camerapapi Loc: Miami, Fl.

Gentlemen, aren't we getting too passionate about this? We all see things differently and we have different opinions.
In the past we accepted grain (noise for digital) as a fact and I still remember those beautiful enlargements when using ISO 400 film in wedding sessions using Hasselblad cameras and Zeiss lenses that had lots of noise and we found them beautiful. Today such grain (noise) seems to be objectionable to so many photographers.
I do use ETTR depending on subject. Many times I extend the dynamic range of my cameras using HDR techniques. I guess each one of us has a way to do things.
Perhaps they are my eyes but I can see grain in EACH of the images posted. As the dynamic range is extended the photographs look better but I still can see chroma noise. To me it is not objectionable. I see good contrast in the shots and to my eyes those shots do not look muddy.
Today we have excellent noise reduction softwares for when using very high ISO settings is a necessity. Dynamic range with new cameras is superb and we can open up to 3 stops of light in the shadow areas without unwanted results most of the time.
I believe the presence of noise in many b&w photographs simply adds to the mood.
Many of my photographs have been saved by ETTR and as I said, it is not a technique for all subjects.
I will not be following this subject so accept or reject my statements at will.

lsimpkins Loc: SE Pennsylvania

This is wrong. An understanding of the physics of semiconductors would show that the noise in the sensor is due to the particular process used to make it. This basically defines the thermal noise (constant with temperature) as well as the shot and/or popcorn noise. So if you add more light to the image, you increase the image-based signal but not the noise which is relatively constant at a given temperature. The net of this is that the signal to noise ratio is greater, giving the appearance of less noise, even though the amount of noise is the same.

Changing ISO only changes the amount of amplification applied to the total signal (made up of the light dependent signal and the semi-constant noise) coming from the sensor array. That amplification increases both the image-based signal and the inherent noise. Since the noise is amplified, it becomes more apparent in the final output.

imagesintime Loc: small town, mid-America

Please explain the physics of how the sensor in your 'digital' camera works.

via the lens Loc: Northern California, near Yosemite NP

Just an FYI on digital and the Zone system. Digital is like slide film and thus you expose for the highlights and let the "shadows fall where they may." (Quote from Ansel Adams.) As you are aware, digital picks up the shadows fairly easily.

via the lens Loc: Northern California, near Yosemite NP

On No. 2, a somewhat technical explanation for why it improves color tones in photographs. It's based on the way a sensor actually works.

ETTR is intended to be used with RAW images. When shooting JPG one would try to go as far to the right as possible but not get anywhere near blown-out as it is difficult to retrieve data with JPG files, so staying in the midtone-low white range. So, why would anyone want to shoot to the right anyway? The answer is to obtain all of the data needed to get the absolute most color tones in the photo (and this is why some people choose to shoot RAW images as JPG images only have 255 tones of color). If you recall, a histogram is a statistical bar graph of a frequency distribution (tones), going from 0 to 255 for an 8-bit file (JPG), and helps us to know if we are exposing correctly. It also relates back to how sensors capture data, in a linear fashion. ETTR means the image histogram is skewed to the right as a slight overexposure but without blowing out the highlights. Most camera histograms show only five stops of tonal value, but many cameras can actually capture 7 stops of tonal value (know your specific camera). In RAW shooting exposing for the highlights is standard practice. Since a 12-bit RAW file (sensor data is commonly quantized to 10, 12 or 14 bits) contains 4096 tones (212 ) and the number of stops (tones) are halved (cameras tend to do this!) you will get 1024 tones in the far right of the RAW histogram. Then the 1024 tones are halved again, giving you 512 tones in the next section, and 256 tones in the next section, and 128 tones in the next section. This breakdown gives you the five sections of your camera and post-processing histogram: Using Lightroom language this equates to the following RAW histogram zones or tones (color or BW):
Whites-2048
Highlights-1024
Exposure-512 (Mid-tones)
Shadows-256
Blacks-128
By exposing to the right you obtain the highest degree of color tones and thus get brighter and/or deeper colors in your image. This is why you can retrieve a slightly blown-out sky with a RAW file but not with a JPG file. Shooting to the right also allows you to take high-key photos, where the image is in the mid-tone range but the background is actually blown out and white, often done in fashion photography. If you want to know more about this you might purchase The Manual of Photography or other books that explain this.

Bill_de Loc: US

Since we are speaking of photography and not science, isn't the appearance the only thing that matters? We use lighting techniques to add the appearance of depth in a photo. It doesn't really matter that it doesn't add depth to the paper it is printed on.

Just my two cents.

selmslie Loc: Fernandina Beach, FL, USA

I had hoped that by now everyone reading this thread would have realized that this is exactly what I am talking about - the appearance or visibility of noise, the noise that can be seen in the final image.

It is, of course, the result of the ratio of the signal (which you can increase with more exposure) to the base noise (which you can safely assume to be relatively constant). The stronger the exposure the higher will be the ratio of signal to noise (S/N) and the less visible the noise becomes until it virtually disappears.

The other factor to consider is the camera's gain. The higher the ISO the higher the gain. The more gain the more that noise is amplified. So both things are working against you at high ISO. You are amplifying the noise more and you are recording less exposure to get the same basic image.

selmslie Loc: Fernandina Beach, FL, USA

Yes and your two cents are worth a million.

There has been entirely too much toil and sweat over the numbers and not enough evidence of the actual appearance of the image, which is all that matters.

This is why I challenged the proponents of ETTR/EBTR to produce evidence that their efforts to expose to the right and immediately return the recorded image to the left through post processing can actually be seen in the final result.

So far nobody has been able to show that it is better than simply exposing normally at a low ISO while watching out for blinkies.

selmslie Loc: Fernandina Beach, FL, USA

I have taken this analysis even further and displayed the results in a couple of spreadsheets here and here.

The point of the first spreadsheet is that there are more than enough tonal values in a normally exposed image of a scene with a 5-7 stop dynamic range to render colors and shades of gray properly at about 30 distinct values per zone.

The second spreadsheet shows that the only way to use the entire physical capacity of the raw sensor it to use base ISO.

tainkc Loc: Kansas City

Well, actually there is a very good reason for it; and going to the left as well. This comes in handy when your model is wearing white against a white background. Same thing goes for black on black and going to the left. This technique helps to separate the two. This technique is usually done in aperture priority obviously even though I have this option in my camera while in manual mode. Like I said, this comes in handy. One still needs to make the appropriate ISO, speed and initial aperture settings. Simple. Look it up. I am not going to do your homework for you. If I recall, B&H even had a tutorial on this subject on what it does and how & when to use it.

selmslie Loc: Fernandina Beach, FL, USA

I hate to refer you back to the entire list of threads but that has already been covered.

selmslie Loc: Fernandina Beach, FL, USA

Do you think I don't understand exposure? We are well beyond that.

The question is whether ETTR/EBTR adds any value to the exercise.

amfoto1 Loc: San Jose, Calif. USA

You're massively over-complicating what is actually a rather simple technique...

ETTR is nothing more a precautionary measure... a little extra effort to avoid accidental under-exposure when using any auto exposure mode. ETTR simply says it's better to err toward slight over-exposure, than it is to under-expose. ETTR recognizes that digital imaging has some "extra headroom" in the highlights, so long as it isn't pushed too far so that they are blown out.

The reason for ETTR is that any time you need to increase or "push" exposure up in post-processing it will also dramatically increase the appearance of any noise in your image.

There will be less noise in image when exposure is "correct" or even if it's "pulled" down a little from slight over-exposure in post-processing.

ETTR isn't a perfect solution. Every shooting situation is different and it merely increases the odds that you'll get usable images when you have to shoot with auto exposure modes.

It will also vary by manufacturer and possibly even by camera model, as well as some user techniques (such as the metering pattern they tend to use). One manufacturer might calibrate their cameras with a slight bias to over-expose, making ETTR unnecessary. Or the metering system in one model might behave differently from that in another. Or someone who uses Spot metering a lot will likely find that their need for ETTR is different from someone who uses Center Weighted or Matrix/Evaluative metering. Depending upon how the Spot metering shooter tends to use their camera, they may need more or less ETTR. It really depends upon the person.

You actually proved ETTR in your initial post. Your own statement there is that noise is slightly less in the 2nd shot where ETTR was employed, compared to the 1st shot where it wasn't. That's the whole point and purpose of ETTR.

Your shot #3 and #4 prove nothing other than that lower ISOs will always produce lower noise than high ISOs. If you have the time and means to use a lower ISO (such as a tripod and stationary subject that will wait while you set it up), that will give always give better results than a higher ISO.

And, with today's cameras, ISO 6400 that you used in your first two shots really isn't high enough to demonstrate a very significant difference.

Try using ISO 25600. Under-expose one shot and slightly over-expose (+1/3 or 2/3 stop) another.... Then "push" the under-exposed image in post processing and "pull" the over-exposed. Best to do the comparison with RAW files (so that in-camera noise reduction and sharpening are both minimized), then apply noise reduction equally to both images in post-processing. Yours may be different, but with most cameras the ETTR image will show less noise.

Any photographer who finds themselves often needing to increase the exposure in their images during post-processing should give ETTR a try to see if that helps. Someone who rarely needs to push their exposure or even tends to over-expose a lot shouldn't use it (frequent over-exposers might even try dialing in some minus Exposure Compensation... "ETTL", I suppose). Usually ETTR is only done with slight "tweaks"... +1/3 or +2/3. But some users find they need more (and some cameras are set to work in 1/2 stop increments, instead of thirds).

ETTR isn't anything new. Back in the days of film cameras a lot of us "calibrated" our own cameras, meters and even film stock using tests. For example, when shooting Fuji Velvia 50 slide film with my EOS-3 cameras, I set them to ISO 50... but with some Konica and Nikon that I used prior I would use ISO 40. Using ETTR is essentially the same thing. It is simply something folks can try, if they tend to under-expose a lot.

It will always be better when there is time to get a more exact exposure. ETTR is for fast shooting situations and only applies to auto exposure modes. And, Exposure Compensation should still be used, in addition to or "on top of" ETTR. ETTR is a "base" adjustment, while E.C. is additional fine tuning, as required by subject tonality or other factors.

So, LIGHTEN UP! (After all, that's the whole point of ETTR, to lighten up your images! )

selmslie Loc: Fernandina Beach, FL, USA

My entire point is that, unless there are demonstrable benefits to ETTR, exposing normally at a low ISO and watching out for blinkies is all you really need to do.

ETTR/EBTR is itself an unnecessary massive over-complication of the exposure process. Nobody has been able to demonstrate that an image captured using ETTR is any better than one captured more easily at a lower ISO.

selmslie Loc: Fernandina Beach, FL, USA

Proponents of ETTR do this all of the time, use a ridiculously high ISO and then slightly overexpose. You get the exact same benefit simply by using a slightly lower ISO.

All that proves is that additional exposure reduces the appearance of noise. We all understand that.

Getting back to reality, if you are using a low ISO and can't see any noise, what value does ETTR add?