This topic is in the 2013 archive. You can read it and like its posts here. It is read only.
Archive 2013ettr - an urban legend?
p.2 #1ettr - an urban legend?
I would think that the test would be done in manual mode only.
Use the meter to set "normal exposure" and then bracket around that point to set your other images.
If any automation is being used I would be skeptical of the results.
Use the meter to set "normal exposure" and then bracket around that point to set your other images.
If any automation is being used I would be skeptical of the results.
Nov 05, 2013 at 10:32 PM
p.2 #2ettr - an urban legend?
RustyBug wrote:
I"m having a hard time believing that an image that was actually 5 stops underexposed would be lifted in post by 5 stops without incurring any noticeable noise or other effects (but I can be made a believer). I gotta wonder if the metering didn't already equalize the EC/ISO via shutter and thus the images were taken at the same exposure, but at different gain rather than truly 2 stops overexposed, 5 stops underexposed.
i assume he did do the test in manual mode, but it doesn't matter i've done similar things with manual or EC on a NEX, the result is the same. clearly you need to spend some time with modern sony sensors. that's what it looks like when you bring up the exposure 5 stops in post.
you actually can see a little bit of noise in the shadow crops at 1600 and 3200 equiv.
I"m having a hard time believing that an image that was actually 5 stops underexposed would be lifted in post by 5 stops without incurring any noticeable noise or other effects (but I can be made a believer). I gotta wonder if the metering didn't already equalize the EC/ISO via shutter and thus the images were taken at the same exposure, but at different gain rather than truly 2 stops overexposed, 5 stops underexposed.
i assume he did do the test in manual mode, but it doesn't matter i've done similar things with manual or EC on a NEX, the result is the same. clearly you need to spend some time with modern sony sensors. that's what it looks like when you bring up the exposure 5 stops in post.
you actually can see a little bit of noise in the shadow crops at 1600 and 3200 equiv.
Nov 05, 2013 at 10:52 PM
p.2 #4ettr - an urban legend?
Okay ... (shaking head) ... if there really is that much latitude, I suppose a light meter is destined to go the way of the dinosaur. I can probably wear a blindfold and guess at the exposure to get within 7 stops. 
Which then raises the question at why so many passionately obsess over 1/3 - 2/3 stop faster when it comes to glass i.e. 50/1.4 vs. 50/1.8 (no need to mention dof variance) ?
Edited on Nov 05, 2013 at 11:15 PM · View previous versions
Which then raises the question at why so many passionately obsess over 1/3 - 2/3 stop faster when it comes to glass i.e. 50/1.4 vs. 50/1.8 (no need to mention dof variance) ?
Edited on Nov 05, 2013 at 11:15 PM · View previous versions
Nov 05, 2013 at 11:08 PM
p.2 #5ettr - an urban legend?
RustyBug wrote:
Okay ... (shaking head) ... if there really is that much latitude, I suppose a light meter is destined to go the way of the dinosaur. I can probably wear a blindfold and guess at the exposure to get within 7 stops.
i can't generally recover more than 1.35 stops of highlights so you still have to watch those, which makes ettr even more silly. with the latest sensors i generally expose make sure i preserve the brightest highlight and then bring everything else up in post. with the shadows i really don't start to see any real effects until 3 stops and it takes a fair bit more to destroy an image. it's amazing how sensors have improved in 5 years.
Okay ... (shaking head) ... if there really is that much latitude, I suppose a light meter is destined to go the way of the dinosaur. I can probably wear a blindfold and guess at the exposure to get within 7 stops.
i can't generally recover more than 1.35 stops of highlights so you still have to watch those, which makes ettr even more silly. with the latest sensors i generally expose make sure i preserve the brightest highlight and then bring everything else up in post. with the shadows i really don't start to see any real effects until 3 stops and it takes a fair bit more to destroy an image. it's amazing how sensors have improved in 5 years.
Nov 05, 2013 at 11:15 PM
p.2 #6ettr - an urban legend?
RustyBug wrote:
Which then raises the question at why so many passionately obsess over 1/3 - 2/3 stop faster when it comes to glass i.e. 50/1.4 vs. 50/1.8 (no need to mention dof variance) ?
this is at iso 100. you can't really push shadows at iso 6400.
Which then raises the question at why so many passionately obsess over 1/3 - 2/3 stop faster when it comes to glass i.e. 50/1.4 vs. 50/1.8 (no need to mention dof variance) ?
this is at iso 100. you can't really push shadows at iso 6400.
Nov 05, 2013 at 11:16 PM
p.2 #7ettr - an urban legend?
Would you say that the sensors have improved that much, or the algorithms have?
sebboh wrote:
this is at iso 100. you can't really push shadows at iso 6400.
So, does that bring ETTR back into the picture strategically for high(er) iso shooting?
sebboh wrote:
this is at iso 100. you can't really push shadows at iso 6400.
So, does that bring ETTR back into the picture strategically for high(er) iso shooting?
Nov 05, 2013 at 11:19 PM
p.2 #8ettr - an urban legend?
RustyBug wrote:
Would you say that the sensors have improved that much, or the algorithms have?
if it was just algorithms we'd be able to recover do something with our old raws to make them almost as good. i really doubt that everybody else has suddenly managed to out program canon too. when you compare raws from pentax, sony, and nikon that all use the same sensor they all behave pretty much the same despite all the differences in circuitry and processing algorithms.
Would you say that the sensors have improved that much, or the algorithms have?
if it was just algorithms we'd be able to recover do something with our old raws to make them almost as good. i really doubt that everybody else has suddenly managed to out program canon too. when you compare raws from pentax, sony, and nikon that all use the same sensor they all behave pretty much the same despite all the differences in circuitry and processing algorithms.
Nov 05, 2013 at 11:28 PM
p.2 #9ettr - an urban legend?
RustyBug wrote:
So, does that bring ETTR back into the picture strategically for high(er) iso shooting?
if you need to shoot at iso 6400 you can't really lower the shutter speed by 2 stops for ettr. also, i think analog gain to boost iso on most of these cameras stops at iso 1600 or 3200 so raising iso to overexpose isn't any different from just boosting exposure in post.
So, does that bring ETTR back into the picture strategically for high(er) iso shooting?
if you need to shoot at iso 6400 you can't really lower the shutter speed by 2 stops for ettr. also, i think analog gain to boost iso on most of these cameras stops at iso 1600 or 3200 so raising iso to overexpose isn't any different from just boosting exposure in post.
Nov 05, 2013 at 11:32 PM
p.2 #10ettr - an urban legend?
Just did a quick shoot to show the ETTR technique (left is 2EV over-exposed over the "right" exposure then -2EV in post) with my non-SLT A99. Please excuse the indoor white balance.
Obviously, there was no changing light. I included the histograms to show how close they are to each other. Noise is quite apparent with the "right" exposure.
So in the end, it's just a technique. There is situation you might want to use ETTR. Most of the time, I over-expose by 0.7EV-1EV. And this is great for a one-shot HDR type of deal (maximize the DR of the sensor).
 |
||
| http://lh5.googleusercontent.com/-PW-bOB1udfg/UnnTlrCmm9I/AAAAAAAAKr4/ajtdH7UcsCM/w1363-h843-no/Compare.JPG | ||
|
Obviously, there was no changing light. I included the histograms to show how close they are to each other. Noise is quite apparent with the "right" exposure.
So in the end, it's just a technique. There is situation you might want to use ETTR. Most of the time, I over-expose by 0.7EV-1EV. And this is great for a one-shot HDR type of deal (maximize the DR of the sensor).
Nov 05, 2013 at 11:32 PM
p.2 #11ettr - an urban legend?
okay, my explanation seems to be crap.
All images were taken at Iso 100
I varied the shutterspeed from 1 second for the Iso 25 shot to 1/125 for the Iso 3200 shot.
Then I compensated for the over/under-exposure in LR, to equalize the brightness
so the "Iso 3200" shot was pushed 5 stops.
I used M mode
Edited on Nov 06, 2013 at 03:41 AM · View previous versions
Edited on Nov 06, 2013 at 03:41 AM · View previous versions
Nov 06, 2013 at 12:28 AM
p.2 #12ettr - an urban legend?
it seems i did a similar test earlier this week in an attempt to determine whether multiple exposures is worthwhile for high dynamic range scenes. shooting directly into the setting sun (not so great for lens contrast ). unfortunately i chose rather large steps on the overexposure side and i focused close to get as much detail out of near subjects, so everything at a distance is a little blurry. still informative i hope.
here's the scene with just adobe defaults applied:
http://farm6.staticflickr.com/5489/10703552684_99f78c8452_o.jpg
all shots were taken at actual iso 100 and f/11. only shutter speed varied between shots, but i did it in A mode by twiddling the EC dial on an RX1.
here's the summary crops:
http://farm4.staticflickr.com/3828/10703757996_6863ab8f70_h.jpg
and a link to where you can see it at 100%.
). unfortunately i chose rather large steps on the overexposure side and i focused close to get as much detail out of near subjects, so everything at a distance is a little blurry. still informative i hope. here's the scene with just adobe defaults applied:
http://farm6.staticflickr.com/5489/10703552684_99f78c8452_o.jpg
all shots were taken at actual iso 100 and f/11. only shutter speed varied between shots, but i did it in A mode by twiddling the EC dial on an RX1.
here's the summary crops:
http://farm4.staticflickr.com/3828/10703757996_6863ab8f70_h.jpg
and a link to where you can see it at 100%.
Nov 06, 2013 at 01:39 AM
p.2 #13ettr - an urban legend?
Shadow details on 5d2 images are crap at the best of times. It's difficult to get much from the shadows if they are even slightly underexposed so if you don't ettr, as much as practical, you only make your life a misery. I'm happy to clip specular highlights, which after all should be pure white, so this is pretty much how I expose in general. I use C1, if it matters.
Nov 06, 2013 at 01:49 AM
p.2 #14ettr - an urban legend?
Since I bought my NEX-7 I use ETTR less than before.
ETTR doesn't have to be an over exposure, it can be an under exposure if it's a high key situation.
FWIW, now days, HDR is far more useful than ETTR is for preserving detail and keeping noise as low as possible.
I think the 5DII was one of the worst high end cameras for shadow noise/banding since it could show up in properly exposed shots in some situations.
ETTR doesn't have to be an over exposure, it can be an under exposure if it's a high key situation.
FWIW, now days, HDR is far more useful than ETTR is for preserving detail and keeping noise as low as possible.
I think the 5DII was one of the worst high end cameras for shadow noise/banding since it could show up in properly exposed shots in some situations.
Nov 06, 2013 at 02:29 AM
p.2 #15ettr - an urban legend?
Phillip Reeve wrote:
All images were taken at Iso 100
I varied the shutterspeed from 1 second for the Iso 25 shot to 1/125 for the Iso 3200 shot.
Then I compensated for the over/under-exposure in LR, to equalize the brightness
so the "Iso 3200" shot was pushed 5 stops.
I used M mode
Thanks Phillip ... that's exactly the BIG CRAYON that I needed. Sorry for the confusion that I wasn't as adept to follow earlier as the others did.
Thanks to Phillip (et al). I'm still "shaking head" a bit, but now in wonderment @ why moreover than disbelief. Now it would be really interesting to see the same test Sony vs. Canon (link to pre-existing examples would be fine).
Judging from your tests (both Phillip & sebboh) ... which of the two exposure strategies would you suggest using as best for highlights, best for shadows?
I hope you don't mind (just say and I'll pull), but here's a rough to see how far I might take the file from the adobe default image ... and the penalties incurred. Highly subjective @ rendering liberties of course.
Edited on Nov 06, 2013 at 10:18 AM · View previous versions
All images were taken at Iso 100
I varied the shutterspeed from 1 second for the Iso 25 shot to 1/125 for the Iso 3200 shot.
Then I compensated for the over/under-exposure in LR, to equalize the brightness
so the "Iso 3200" shot was pushed 5 stops.
I used M mode
Thanks Phillip ... that's exactly the BIG CRAYON that I needed. Sorry for the confusion that I wasn't as adept to follow earlier as the others did.
Thanks to Phillip (et al). I'm still "shaking head" a bit, but now in wonderment @ why moreover than disbelief. Now it would be really interesting to see the same test Sony vs. Canon (link to pre-existing examples would be fine).
Judging from your tests (both Phillip & sebboh) ... which of the two exposure strategies would you suggest using as best for highlights, best for shadows?
I hope you don't mind (just say and I'll pull), but here's a rough to see how far I might take the file from the adobe default image ... and the penalties incurred. Highly subjective @ rendering liberties of course.
Edited on Nov 06, 2013 at 10:18 AM · View previous versions
Nov 06, 2013 at 05:05 AM
p.2 #16ettr - an urban legend?
ETTR was (and still is) necessary to get the best results from most any DSLR, where the histogram and preview are based on your JPEG settings and do not accurately reflect the actual RAW data being captured.
If you don't shoot RAW, it's probably not a big deal.
If you don't shoot RAW, it's probably not a big deal.
Nov 06, 2013 at 06:50 AM
p.2 #17ettr - an urban legend?
Phillip Reeve wrote:
okay, my explanation seems to be crap.
*All images were taken at Iso 100
*I varied the shutterspeed from 1 second for the Iso 25 shot to 1/125 for the Iso 3200 shot.
*Then I compensated for the over/under-exposure in LR, to equalize the brightness
*so the "Iso 3200" shot was pushed 5 stops.
*I used M mode
Hi Phillip,
So, let's take a look at your experiment and see what it is that was accomplished. By your explanation the exposures went like this...
pseudo ISO 25 @ 1 second
50 @ 1/2
100 @ 1/4
200 @ 1/8
400 @ 1/15
800 @ 1/30
1600 @ 1/60
3200 @ 1/125
...correct? This being so, then the exposures that were ETTR in this experiment were the pseudo ISO 50 @ 1/2 second and the 25 @ 1 second images which, obviously, were over-exposed when you took them. You then created the pseudo ISO's of 50 and 25, respectively, by bringing those image exposures "down" in LR to visually match (as close as possible) the median exposure, ISO 100 @ 1/4 second.
Now, this all being correct, so far, then the rest of the images from pseudo ISO 200 @ 1/8 second up to pseudo ISO 3200 @ 1/125 second were all under-exposed when you took them—or, in this case, exposed-to-the-left (ETTL...not to be confused with E-TTL for flash metering!)—and you had to bring all of those exposures "up" in LR to, again, match (as close as possible) the median ISO 100 @ 1/4 second image.
So far, so good?!
This then makes the in-the-field SOOC exposure sequence +2, +1, 0, -1, -2, -3, -4, -5 as hiepphotog posted earlier.
And, this would then have to make the exposure compensation sequence in LR, -2, -1, 0, +1, +2, +3, +4, +5 in the attempt to create the normalized images from pseudo ISO 25 to pseudo ISO 3200.
Everyone still on-board? Great!
So, the question becomes...what should one expect to see in this sequence of images SOOC?
In a nutshell, the over-exposed images will show the data getting pushed to the "right" in the histogram with, perhaps, the highlights getting "blown" and becoming lost in pure unrecoverable white. And the under-exposed images will show the data getting pushed to the left in the histogram with, perhaps, the shadows getting "crushed" and becoming lost in pure unrecoverable black.
And then the question becomes...which of these exposure provides the most exposure latitude for the scene that has been shot? And here, a quick and simple forensic investigation reveals the truth...
Examining the OP's sequence of "normalized" images we can still see the effects of his exposure choices. The "ISO 25" (over-exposed 2-stops) image clearly shows the lost highlights we expected (note the leaf in the left column". And the "ISO 3200" (under-exposed 5-stops) no longer has enough highlight detail to bring the exposure of the leaf in the left column back up to the exposure level of the other images.
But what of the shadows? Well, I opened the OP's sequence image in an editor and then adjusted the Gamma (middle slider in the Levels histogram control) to see what was hiding in the dark area in the right column. And, what do we find, even after the OP had tried to normalize these images? That there is clearly more preserved details to be found in the shadow areas of the "ISO 25" and "ISO 50" images and that the shadow detail quickly begins to show sensor noise from the actual ISO 100 through "ISO 3200" images.
So, what's the conclusion here?
Two things...The OP has demonstrated that ETTR is an effective method for preserving shadow detail. And that employing ETTR needs to be balanced so one does not blow their highlights. For this scene an ETTR of somewhere between +1 and +2 stops would have been ideal/appropriate.
Well, that's how this ref is calling it.
Jimmy G
okay, my explanation seems to be crap.
*All images were taken at Iso 100
*I varied the shutterspeed from 1 second for the Iso 25 shot to 1/125 for the Iso 3200 shot.
*Then I compensated for the over/under-exposure in LR, to equalize the brightness
*so the "Iso 3200" shot was pushed 5 stops.
*I used M mode
Hi Phillip,
So, let's take a look at your experiment and see what it is that was accomplished. By your explanation the exposures went like this...
pseudo ISO 25 @ 1 second
50 @ 1/2
100 @ 1/4
200 @ 1/8
400 @ 1/15
800 @ 1/30
1600 @ 1/60
3200 @ 1/125
...correct? This being so, then the exposures that were ETTR in this experiment were the pseudo ISO 50 @ 1/2 second and the 25 @ 1 second images which, obviously, were over-exposed when you took them. You then created the pseudo ISO's of 50 and 25, respectively, by bringing those image exposures "down" in LR to visually match (as close as possible) the median exposure, ISO 100 @ 1/4 second.
Now, this all being correct, so far, then the rest of the images from pseudo ISO 200 @ 1/8 second up to pseudo ISO 3200 @ 1/125 second were all under-exposed when you took them—or, in this case, exposed-to-the-left (ETTL...not to be confused with E-TTL for flash metering!)—and you had to bring all of those exposures "up" in LR to, again, match (as close as possible) the median ISO 100 @ 1/4 second image.
So far, so good?!
This then makes the in-the-field SOOC exposure sequence +2, +1, 0, -1, -2, -3, -4, -5 as hiepphotog posted earlier.
And, this would then have to make the exposure compensation sequence in LR, -2, -1, 0, +1, +2, +3, +4, +5 in the attempt to create the normalized images from pseudo ISO 25 to pseudo ISO 3200.
Everyone still on-board? Great!
So, the question becomes...what should one expect to see in this sequence of images SOOC?
In a nutshell, the over-exposed images will show the data getting pushed to the "right" in the histogram with, perhaps, the highlights getting "blown" and becoming lost in pure unrecoverable white. And the under-exposed images will show the data getting pushed to the left in the histogram with, perhaps, the shadows getting "crushed" and becoming lost in pure unrecoverable black.
And then the question becomes...which of these exposure provides the most exposure latitude for the scene that has been shot? And here, a quick and simple forensic investigation reveals the truth...
Examining the OP's sequence of "normalized" images we can still see the effects of his exposure choices. The "ISO 25" (over-exposed 2-stops) image clearly shows the lost highlights we expected (note the leaf in the left column". And the "ISO 3200" (under-exposed 5-stops) no longer has enough highlight detail to bring the exposure of the leaf in the left column back up to the exposure level of the other images.
But what of the shadows? Well, I opened the OP's sequence image in an editor and then adjusted the Gamma (middle slider in the Levels histogram control) to see what was hiding in the dark area in the right column. And, what do we find, even after the OP had tried to normalize these images? That there is clearly more preserved details to be found in the shadow areas of the "ISO 25" and "ISO 50" images and that the shadow detail quickly begins to show sensor noise from the actual ISO 100 through "ISO 3200" images.
So, what's the conclusion here?
Two things...The OP has demonstrated that ETTR is an effective method for preserving shadow detail. And that employing ETTR needs to be balanced so one does not blow their highlights. For this scene an ETTR of somewhere between +1 and +2 stops would have been ideal/appropriate.
Well, that's how this ref is calling it.
Jimmy G
Nov 06, 2013 at 08:52 AM
p.2 #18ettr - an urban legend?
jimmy462 wrote:
Two things...The OP has demonstrated that ETTR is an effective method for preserving shadow detail. And that employing ETTR needs to be balanced so one does not blow their highlights.
I'm still confused about the different ways people define ETTR. Is the general definition to overexpose the midtones to get less noise in the shadows? I often think that underexposing to align the highlights to the right end of the histogram also means exposing to the right (because you're aligning it to the right), but I guess that's called exposing to the left because you're underexposing and moving the levels to the left?
Maybe simply saying expose for the shadows or highlights leaves less room for confusion?
Two things...The OP has demonstrated that ETTR is an effective method for preserving shadow detail. And that employing ETTR needs to be balanced so one does not blow their highlights.
I'm still confused about the different ways people define ETTR. Is the general definition to overexpose the midtones to get less noise in the shadows? I often think that underexposing to align the highlights to the right end of the histogram also means exposing to the right (because you're aligning it to the right), but I guess that's called exposing to the left because you're underexposing and moving the levels to the left?
Maybe simply saying expose for the shadows or highlights leaves less room for confusion?
Nov 06, 2013 at 09:08 AM
p.2 #19ettr - an urban legend?
I think exposing to the right means that you have to decide which highlights you want to preserve and which you can sacrifice. If you want to preserve all the highlights, in some situations, this means severe underexposure. But normally ETTR tends to overexpose more often than not.
Nov 06, 2013 at 09:12 AM
p.2 #20ettr - an urban legend?
^But that makes it an ambiguous term. I'd think that exposing for the highlights means you clip no highlights, and exposing for the shadows means you don't clip shadows (and necessarily clip highlights if the scene's DR exceeds the camera's).
Nov 06, 2013 at 09:40 AM