craygc
Well-known
The problem that relying on the histograms for exposure mins and maxs is that what it's showing you is a jpeg representation. You're actually going to have a lot more latitude in the raw file.
Invariance
- Thread starter Bill Pierce
- Start date
- Latest activity Latest activity:
- Replies 33
- Views 2K
The problem that relying on the histograms for exposure mins and maxs is that what it's showing you is a jpeg representation. You're actually going to have a lot more latitude in the raw file.
peterm1
Veteran
Thanks Tim I will bear that in mind. For most of my time with my M8 I have tended to view it the same way as your M9 - good for bright scenes but disappointing in situations in which old Leica film M cameras excelled - available light when light was a little lacking. And of course in high contrast situations where CCD sensors do not cope well.
I was of course aware of the advice about shooting at native ISO and deliberately under exposing as my previous post indicated and have started trying it and found it can be useful. Though I have not been terribly serious about it. So for most day to day shooting I have been lazy and tend to just use auto ISO - though I keep my sensitivity limits between 160 iSO and 640 IS0 so I do not go overboard in this respect.
However I should really start to seriously use the technique of shooting at 160 ISO (base ISO), under exposing by a stop or more depending on the lighting and then try pushing the image in post, as my main every day technique. I have done it half heartedly / experimentally and while I have found it works I need to put it to serious use. I guess its habit that has prevented me from doing so, and the fact that the M8 does not have an EV dial. This means if I need to vary this parameter due to an ambient lighting change I need to delve into menus. Which is a pain. I suppose the alternative is just to set aperture and shutter speed wholly manually and deliberately under expose if necessary in that manner.
May I ask, have you tried under exposing by larger amounts than 1 EV? If so how well does this work as a technique with your sensor? (If the ambient lighting is very poor I do not necessarily want to have unduly blurry shots due to low shutter speed so more than 1 EV may be needed in some situations).
retinax
Well-known
I took advantage this often when I shot digital. But while it worked great for b&w, I noticed that the colors tended to go off. Maybe that was just my software, Darktable?
Timmyjoe
Veteran
Hi Peter,
For some reason I can't find the original files for the image below, but from recollection, this was shot on my M9, ISO 640 (the max ISO I used on the M9), and EV -1, then corrected in Lightroom, (as per the instructions in the links I posted above.)
I've gone as far as EV -2 with the M9 but for most uses, EV -1 works better. It's a trial and error thing, but you could easily test it by shooting a low light image as EV -1 then EV -2, and see which one looks better to you when you correct exposure in Lightroom.
Best,
-Tim
ptpdprinter
Veteran
How does underexposure with correction in Lightroom comport with the admonition to expose to the right. They seem at odds.
Gregm61
Well-known
Yep, you figure out pretty quick how far "off the scale" on the right-hand side you can go with the preview and wind up with a file that holds all of the highlight detail. It gets you way closer than taking a shot or two and often having to re-shoot because you couldn't see the exposure preview in the optical finder beforehand.
Contarama
Well-known
+1 what he said
And dr is maximized as well
It would be interesting to see how all of this applies to a Fuji S5 Pro
Dogman
Veteran
willie_901
Veteran
Here's a some data. These data show read-noise levels vs ISO setting normalized for ISO gain. They estimate read-noise levels at the photo-diode level.
FUJIFILM
Nikon
Olympus
SONY
The single large change in sensor read-noise levels is due an increase in photodiode conversion gain. More conversion gain reduces read noise at the expense of maximum photo-diode full-well capacity (dynamic range). In bright light always use the low ISO setting range. In low light use the high ISO range.
These data depict the two pseudo ISO-invariant ranges for some of these cameras. Read noise is most obvious in shadow regions. Increasing DC signal gain (i.e. increasing ISO setting) does not affect the noise level within ~ +/- 1/3 stop.
Note: These data are from statistical analyses of un-rendered, raw-file data.
Here's some data for Leica M bodies.
Input-referred Read Noise VS ISO
Shadow-Region Improvement vs ISO
willie_901
Veteran
They aren't.
Underexposure always reduces the analog signal signal-to noise ratio. Exposure only depends directly on shutter time and, or aperture. This means using practical shutter times and, or apertures often results in unavoidable sensor underexposure. There are two very different ways to compensate for sensor underexposure.
At the camera's native (base) ISO maximizing exposure (shutter time and aperture) maximizes the signal-to-noise ratio for the data. By coincidence maximizing exposure produces an expose-to-the-right histogram. Using a tripod with static subjects means the sensor exposure can be maximized. But subjects in motion require shorter shutter times. Then practicality demands sensor underexposure.
When practically requires underexposure, at the native ISO setting the rendered image will appear too dark. There are two ways to increase the final image's brightness. One is to increase the brightness digitally in post-production. The other is to increase the gain of the analog signals before they are digitized in-camera. This is accomplished by the setting the camera's ISO setting above the native value.
In the first case, the in-camera histogram will not be biased to the right due to sensor underexposure. In the second case the histogram may appear to be biased to the right. But it is identical to the first histogram. Increasing ISO also changes the exposure index calibration which simply shifts the histogram scale.
In the first case the histogram is a guide to avoid exceeding photo-diode full-well capacity. But in the second case the histogram becomes a guide to avoid exceeding the analog-to-digital converter's maximum signal threshold (clipping). These are very different situations.
In low light underexposure is often unavoidable. For some camera designs, increasing analog signal gain after the shutter closes increases the read noise levels. Underexposing at a lower ISO and increasing the brightness in post-production will result in cleaner shadow regions. However, now the best possible post-production rendering requires using raw files. In-camera JPEGs lossy compression destroys some of information needed to properly brighten the image.
With ISO-invariant designs the read-noise is essentially constant with respect to ISO setting. Shadow region SNR is not affected by read noise as ISO increases. One can use base ISO, raw files and just set the shutter time and aperture as needed. The only concern is exceeding sensor full-well capacity in bright light. Otherwise the histogram and, or meter can be ignored. When underexposure becomes unavoidable it will be impossible to clip the ADC. Image brightness is optimized in post-production.
Underexposure always increases the relative photon noise level. Camera ISO settings (electronic gain) have no impact on photon noise levels. Since read noise is essentially constant, the noise levels for ISO-invariant cameras are dominated by changes in photon noise. Otherwise, both read noise and photon noise degrade IQ as underexposure increases and the ISO setting is increased to brighten the rendered image.
Bill Pierce
Well-known
On a practical level, for me, expose to the right has always been the way to capture the greatest brightness range in the file. With today’s sensors that can be pretty impressive, and many photographs are not going to need it. But when they do, I set my camera to its base ISO and bracket my exposures, selecting the most generously exposed one that still holds highlight detail. I think with early digital you looked to “exposure to the right” not only to maximize brightness range but to minimize noise in the darker values of your final print. Today I just use it to go for that tonal range, and it’s not something I do a lot. Truth is, I’m sort of an “expose to the left” person having spent a lot of years shooting transparency film, worrying about highlight detail and learning to live with the shadow detail it provided. The same phobia seemed to produce decent digital files.
I think willie 901 is the expert here. If he says this is wrong, I’ll have to consider changing my ways. I hope “exposing to the left” is pretty good standard practice and “exposing to the right” is good specialty practice.
kbb
Well-known
Bill, you've nailed the 2 most important points for the actual photographer:
1. ETTRight was desirable, and even necessary, for the early limited-range sensors.
2. ETTLeft is desrable nowadays for extreme wide range scenes, especially ones with super bright spots and/or specular highlights in which the photographer wishes to preserve detail.
1. ETTRight was desirable, and even necessary, for the early limited-range sensors.
2. ETTLeft is desrable nowadays for extreme wide range scenes, especially ones with super bright spots and/or specular highlights in which the photographer wishes to preserve detail.
pgk
Well-known
Whilst I can see logic and reasoning in the posts here, am I not correct in thinking that this would all be easy to put into practice IF the RAW file had not bee subjected to the manufacturer's adjustment prior to being written to the memory card? My experience with ETTR almost always leads to some discrepancy in tonality which I assume to be because the manufacturer has tweaked the RAW file for reasons which they anticipate will optimise it. So RAW files are not truly RAW and are subjected to some form of adjustment.
willie_901
Veteran
That's what I do. I assume you're using raw files.
Exactly.
Regardless of what camera we use the goal is to achieve the highest level of useful exposure without exceeding the full-well capacity of the sensor. This maximizes the signal and minimizes the photon noise levels. This is all we can do.
The difference between early and contemporary digital involves relative read noise levels. Contemporary designs have very low read noise levels over a wide range of ISO values (at least 4 stops above base ISO). Even so, deep shadow regions (more than 5 stops underexposed compared to highlight regions) can be affected by read noise even at base ISO. Relative read noises levels are much higher for early designs. Different early designs have different read noise vs ISO profiles. So some ISO values should avoided in low light.
The first place I heard about this was in a long thread about ETTR in another Forum where Emil Martinec wrote this about ETTR:
"What is the appropriate mantra?* I would prefer "Maximize Exposure"; maximize subject to three constraints:
(1) maintaining needed DoF, which limits how much you can open up the aperture;
(2) freezing motion, which limits the exposure time;
(3) retaining highlight detail, by not clipping wanted highlight areas in any channel.*(emphasis mine)
Note that ISO is not part of exposure. Exposure has only to do with aperture and shutter speed. Maximizing exposure guarantees that one captures as many photons as possible subject to photographic constraints, and therefore optimizes S/N."
Step 3 implies one should intentionally overexpose highlights that are not aesthetically relevant. Examples are specular highlights during daytime and bright point source lights at night. This is why I auto-bracket aperture by +/- 1/3 stops. I can keep the raw file that retains the important highlight regions.
There are three disadvantages to this method.
- You have to use raw files
- Chimping is compromised since the in-camera JPEG will often be too dark
- There is some uncertainty about judging deep shadow region IQ
- Camera profiles used in some post-production rendering software generates hue twist with large increases in rendering brightness (the Exposure slider). This means increasing brightness may require selective changes in hue rendering parameters. Sometimes this is most obvious for skin-tone rendering.
The advantage is the raw data will contain the maximum possible amount of information. In other words, SNR will be optimized.
A minor advantage is you can ignore thinking about choosing a camera ISO setting – or wondering if auto-ISO is letting you use the optimum shutter time and, or aperture.
