AhamB wrote:
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?
Correct, increase exposure without blowing the highlights...that's ETTR.
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?
That "align the highlights to the right of the histogram" is a fair method for determining, in the field, that one's highlights haven't gone overboard when increasing exposure. That folks are using the same method (align the highlights to the right of the histogram) to bring down their exposures in an attempt to preserve their highlights makes sense, as well, and in that scenario they are exposing to the left. I can see where folks can get confused.
Maybe simply saying expose for the shadows or highlights leaves less room for confusion?
That works for me, it's all just different language. I, personally, expose for my subject and if that subject is in the shadows I might not necessarily mind losing my highlights. Or vice versa, that my subject is well lit I might not mind crushing my blacks. It's all subjective based on what feel we're looking to achieve with our image. There are no hard set rules, expose to suit your desired goals. All we're talking about here are techniques to achieve those goals and "feels".
^^ Yes, ETTR will overexpose. But the point is you can recover that blown out highlight later in post. Often time, you don't have to sacrifice much (if at all) in return for a better shadow detail/noise profile. Considering Sony tends to under-expose a bit to preserve the highlight, pushing to the right might actually help to bring back the "normal" exposure.
ISO 100 @ 1 second
ISO 100 @ 1/2
ISO 100 @ 1/4
ISO 100 @ 1/8
ISO 100 @ 1/15
ISO 100 @ 1/30
ISO 100 @ 1/60
ISO 100 @ 1/125
Hi RustyBug,
My apologies if I fanned the flames of confusion, I thought I was being clear. Your sequence...
ISO 100 @ 1 second
ISO 100 @ 1/2
ISO 100 @ 1/4
ISO 100 @ 1/8
ISO 100 @ 1/15
ISO 100 @ 1/30
ISO 100 @ 1/60
ISO 100 @ 1/125
...is 100% spot on with what the OP did in the field. The way I portrayed that data was an attempt to show how the OP achieved his pseudo ISO's in LR. The long-hand version between your version and mine should read...
ISO 100 @ 1 second = pseudo ISO 25
ISO 100 @ 1/2 = pseudo ISO 50
ISO 100 @ 1/4 = ISO 100
ISO 100 @ 1/8 = pseudo ISO 200
ISO 100 @ 1/15 = pseudo ISO 400
ISO 100 @ 1/30 = pseudo ISO 800
ISO 100 @ 1/60 = pseudo ISO 1600
ISO 100 @ 1/125 = pseudo ISO 3200
Again, my apologies if my shorthand confused matters!
Phillip Reeve wrote:
I was out in the woods this morning and because the light was boring and I had a tripod with me I shot a series to see how pushing or oulling affects image quality.
My conclusion based on this quick test is that with a low contrast scene there is no reason to expose to the right, what do you guys think?
I thought the purpose of ETTR was to maximize the amount of information collected by the sensor.
I think you did an ISO test, not an ETTR test, but I might be wrong.
To do an ETTR test, wouldn't you keep the ISO constant, and vary the exposure and compare the results?
No worries Jimmy G ... to further (hopefully) enhance my understanding of what we are saying here, is it then correct that the pseudo ISO 3200 is in essence the same capture @ RAW in camera base ISO with two different locations (one in camera vs. one in post) for applying the gain as if we had shot it at ISO 3200 and it would have allowed us to have a "normal" exposure with a shutter of 1/125 from the camera settings?
So, in essence ... everything we shoot is really at base ISO, just a matter of where/how much the gain algorithm is being applied (i.e. in camera @ OEM vs. in post OEM/3rd party), is that correct?
RustyBug wrote:
So, in essence ... everything we shoot is really at base ISO, just a matter of where/how much the gain algorithm is being applied (i.e. in camera @ OEM vs. in post OEM/3rd party), is that correct?
That is correct since you use the same shutter speed and aperture (same amount of light). ISO is just the internal gain. That begs the question which one would be a more "efficient" gain (in camera vs. in post). On the same topic, it's why many people argue on the merits of overexposing (higher ISO with faster shutter speed) vs. right exposure (lower ISO with slower shutter) in a high ISO shot for a cleaner noise. Rule of thumb is that you should not underexpose your high ISO shot.
sebboh wrote:
......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.
Agreed.
If anything I tend to expose to the left (at least in terms of low ISO). I find I can bring detail out of the shadows much easier than I can recover the highlights.
RustyBug wrote:
No worries Jimmy G ... to further (hopefully) enhance my understanding of what we are saying here, is it then correct that the pseudo ISO 3200 is in essence the same capture @ RAW in camera base ISO with two different locations (one in camera vs. one in post) for applying the gain as if we had shot it at ISO 3200 and it would have allowed us to have a "normal" exposure with a shutter of 1/125 from the camera settings?
So, in essence ... everything we shoot is really at base ISO, just a matter of where/how much the gain algorithm is being applied (i.e. in camera @ OEM vs. in post OEM/3rd party), is that correct? ...Show more →
There are two types of in-camera gain: analog and digital. Only the digital gain is similar to what you can do in post-processing and this is why people sometimes refer to "fake ISOs" when referring to in-camera ISO settings where digital gain is being applied. It is no better than just getting the raw file and pushing it in post.
Analog gain is different, though. It is applied before the analog-to-digital conversion happens in-camera and cannot be replicated in post.
RustyBug wrote:
No worries Jimmy G ... to further (hopefully) enhance my understanding of what we are saying here, is it then correct that the pseudo ISO 3200 is in essence the same capture @ RAW in camera base ISO with two different locations (one in camera vs. one in post) for applying the gain as if we had shot it at ISO 3200 and it would have allowed us to have a "normal" exposure with a shutter of 1/125 from the camera settings?
So, in essence ... everything we shoot is really at base ISO, just a matter of where/how much the gain algorithm is being applied ...Show more →
Hi RustyBug (I love that "handle" BTW),
If I'm understanding your question correctly...that shooting "pseudo ISO 3200" (as the OP did here, shooting an under-exposed ISO 100 at 1/125 image) will yield the same results as actually setting the camera to ISO 3200 and shooting a 1/125 second exposure?...the answer would be, no.
The reason for this is that each camera manufacturer employs their own system of signal amplifiers and algorithms to get their specific camera sensors to imitate/simulate the varying ISO's. The application of actual, in-camera signal gain from ISO 100 to 3200 is going to be applied differently not only between different image sensors, but also in how each of the manufacturers employs those sensors with the associated internal electronics and programming to perform the final sensor readout and processing to the card. The RAW image on your card should never be confused with the actual data dump that comes off of the image sensor itself, in-camera processing is done, already, at this early stage, just to create, say, a 14-bit, er, "RAW" image. Think of it as an electronic recipe. The JPG images then go through another sequence of steps to push (crunch, maim, or otherwise distort) certain numerical values into the user-specified color space (sRGB, Adobe RGB) and the 8-bit JPG depth. Pureed soup anyone?
How we can see this in the histogram is to reenact the OPs experiment...shoot a pseudo ISO 3200 (e.g. ISO 100 underexposed 5-stops) and shoot a camera ISO 3200 exposed 5-stops differently from the first and then open both images up in post. What one will discover when they stretch the pseudo ISO 3200 image to approximate the in-camera ISO 3200 image is that the histograms will be different between the two. The in-camera ISO 3200 image will provide more image and color information (read: data). The reason for this is that both methods will come up with different numerical values for each represented pixel. One is the LR (or other software) version of an interpreted ISO 3200 and the other is the in-camera soup recipe for an ISO 3200 image.
Are either of these true ISO 3200? No. They're both extrapolations off of the base ISO in the camera.
Which of these two methods will yield a more usable ISO 3200 image data set? The in-camera version, since it is dealing with less math than the software version. Or, to put it another way, the in-camera version is an interpretation, the pseudo software version is an interpretation of an interpretation.
jimmy462 wrote:
Hi RustyBug (I love that "handle" BTW),
Rusty ShutterBug @ dang I forgot a lot ... just for clarification, but I figure you already got that.
Are either of these true ISO 3200? No. They're both extrapolations off of the base ISO in the camera.
That is the essence of my question.
the in-camera version is an interpretation, the pseudo software version is an interpretation of an interpretation.
But, isn't the in-camera version also an interpretation/extrapolation of an interpretation (the RAW) as is the software version an interpretation/extrapolation of an interpretation (the RAW). Bearing in mind that the OEM software can be equally (is that correct ) applied to the RAW in post or in camera ... AND/OR ... third party software interpretation/extrapolation can be applied to the RAW (i.e. same as the pp application of the OEM software), or following the OEM (in camera or post) applied after the RAW.
Then, aren't all three routes (in camera OEM, out of camera OEM, out of camera third party) really secondary interpretations/extrapolations (albeit somewhat different algorithms) of the primary (RAW) interpretation. Only then wouldn't a jpg coming out of the camera and being further processed be an interpretation/extrapolation (third party) of an interpretation/extrapolation (jpg) of an interpretation (RAW)?
The objective of ETTR for me is to improve detail and reduce noise in important shadow areas. It is accomplished by increasing exposure as much as possible without clipping or distorting important highlights.
Shadow detail can be improved since ETTR can allow the linear raw file to differentiate more finely between different tonal values in the shadow areas and avoid posterization artifacts for our non-linear viewing. Here is a good explanation from Ron Bigelow, including a flow-chart that covers the process from photons to developed raw file.
Apparent noise can be reduced because ETTR is essentially increasing the signal to noise ratio. The noise floor remains fixed but the signal is being increased.
Note how I underlined "important" in the paragraph above. If shadows aren't important in a given shot, then ETTR is probably not necessary. If shadows are so unimportant and highlights are so important in a given shot, you may even want to consider ETTL so that highlights are guaranteed to be unclipped and undistorted.
IMO, highlights can be clipped or distorted in ETTR if those highlights are not important. Determining whether highlights and shadows are important and to what extent is where photographers need to make use of their brains.
ETTR is not very useful for scenes with wide dynamic range where all tones are "important". Something is probably going to be sacrificed in that scenario unless the lighting is modified or multiple exposures are taken for an HDR composite.
Where I often find ETTR comes in handy is for non-flash, indoor shots. I'm shooting at higher ISOs and the "normal" exposure will often be bunched up around the middle of the histogram with a trail into the shadows. If there are any strong highlights, they are often light sources captured within the frame and I may not care whether those spots are blown or accurately captured. Under those situations, an ETTR can be helpful to reduce noise and improve shadow detail and I often find it worth bumping up the ISO to achieve ETTR, as long as I am not using a "fake" ISO that is employing digital gain instead of analog.
Regarding the OP's test shots, since the main purpose of ETTR is to improve the rendering of shadow tones, I think that is where you need to focus the comparison. Here is a re-edit of the OP's samples, concentrating on enlargements of the shadow/leaves:
michael49 wrote:
If anything I tend to expose to the left (at least in terms of low ISO). I find I can bring detail out of the shadows much easier than I can recover the highlights.
This ^. Some color channels can blow out with the histogram looking just fine (reds and yellows, ugh) and so if anything I am more likely to ETTL to preserve colors.
However, I shot 4/3 for a while a few years back, and ETTR made a huge difference then. Seems like the camera bodies then tended to underexpose, and were unforgiving for lifting shadows. It was less of a big deal when I moved to Canon, and now is mostly a non-issue with Exmoor sensors.
Now days, I do ETTR a bit with Sony if I'm in really low light (high ISO), and I know highlights are going to make the camera underexpose. Then I'd rather blow out a few highlights and have less noise in the overall image. I feel like there is only so much you can pull up exposure at ISO3200 and beyond w/o IQ becoming friable.
mco_970 wrote:
Some color channels can blow out with the histogram looking just fine (reds and yellows, ugh) and so if anything I am more likely to ETTL to preserve colors.
If you are shooting Raw, I recommend you double-check your raw processing. In my experience, blown reds and yellows are usually a raw conversion problem (going from wide-gamut to small-gamut) more than a true exposure problem.
JohnJ wrote:
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.
The 5dII surely is worse than the Nex-6 but it isn't that much, a little over 1 EV
Phillip, that may only be one stop in some technical way of measuring it, but just looking at it, it is horrendous. Whereas the Sony has nice crisp, monochrome noise, the Canon has colour sprinkles, banding and blotching. One stop will nowhere near fix that, visually, IMO. I would be surprised if two stop would.
Canon really needs to bite the bullet and get a new fab up and running.
RustyBug wrote:
Rusty ShutterBug @ dang I forgot a lot ... just for clarification, but I figure you already got that.
Thanks, I needed a good chuckle today! But I'm thinking you ain't as rusty as your lettin' on...
That is the essence of my question.
But, isn't the in-camera version also an interpretation/extrapolation of an interpretation (the RAW) as is the software version an interpretation/extrapolation of an interpretation (the RAW). Bearing in mind that the OEM software can be equally (is that correct ) applied to the RAW in post or in camera ... AND/OR ... third party software interpretation/extrapolation can be applied to the RAW (i.e. same as the pp application of the OEM software), or following the OEM (in camera or post) applied after the RAW.
Then, aren't all three routes (in camera OEM, out of camera OEM, out of camera third party) really secondary interpretations/extrapolations (albeit somewhat different algorithms) of the primary (RAW) interpretation. Only then wouldn't a jpg coming out of the camera and being further processed be an interpretation/extrapolation (third party) of an interpretation/extrapolation (jpg) of an interpretation (RAW)?...Show more →
Yeah, you caught me doing some more shorthand here...the "interpretation of an interpretation" I was referring to was that final step needed for stretching the under-exposed ISO 100 1/125-second histogram out to make it a pseudo ISO 3200. I just omitted all those extra "interpretations" along the way up to that point for clarity's sake. Thems was a whole bunch of extra "interpretations" to type, and I wasn't gonna do it!
RustyBug wrote:
No worries Jimmy G ... to further (hopefully) enhance my understanding of what we are saying here, is it then correct that the pseudo ISO 3200 is in essence the same capture @ RAW in camera base ISO with two different locations (one in camera vs. one in post) for applying the gain as if we had shot it at ISO 3200 and it would have allowed us to have a "normal" exposure with a shutter of 1/125 from the camera settings?
So, in essence ... everything we shoot is really at base ISO, just a matter of where/how much the gain algorithm is being applied (i.e. in camera @ OEM vs. in post OEM/3rd party), is that correct? ...Show more →
No, that is not correct. The main problem is addition of noise sources, and how they interact.
I might have trouble simplifying this, but I'll give it a (short) shot.
You can simplify the noise additions to the signal in the camera into two stages - one is close to the pixel, one is close to the A/D conversion. The "real" analog ISO amplifier is closer to the pixel. I'll call them "n1" and "n2" here. NB - this is not a normal convention.
In low light, you need to amplify the signal to make it brighter - the question is just where you do it.
If you do it in post, you're basically applying a multiplication to the SUM of n1 and n2.
If you do it in the analog ISO amplifier close to the pixel, you're just multiplying n1 - the late stage noise n2 is the same as before, and is added as normal without multiplication!
Say we need a 10x brightening (go to ISO1000, if the camera's base ISO is 100). I'll throw in some arbitrary numbers of n1=5 and n2=20 just to show what happens.
Brighten in post:
Total noise = (brightening factor) * (sqrt(n1^2+n2^2))
>>>> 10 * sqrt( 5^2 + 20^2 ) = 206.1
Brighten in analog, with ISO amplifier:
Total noise = sqrt ( (brightening factor * n1)^2 + n2^2)
>>>> sqrt( (10*5)^2 + 20^2) = 53.9
-the noise is much lower when you use the on-sensor ISO amplifiers, because of the fact that you don't have to multiply the late stage noise too...
In cameras with very low n2 noise (late stage noise) you don't lose as much by amplifying in post. Say that we lower n2 to "10" in stead - that's like getting quieter A/D converters and a cleaner signal path.
Post : 10 * sqrt( 5^2 + 10^2 ) = 111.8
ISO : sqrt( (10*5)^2 + 10^2) = 50.9
Note that the "push in post" noise got a LOT lower - but the "amplify with analog ISO" didn't get much better at all. That's because of the n1 having to be amplified no matter what, there's no getting around that if the image is to get brighter.
That is (fairly simplified) why you can push a Sony sensor quite high from base ISO without the image falling apart, but you can't do it with a Canon sensor. The Canon sensor has good high ISO in camera, since it has excellent n1 noise. But it's limited at low ISOs by the rather strong n2 noise, and that's amplified by pushing in post.
This is also why Canon sensors have lower base ISO DR's, since DR is dependent on both n1 and n2. Higher ISOs are almost purely dependent on n1.
jimmy462 wrote: Thanks, I needed a good chuckle today! But I'm thinking you ain't as rusty as your lettin' on...
That's only cause a lot of good fellow FM'ers have been greasin' my squeaky wheels since I signed on with Fred and company. Trust me, I was WAY MORE rusty back then, but I still have areas that need more shaken off. Likely I'll never be finely polished, so if you're a fan of rust or patina ...
Joakim,
Diggin' the numbers ... I can follow well enough.
My question is what distinguishes low n1 vs. high n1 or low n2 vs. high n2 in different cameras?
I'd mostly heard that large pixels = low noise for low iso, small pixels = low noise for high iso (or some corollary along those lines) ... never really sure-sure if I understood it correctly or why? I shoot mostly base ISO 160 with large pixels on SLR/C, and occasionally shooting some ISO 3200 with 1D II N for low light (also shoot 1D II N @ base ISO 100 - ISO 400 for "action" @ sufficient light).
Given those two cameras ... and maybe something like the Pentax K3 (looked at today), how do we know when to better harness n1 vs. n1 + n2?
Thanks, I needed a good chuckle today! But I'm thinking you ain't as rusty as your lettin' on...