount of light for this test, that "EI=3200 midtone signal" is then being amplified by a greater amount than the "EI=200 -4 from midtone signal", affecting the noise introduced after amplification—which in modern sensors is mostly from the ADC process itself.
The 'SNR at 18%' values of the Nikon D850, at base ISO, are a significant 6.5dB higher than the Olympus E-M1 MkII.
I am not sure what that comparison has to do with either the subjects of this thread (comparing "highlight based exposure strategies) or of my post (how to assess photographically relevant DR and SNR levels) — but it is about what is expected if the noise levels are dominated by photon shot noise and the sensor's well capacity is about the same per unit area, which would predict 6dB ("one stop") for a quadrupling of sensor area. Half of that 0.5db (1/12 stop) gap goes away once one corrects for the fact that EI indices used in DXO's comparison differ between the cameras and the EI for each camera differs from its ISO SSat speed ("base ISO speed") by a different amount. So then the measurements are within about 1/24 stop of a simplistic prediction based on photon shot noise and sensor area alone.
P. S. I just noticed that DP already said the same thing far more succinctly: "... photon shot noise ... die size ..."
-
The resolving power of your lenses is fixed by their optics - compared to lower res sensors, the D850 sensor will provide a better chance of exposing the limits of those optics and the flaws in focus technique.
kirk
I think it would be clearer and more truthful to say that the resolving power of a lens is limited by its optics. Do you see the difference?
The resolution of a photograph, using the best techniques such as a tripod, or sufficiently fast shutter speed when hand-held, and accurate focusing of course, is always a combination of sensor resolution and lens resolution.
If you increase the resolution of either one, you increase the final resolution in the photographic image. I tested this for myself years ago when Canon introduced the 15mp 50D, which has 50% more pixels than its previous model, the 10mp 40D.
Using the same Canon prime lens with both cameras, I took many shots of a test chart, at various apertures, using a tripod and manual focus. Whilst I can't remember the precise details without digging into my old records, I recall in general terms that the resolution differences in the centre of the image were approximately equal to, and sometimes greater than, a change in aperture by one F/stop.
In other words, the 50D at F8 was at least as sharp as the 40D at the lens' sharpest aperture of F/5.6, and the 50D at F11 would be at least as sharp as the 40D at F8, and so on.
Now, it's also true that certain weaknesses in a lens will become more apparent through comparison with the higher resolution in the center. For example, if a lens has rather poor resolution at the edges and corners, a higher-pixel-count sensor will not improve that resolution at the edges as much as it will improve the resolution in the center, so the edges only appear worse in comparison with the significantly better resolution at the center.
However, if resolution at the edges is an important consideration in a particular shot, the higher-resolving sensor allows you to stop down to improve edge performance, whilst still maintaining the same, or even slightly worse, center resolution that the lower-resolving sensor provides at the sharper aperture.
If you don't believe me, check out the lens tests at DXOMark. There are numerous examples of the same lens being tested on different camera bodies. The body with the higher pixel count always results in the same lens having a higher score. Here's one example in the attached image, comparing the performance of the Nikkor 14-24mm on the Nikon D700 and D810. The jump in performance is huge, but so is the jump in pixel-count, of the D810. ;)
-
You have completely overlooked the EI shift part of my comparison.
No I haven't. Did you not read my previous comments about Canon's DR being equal to Nikon's DR at higher ISOs? Nikon's DR advantage tends to be greatest at base ISO and 2 or 3 of stops higher, say to ISO 800. Beyond ISO 800 DR tends to be very similar. This is something useful to know.
The point I am making is that one cannot get any reliable information on the quality of the shadows by looking at the graph of 'SNR at 18%' values. Period.
To demonstrate this fact, I've attached an image comparing the SNR and DR values of the FF Canon 5D and the 'cropped format' Nikon 7200. Amazingly, SNR is about the same for both cameras, across the entire ISO range, despite the difference in format size.
However, it would be a huge mistake for anyone to deduce from the SNR graph that the noise in the shadows, from both cameras, is approximately the same. The difference in DR is so huge, it would have to slap you on the face. ;)
Perhaps we should start a new thread, maybe in the Coffee Corner, titled, "The Denial of the results of scientific camera tests."
I find it amusing that so many posters, especially at Dpreview, have claimed the DXO tests are fraudulent, or inaccurate, or a misrepresentation of the situation. But such posters never show their own valid comparisons to debunk the DXO results. The best they can do is sometimes show out-of-camera jpeg images, which only demonstrate that Camera A has better 'in-camera' jpeg processing than Camera B.
-
SNR @ 18% is where shot noise already dominates other noises sources (those affecting deep shadows), so there sensors of similar generation & technology differs just by a die size... FF shall be 2 stops better than m43 being twice bigger (when equalized to the same mp and comparing more or less similar tech generations - which D850 and E-M1mkII are, both being Sony Semi tech)
being 4 times bigger of course, was typing too fast ...
-
6dB ("one stop") for a quadrupling of sensor area.
except for this matter a "stop" here will be 3dB = you get twice better SNR (6dB) by getting 2 stops more exposure = either by 4 times bigger sensor area (let us not get into vignetting, angle of light to sensels away from the center, etc) or 4 times longer exposure time or 2 stops wider aperture or combination of ...
3dB = 20log(1/sqrt(2)) = 1 stop more exposure (light collected by bigger die size,longer exposure time, wider aperture or combo)
-
The point I am making is that one cannot get any reliable information on the quality of the shadows by looking at the graph of 'SNR at 18%' values. Period.
true
-
The point I am making is that one cannot get any reliable information on the quality of the shadows by looking at the graph of 'SNR at 18%' values.
I partially agree, but let me put it this way: both the "SNR 18%" and the "DR graphs give limited partial information about usable DR, in the sense of what subject brightness range can be handled with adequately high SNR in the darkest relevant parts of the image. Some of the limitations are:
- SNR 18% assesses only shot noise and "upstream" noise introduced by the camera before the variable amplification for ISO speed adjustment is introduced, so n particular omits quantization noise from the ADC process.
- DXO's DR overestimates by going down to SNR=1 instead of stopping at a more appropriate level. The number of stops to subtract to get a "photographically useful DR" are not known to be the same for different cameras (or even different ISO setting on the same cameras) so it is not jus a matter of "a higher DXO DR score means more useful DR by roughly the same increment".
Aside: the DXO graphs suggest that
(a) the D850 has an unusually low true SSat (aka "base ISO speed": 44, vs 70 for the Sony A7R III and 87 for the Olympus OM-D E-M1 Mk II), and
(b) this is due to lower QE rather than deeper wells.
This may be a design decision weighed more to color accuracy (with narrower pass bands for the CFA) than low light performance, compared to other cameras.
To see the comparison go to DXO comparison (https://www.dxomark.com/Cameras/Compare/Side-by-side/Sony-A7R-III-versus-Nikon-D850-versus-Olympus-OM-D-E-M1-Mark-II___1187_1177_1136) and select Measurements -> SNR 18% -> Screen, for a per pixel comparison. The surprising thing is that the D850 and EM1MkII have the same SNR curve, indicating equal photo-electron counts at equal ISO speed, despite the larger D850 photosites.
Or else something is weird with the way that DXO measures for those SNR graphs.
-
except for this matter a "stop" here will be 3dB = you get twice better SNR (6dB) by getting 2 stops more exposure = either by 4 times bigger sensor area (let us not get into vignetting, angle of light to sensels away from the center, etc) or 4 times longer exposure time or 2 stops wider aperture or combination of ...
3dB = 20log(1/sqrt(2)) = 1 stop more exposure (light collected by bigger die size,longer exposure time, wider aperture or combo)
I was using "one stop" loosely, to mean a doubling, so I think we are saying same thing: "one stop more light per photosite" meaning a doubling of light received by the photosite (whether by doubling photosite area or doubling exposure duration or increasing aperture by one stop) will increase SNR by sqrt(2), which is 3dB in the scale used by DXO, so it takes two stops more exposure to double the SNR.
-
Not to change the subject :) but I have read the earlier parts of this thread with new understanding and went about trying some experiments to where I cross the overexposure line with my D850. At ISO 64 with a test image image including a bright sky, I was able to overexpose by two stops and still had some headroom to fully recover highlights in Lightroom. This seems to be corroborated in RawDigger (Fig 1), where I see for the NEF version 27.9% avg G, 5.9% B, and 0% R O/E; while the Max stat values for G/B channels are both around 15.8K, R channel only comes to 11.2K. I realize this could vary per image. Upon converting to DNG during import (zeroed), the RawDigger histograms look exactly the same to me (Fig 2), the O/E stats are a little different only in the G channel average (27.4%). OK, I am not too concerned at this point and my results seem to be close to those discussed earlier by others.
However, when I compare the NEF and DNG histograms in FastViewer, I get slightly different results for each (Fig3&4, respectively).
My questions: Practically, this maybe makes no difference, or does it? I have read that Adobe makes some hidden compensation adjustments, but again the histograms are from zeroed files and shouldn't the two programs from the same vendor yield the same result? I didn't go above +2EV, but perhaps there is even a bit more overhead room? Also, if on average, I am shooting 2 stops underexposed, that means I am throwing away 75% of my data, if I understand the discussion!!!
Andrew, went through your histogram video and 40 minutes seemed like 10!!! Lot's of great info on the Libraw site (https://www.fastrawviewer.com/Libraw-products). Thanks for the discussion folks!
-
shouldn't the two programs from the same vendor yield the same result?
FRV does rough calculations by design - RawDigger does precise ... when in doubt believe RawDigger
PS: plus FRV has an option "Apply Adobe hidden exposure correction" - check if it is set or not
-
The surprising thing is that the D850 and EM1MkII have the same SNR curve, indicating equal photo-electron counts at equal ISO speed, despite the larger D850 photosites.
Yes. That is surprising and is also good news. If Olympus were to get a full frame sensor manufactured, using the same quality and size of pixels as the E-M1 Mk II, it would be approximately an 80mp sensor (depending on aspect ratio), with better 'SNR at 18%' than the D850. That's technological progress.
Unfortunately, the DR of the smaller E-M1 pixels is significantly worse, as shown in attached image of the DXO graph.
An example of the practical implications is, if the D850 image were cropped to 20mp with a 4:3 aspect ratio, skin tones or similar levels would we just as good as a full frame shot with the E-M1 MkII at its base ISO, but the DR of the cropped D850 shot would be about one stop better, resulting in better shadow detail and lower noise in the shadows. Agreed?
-
SNR @ 18% is where shot noise already dominates other noises sources (those affecting deep shadows), so there sensors of similar generation & technology differs just by a die size... FF shall be 2 stops better than m43 being twice bigger (when equalized to the same mp and comparing more or less similar tech generations - which D850 and E-M1mkII are, both being Sony Semi tech)
Wrong. For the same exposure, the FF sensor collects 4 times as much light as the M4/3 sensor, but the SNR advantage (for equal technologies) is NOT 2 stops but just 1 stop, and the advantage FF vs APS is not 1+1/3 stops but just 2/3 stops. I.e. the improvement of collecting 4 times as much light over four times the surface (1 stop), is lower than the improvement of collecting 4 times as much light over the same surface with +2EV extra exposure (2 stops).
This is about noise statistics, just compare the Canon 7D II vs Canon 5DS (I take this example since it's the only one in the market where 5DS pixels are exactly the same technology and size as 7D II pixels):
SNR18%: 4dB gap (2/3 stops):
(http://www.guillermoluijk.com/misc/5ds_vs_7d2_db.png)
DR 2/3 stops gap:
(http://www.guillermoluijk.com/misc/5ds_vs_7d2_ev.png)
Since most M4/3 and APS cameras usually have denser sensors than FF cameras, i.e. smaller photocells than in the previous 5DS vs 7D II comparison, those gaps can be slightly increased for technological and QE reasons, but the mathematical starting point of reference is: 1 stop noise advantage FF vs M4/3 (not 2 stops), and 2/3 stops noise advantage FF vs APS (not 1+1/3 stops).
To understand the reasons for this just look at how 4 pixels bin into one: noise is added in quadrature, so SNR is doubled (1 stop of advantage), not multiplied by 4 (which would if there were 2 stops of noise improvement):
http://theory.uchicago.edu/~ejm/pix/20d/tests/noise/noise-p3.html#pixelsize
That is why sensor technology is far more important than sensor size for noise performance. The following infography shows the crop on a A7R III sensor that would provide the same DR as the entire sensor of the most popular FF cameras. Look at the tiny sensor (0,9Mpx) with A7R III technology that would provide the same DR as the whole Canon 6D II sensor:
(http://www.guillermoluijk.com/misc/sensorsizesvsa7riii.png)
Thanks to this statistical fact smartphone cameras are achieving 'unexpected' great performance with minuscule sensors.
Regards
-
Wrong. For the same exposure, the FF sensor collects 4 times as much light as the M4/3 sensor, but the SNR advantage (for equal technologies) is NOT 2 stops but just 1 stop
you missed my next post in that thread with the math... let me help you with the quote from above
you get twice better SNR (6dB) by getting 2 stops more exposure = either by 4 times bigger sensor area (let us not get into vignetting, angle of light to sensels away from the center, etc) or 4 times longer exposure time or 2 stops wider aperture or combination of ...
at no point in time above I claimed that 4 times bigger/longer/wider means 4 times more SNR-wise...
-
Yes. That is surprising and is also good news. If Olympus were to get a full frame sensor manufactured, using the same quality and size of pixels as the E-M1 Mk II, it would be approximately an 80mp sensor (depending on aspect ratio), with better 'SNR at 18%' than the D850. That's technological progress.
It might also be good news now for the D850's target users: trading away a bit of a modern sensor's excellent low light performance for improved color accuracy could make many "landscape/enough light" users very happy.
Unfortunately, the DR of the smaller E-M1 pixels is significantly worse, as shown in attached image of the DXO graph. ... the DR of the cropped D850 shot would be about one stop better
Of course! Have we not all agreed long ago that when a sensor four times larger (or a crop to equal pixel count with photosites twice as large) can be given full exposure (not constrained to substantial "photosite underexposure" by the need for enough DOF and high enough shutter speed limits), it will give better image quality? Larger formats are certainly some times for some uses worth their greater cost and bulk! (True even when comparing 36x24mm to 44x33mm to 54x40mm.)
But Ray, you should know by now that you will not impress me with DXO's "total engineering dynamic range" graphs, with an unknown number of a camera's lowest stops having uselessly low SNR, and that number of "junk stops" of DR unlikely to be the same for all cameras. Are there any good up-to-date photographic dynamic range graphs for recent cameras, from Bill Claff or Jim Kasson or other sources?
-
But Ray, you should know by now that you will not impress me with DXO's "total engineering dynamic range" graphs, with an unknown number of a camera's lowest stops having uselessly low SNR, and that number of "junk stops" of DR unlikely to be the same for all cameras. Are there any good up-to-date photographic dynamic range graphs for recent cameras, from Bill Claff or Jim Kasson or other sources?
If you don't want to use 0 dB as the floor for DR calculations as per DXO, you can use their full SNR plots to calculate the DR for other noise floors as discussed here (http://forum.luminous-landscape.com/index.php?topic=42158.0).
Regards,
Bill
-
If you don't want to use 0 dB as the floor for DR calculations as per DXO, you can use their full SNR plots to calculate the DR for other noise floors as discussed here (http://forum.luminous-landscape.com/index.php?topic=42158.0).
Regards,
Bill
Thanks for that reminder, Bill. This is all 'deja vue' for me. I've been a bit puzzled as to why I couldn't find the full SNR plots for the cameras I've been comparing. I wondered if DXOMark had discontinued the provision of such graphs.
However, after a bit of inquiry on the internet, I was reminded that the full SNR plots are not available from the 'camera comparison' page. One has to do a search for the specific camera measurements, where an additional heading of 'Full SNR' is included.
BJL should be ecstatic to discover this. ;D
-
If you don't want to use 0 dB as the floor for DR calculations as per DXO, you can use their full SNR plots to calculate the DR for other noise floors as discussed here (http://forum.luminous-landscape.com/index.php?topic=42158.0).
Thanks Bill! This does look like the core information, as used by Bill Claff and others to estimate "photographically useful dynamic range". (Once you mentioned it, I had guessed that it would show only in the single camera measurements, not the camera comparisons [too many curves!], but then discovered that it is also only available in the "desktop" version of the site, not on phones.)
On one hand, at what I find interesting SNR levels, from 12dB [SNR of 8:1] up, the results are close to what one would expect from shot noise alone: about one more stop for the D850 than for the EM1MkII at their lowest ISO settings (ones with minimum gain, so sending full wells to about maximum raw level rather than clipping.) It does seem that at "equal total exposure" [see note], so with suitably higher EI with the larger sensor, the advantage goes slightly the other way—but care is required due to the different pixel counts and the different gain levels used at the same ISO setting.
My raw observations, not yet carefully analyzed, are that
1) At 12dB [SNR 4:1] and rising, the EM1MkII is increasingly "better" (curves to the left) at equal ISO settings.
2) At 24B [SNR 16:1] the curves are offset by about one stop on the ISO speed settings scale: curves at a given ISO speed setting on the D850 roughly match those for the EM1MkII at twice the ISO speed setting.
3) On the other hand, at 6dB [SNR 2:1] and below, the D850 is about one stop better, and a bit more than one stop way down at 0dB [SNR 1:1].
Note: "Equal total exposure" meaning equal amount of light delivered to the sensor, as when both DOF and exposure time are equal.
-
That would not work. Incident light metering ignores highlights.
Best regards
Erik
Use an incident light meter?
Kent in SD
-
Thanks Bill! This does look like the core information, as used by Bill Claff and others to estimate "photographically useful dynamic range". (Once you mentioned it, I had guessed that it would show only in the single camera measurements, not the camera comparisons [too many curves!], but then discovered that it is also only available in the "desktop" version of the site, not on phones.)
On one hand, at what I find interesting SNR levels, from 12dB [SNR of 8:1] up, the results are close to what one would expect from shot noise alone: about one more stop for the D850 than for the EM1MkII at their lowest ISO settings (ones with minimum gain, so sending full wells to about maximum raw level rather than clipping.) It does seem that at "equal total exposure" [see note], so with suitably higher EI with the larger sensor, the advantage goes slightly the other way—but care is required due to the different pixel counts and the different gain levels used at the same ISO setting.
My raw observations, not yet carefully analyzed, are that
1) At 12dB [SNR 4:1] and rising, the EM1MkII is increasingly "better" (curves to the left) at equal ISO settings.
2) At 24B [SNR 16:1] the curves are offset by about one stop on the ISO speed settings scale: curves at a given ISO speed setting on the D850 roughly match those for the EM1MkII at twice the ISO speed setting.
3) On the other hand, at 6dB [SNR 2:1] and below, the D850 is about one stop better, and a bit more than one stop way down at 0dB [SNR 1:1].
Note: "Equal total exposure" meaning equal amount of light delivered to the sensor, as when both DOF and exposure time are equal.
In case anyone is confused, I'd like to emphasise the point that these 'Full SNR' graphs relate to the pixel, not the full frame of the sensor. It might be more revealing if DXO were to provide the Full SNR plots for the entire sensor, normalized to the usual 8mp..
Another point that should be emphasized, which Emil Martinec also made, is that Photographic DR is a very subjective definition that can vary wildly from individual to individual, and also be influenced by the display size of the image, whether print or screen.
The engineering definition that DXO uses, is at least objective and consistent. If Camera A has 14 EV of DR, and Camera B has 12 EV of DR, then it's reasonable to presume that both cameras will produce unacceptable shadows at their DR limits.
However, it's also reasonable to presume that the shadows in a shot from Camera A, at 12 EV of DR, will be noticeably better than the shadows in the same shot from Camera B at it's engineering limit of 12EV.
The attached image of graphs comparing the Full SNR of the D850 with the E-M1 MkII, the D850 at the top, shows that the SNR of the D850 pixel is generally better. However, it's not clear where the SNR 18% level lies on the graph.
Interestingly, at maximum exposure, ie. full well capacity, the D850 pixel has about 5dB better SNR. At 1% on the Logarithmic scale, which represents reasonably deep shadows, it has about 4dB better SNR. This just goes to show how limited the SNR 18% plot is.
When one extrapolates all the values on the Full SNR graphs to full frame values, with approximately 4x the number of photons reaching the D850 sensor, the noise advantages of the D850 are apparent across the entire range, from the deepest shadows to the brightest highlights.
-
Another point that should be emphasized, which Emil Martinec also made, is that Photographic DR is a very subjective definition that can vary wildly from individual to individual, and also be influenced by the display size of the image, whether print or screen.
You make that sound like a bad thing! But apart from the gratuitous pejorative "wildly", what you and Emil are saying is that Photographic dynamic range is a measure adapted to be relevant to the actual needs of a particular photographer and particular intended uses. To me, that is far more useful than a consistent but in itself irrelevant measure based on a SNR lower than could be of any direct practical use for assessing image quality. That is the usual sorry end when one attempts to reduce a complex reality to a single number. (For worse examples, see DXOMark's single number scores for "portrait", "landscape", "sports", and even overall scores for "sensor" and "lens".) That is why more useful assessments of a tool as complicated as a camera need a healthy range of measurements from which practical conclusions can be derived according to actual practical needs — like those complete SNR curves. It is like understanding and using MTF curves vs a single "bragging rights" number for "resolution" or "sharpness".
If you crave a number to use in comparisons that is consistent and clearly defined, one can choose a particular standardized PDR, as Bill Claff does. I would go for something like a low end where pixel level SNR = 5:1. Using SNR = 1:1 is equal arbitrary relevant to photographic needs.
-
Just in time:
https://www.fastrawviewer.com/blog/determining-practical-dynamic-range
Kirk
-
Just in time:
https://www.fastrawviewer.com/blog/determining-practical-dynamic-range
Kirk
Quite a lot to think about there. Maybe — in the spirit of Bill Claff's choice of standardized print size and viewing distance and Ray's desire for something similar in the form of DR after downsampling to 8MP — there is an informative curve of "usable DR vs resolution", or "resolution at various levels of scene brightness". At one extreme, even an image file with SNR<=1 per pixel might be salvageable by sufficient downsampling. In fact, that is sort of what we do with traditional black and white film, where the raw signal is all 0's and 1's, because each silver halide grain is either exposed or not, but there are many billions of "chemical pixels".
-
Quite a lot to think about there. Maybe — in the spirit of Bill Claff's choice of standardized print size and viewing distance and Ray's desire for something similar in the form of DR after downsampling to 8MP — there is an informative curve of "usable DR vs resolution", or "resolution at various levels of scene brightness". At one extreme, even an image file with SNR<=1 per pixel might be salvageable by sufficient downsampling. In fact, that is sort of what we do with traditional black and white film, where the raw signal is all 0's and 1's, because each silver halide grain is either exposed or not, but there are many billions of "chemical pixels".
Photographic Dynamic Range is what you decide through your own tests in relation to particular circumstances which can vary. The print size will also have some bearing on the amount of shadow noise and detail which is acceptable. Noise can be significantly reduced if one is prepared to sacrifice detail, which is less of a concern on the smaller print.
When I bought the Nikon D7000 to replace my Canon 50D, the main attraction was its high DR, almost 2.5EV better than the 50D at base ISO. Shortly after taking delivery of the D7000, I made a thorough comparison of the DR capability of both cameras by taking a serious of shots of the same target in the same lighting conditions, starting with an ETTR shot with both cameras, then reducing exposure successively, one stop at a time, with each camera until I reached 13 &2/3rds stops of underexposure, the DR limit of the D7000.
At the different DR limits of both cameras, the images were similarly degraded, but some detail was still discernible. At 13 stops underexposure on the 50D, no detail at all was discernible. It was all pure noise.
However, at 11 stops underexposure, the D7000 image was acceptable, but not the 50D image because it was at its DR limit. At 9 stops underexposure, the 50D was acceptable for shadows, but the D7000 was significantly better and more detailed.
In other words, the PDR of the D7000 was about 11.5 EV and the PDR of the 50D about 9 EV, which corresponds with the DXO engineering DR differences of about 2.5 EV. At 8 stops underexposure, the D7000 had an insignificant edge, and at 7 stops underexposure there was no discernible difference.
The issues as I see them, are:
(1) How to determine the PDR of your camera? Do you just accept someone elses view, or do you take the trouble to do your own testing?
(2) How do you determine the DR (or SBR) of the scene you are about to shoot?
I've already described the first process, so what about the second process? If you have plenty of time to organize everything in a studio, then no problem. However, if you are taking numerous shots of various scenes as you travel, there needs to be a quick and reliable process.
The quickest process is to simply automatically bracket all exposure, which is what I used to do when using the older Canon DSLRs. However, bracketing most shots fills up you memory card faster and involves a lot more processing time for those shooting in RAW mode.
With my Nikon cameras, it has become apparent that by far the majority of my shots do not require bracketing if I take the trouble to find the optimum ETTR exposure.
If I do get the impression that a particular scene might exceed the PDR capabilities of my camera, then it takes me a little more time than it takes to get a ETTR shot, because a mathematical calculation is required.
Using a single focusing square, and camera in manual exposure mode, I simply move the focusing square to the brightest part of the image and adjust exposure with my thumb on a wheel untill the meter in the viewfinder is at the far right where the + sign changes into an arrow head. That is the correct exposure for an ETTR shot, which I've determined through my own tests. No calculations are required, except if the shutter speed is too slow for a hand-held shot, in which case I'll raise the ISO and readjust the exposure.
If I think the lighting in the shadows, in another part of the composition, might be inadequate for PDR quality, I'll simply move the focusing square to those shadows and reduce the shutter speed, with thumb on wheel, till the camera's meter moves to the extreme right, as though I'm taking an ETTR shot of the shadows.
I then have to do a mathematical calculation to convert the differences in shutter speeds into numbers of stops or EVs. Alternatively, as I move the exposure adjustment wheel, I could count the number of clicks involved before the exposure meter reaches the far right. If the cameras is set for 1/2 a stop exposure increments, then I could divide the number of clicks, or changes in exposure, by 2 to get the difference in EV range between the highlights and the shadows. However, the D810 allows for 1 stop changes in exposure, which would make it even easier.
If the difference between the two ETTR readings, one for the brightest highlights, and one for the deepest shadows which have meaningful detail, is greater than my calculated PDR for my camera, say 11 stops, then I need to bracket exposure.
However, even if one has gained a general PDR rating for one's camera through one's own testing, this rating can change according to circumstances. A good example would be the photographing of the amazing ruins at Angkor Wat. In a particular scene there might be an amazing and detailed engraving in the stone, in the deepest shadows. After using my method to check the lighting conditions in those shadows, one might find that the shadows are in fact within the PDR range of one's camera. On the other hand, because the content of the shadows is so interesting, one might decide to bracket exposure in order to get the maximum quality and detail in those shadows.
This is the one major reason why I prefer the higher DR of the Nikon cameras. There are far fewer occasions when these sorts of time-consuming difficulties arise. Most of the time I can simply take an ETTR shot and be confident that the shadows will be okay, at base ISO and/or a couple of ISO stops above.
-
This is the one major reason why I prefer the higher DR of the Nikon cameras. There are far fewer occasions when these sorts of time-consuming difficulties arise. Most of the time I can simply take an ETTR shot and be confident that the shadows will be okay, at base ISO and/or a couple of ISO stops above.
I agree that with the latest Nikon sensors (and many others) DR is often not a limiting factor. I also find that bracketing is seldom necessary with my D850 if one uses proper ETTR exposure at base ISO. If one has to increase the ISO, jumping directly to ISO 400 makes sense because this is where the Aptina dr-pix kicks in. See Here (http://www.photonstophotos.net/Charts/PDR_Shadow.htm#Nikon%20D850).
At ISOs over 400 one can protect the highlights by shooting dark and increasing exposure in post. To obtain a decent LCD preview, I shoot dark only by one or two stops.
Regards,
Bill
-
I agree that with the latest Nikon sensors (and many others) DR is often not a limiting factor. I also find that bracketing is seldom necessary with my D850 if one uses proper ETTR exposure at base ISO. If one has to increase the ISO, jumping directly to ISO 400 makes sense because this is where the Aptina dr-pix kicks in. See Here (http://www.photonstophotos.net/Charts/PDR_Shadow.htm#Nikon%20D850).
At ISOs over 400 one can protect the highlights by shooting dark and increasing exposure in post. To obtain a decent LCD preview, I shoot dark only by one or two stops.
Regards,
Bill
Yes. From ISO 400 onward the D850 appears to be essentially ISO-invariant, according to the DXO graph, although the accumulation of very insignificant degrees of DR loss for each stop of underexposure, instead of using the same exposure at the appropriately higher ISO, does eventually lead to a noticeable effect, according to the DXO DR graph.
For example, if one were to underexpose by 7 stops at ISO 400, instead of raising ISO to 51,200, one would lose 0.5EV of DR, which is at the threshold of significance.
Perhaps you could test this, Bill, to confirm the accuracy of DXOMark's results, and determine whether that claimed improvement of 0.5EV is noticeable. I don't have a D850 yet. My D810 is still functioning well, despite my having dropped it from a shopping trolley onto a concrete slab some time ago, which produced an open gash in the corner of the camera, which I've plugged up with glue so that dust and water will not get in.
-
I also find that bracketing is seldom necessary with my D850 if one uses proper ETTR exposure at base ISO.
And I find that ETTR exposure at base ISO is seldom necessary with proper bracketing!
More seriously, thanks Bill for the confirmation that with the latest sensor technology, settings beyond about ISO 400 and analog gain beyond that level are mostly for convenience (making in-camera reviews, live view previews and default JPEG conversions look roughly right, for example) rather than being of much importance to image quality.
-
One wonders how Ansel Adams managed to take a photo without access to a computer and this 5 pages of priceless information?
In the good old days you took a guess or measured it.
Now I put the camera on manual, point the camera at the sky or brightest part of the shot, adjust until the pointer is against the right hand stop but not blinking. Then shoot once and move on. (:-)
-
Ansel Adams's technique was the complete opposite of "taking a guess". The Zone System requires careful light meter measurements of multiple points in the scene, coupled with detailed knowledge of the response characteristics of film and developer. Modern digital sensors have greatly simplified the process of getting an acceptable exposure, and there is no longer a development process, as digital sensors have linear response curves. The creative part of image processing is now in what Adams called printing, and what we call post-processing and printing.
-
Ansel Adams's technique was the complete opposite of "taking a guess". The Zone System requires careful light meter measurements of multiple points in the scene, coupled with detailed knowledge of the response characteristics of film and developer.
Yes! And we can and should do the same with digital (raw capture).
Modern digital sensors have greatly simplified the process of getting an acceptable exposure, and there is no longer a development process, as digital sensors have linear response curves. The creative part of image processing is now in what Adams called printing, and what we call post-processing and printing.
There still is development; the raw processor. But you're absolutely correct if and when we view a raw Histogram. When doing so, no development and where we really should examine exposure san's development.
-
Ansel Adams's technique was the complete opposite of "taking a guess".
Not that I said Ansel was guessing but that depends what you mean by a guess. If he did make a guess then it was a very good guess because he was a very good photographer. I am not sure that he had an exposure meter in his plate camera or that he come up with the Zone System before the time of his earliest works. The Zone System was documented as a teaching aid for other photographers in about 1940. He did not need to bracket because he had skill and only about 12 shots available for the day anyway.
The point was that if you don't capture within acceptable limits then there is no amount of post production creativity that will fix it.
-
Minor White once described a photography outing with Ansel and Edward Weston. At Ansel's insistence, Edward was using a light meter.
Edward's technique, according to Minor, was to wave the meter around, then glare at it and mutter "It's wrong!"
Then he proceeded to set the exposure his eye told him was correct.
-
Minor White once described a photography outing with Ansel and Edward Weston. At Ansel's insistence, Edward was using a light meter.
Edward's technique, according to Minor, was to wave the meter around, then glare at it and mutter "It's wrong!"
Then he proceeded to set the exposure his eye told him was correct.
But was it?
-
But was it?
Judging from the many of his prints that I've seen, I'd say it was. It was certainly more reliable than my eye.
-
Judging from the many of his prints that I've seen, I'd say it was.
Maybe thanks to extensive "Post-processing" during development and printing?
I've manipulated images myself during my wet darkroom days, both during the film development process, by using the temperature of my fingers, and easier (because it's easier to see what's happening) by dodging and burning during the exposure of the print and again using my fingers.
Cheers,
Bart
-
Judging from the many of his prints that I've seen, I'd say it was. It was certainly more reliable than my eye.
A print informed you of the exposure of the neg? I think not!
-
I've manipulated images myself during my wet darkroom days, both during the film development process, by using the temperature of my fingers, and easier (because it's easier to see what's happening) by dodging and burning during the exposure of the print and again using my fingers.
Ah yes, the finger 'trick'. Trick because in my first trimester in photography school, everything had to be shot 4x5, B&W and all prints made had to be 'straight' with zero dogging and burning. The idea in those very old days was, we students needed to learn to properly expose and light and no print manipulation was allowed. Contact proofs on the back of the print board had to be supplied so the instructor could verify we didn't dodge and burn the 8x10 print itself. We fudged this all a tad using the wet finger trick; it did work unless we went overboard and the sharper instructor's would catch us. Then the mounted print would come back from critique with "Remake" on it. We had to start from scratch with the entire assignment.
It was a tough lesson and a tough first trimester but we learned to expose and light with little regard to darkroom manipulating.
Of course this was a good decade before anyone had any idea what Photoshop would or could be.....
-
Ah yes, the finger 'trick'. Trick because in my first trimester in photography school, everything had to be shot 4x5, B&W and all prints made had to be 'straight' with zero dogging and burning. The idea in those very old days was, we students needed to learn to properly expose and light and no print manipulation was allowed. Contact proofs on the back of the print board had to be supplied so the instructor could verify we didn't dodge and burn the 8x10 print itself. We fudged this all a tad using the wet finger trick; it did work unless we went overboard and the sharper instructor's would catch us. Then the mounted print would come back from critique with "Remake" on it. We had to start from scratch with the entire assignment.
It was a tough lesson and a tough first trimester but we learned to expose and light with little regard to darkroom manipulating.
Of course this was a good decade before anyone had any idea what Photoshop would or could be.....
I never went to a photography school but I did go through a period of time where I shot only color transparencies for the same reasons.
Exposure had to be nailed and white balance only corrected through camera filters.
-
:-X
Polite is all that is required. Rude isn't.
Dave S
:-X Well shut my pie hole!
-
Glad to see the great interest in ETTR, I encourage you all in this thread to consider the possibility of signing for this CHANGE.ORG request to camera makers to include an option of having RAW histograms and RAW highlight clipping warnings, which would help a lot in achieving easily and quickly the perfect ETTR on our cameras:
http://forum.luminous-landscape.com/index.php?topic=124450.0
Spreading the link among your photographic community would be highly appreciated for all of us who would like to have such tools.
Regards
-
Not that I said Ansel was guessing but that depends what you mean by a guess. If he did make a guess then it was a very good guess because he was a very good photographer. I am not sure that he had an exposure meter in his plate camera or that he come up with the Zone System before the time of his earliest works. The Zone System was documented as a teaching aid for other photographers in about 1940. He did not need to bracket because he had skill and only about 12 shots available for the day anyway.
The point was that if you don't capture within acceptable limits then there is no amount of post production creativity that will fix it.
He certainly made a (very) informed guess for his Moonrise, Hernandez, New Mexico shot :- see http://anseladams.com/ansel-adams-anecdotes/
Dave
-
Minor White once described a photography outing with Ansel and Edward Weston. At Ansel's insistence, Edward was using a light meter.
Edward's technique, according to Minor, was to wave the meter around, then glare at it and mutter "It's wrong!"
Then he proceeded to set the exposure his eye told him was correct.
Here's the version from Adams himself (page 30 of The Negative)
I can recall seeing Edward Weston, who was not particularly of scientific persuasion, using his meter in rather unorthodox ways. He would point it in several directions, take a reading from each, and fiddle with the dial with a thoughtful expression. "It says one-quarter second at f/32, I'll give it one second."
But Adams continues
His approach was empirical, based on long experience combined with very deep sensitivity and intuition, and his extraordinary results speak for themselves.
-
Thanks for the clarification, BJL. I should have looked in my copy of The Negative.
I heard Minor's version in one of his workshops.
I agree with Adams' assessment of Edward's approach.
-Eric
-
Ansel Adams's technique was the complete opposite of "taking a guess". The Zone System requires careful light meter measurements of multiple points in the scene, coupled with detailed knowledge of the response characteristics of film and developer.
Is it fair to say that the Zone System goal is to choose a placement of one level so that all other relevant levels of the scene fall as much as possible in the straight line section, rather than any important highlight or shadow details being too far onto the shoulder or toe? What I do not know of any film photographer doing is pushing the levels as far up the straight line section as possible, and then having to bring levels back down in development or printing. It was more "expose to the middle", with a very sophisticated interpretation of "middle".
There was of course a version of aiming for maximum exposure: using the slowest film that can get the job done.
-
Ansel Adams's technique was the complete opposite of "taking a guess". The Zone System requires careful light meter measurements of multiple points in the scene, coupled with detailed knowledge of the response characteristics of film and developer. Modern digital sensors have greatly simplified the process of getting an acceptable exposure, and there is no longer a development process, as digital sensors have linear response curves. The creative part of image processing is now in what Adams called printing, and what we call post-processing and printing.
It is interesting that Ansel Adams and the Zone System was brought up in a discussion of ETTR. When dealing with what he called a short scale subject (DR of the scene less than the DR of the capture medium) Ansel would let the high values fall on Zone VI rather than Zone VII so that the resulting negative would have less grain and "wasted density" while still capturing the full scale of the scene (see pp 68 and 69 of The Negative). Photon counting statistics apply to film as well as digital, but with the films of his time, I think that film grain would swamp photon noise. An analysis of photon noise as applied to modern film would be interesting, but of limited practical value because of who uses film these days.
Regards,
Bill
-
with zero dogging and burning.
Excellent! A Freudian slip, perhaps?