Expose to the left?

[tonysmith]

I found the following in the "Photography Tips" section of the website of the International Centre of Photography in New York:

Shadows and Highlights

This method applies to digital cameras, when capturing in the RAW format, as much as it does to film cameras. When shooting negative film, color, or b&w, the rule of thumb is to shoot for the shadows and develop for the highlights. What this means is: you should set your exposure meter so that you will capture all of the shadow detail that you want and process/develop your film to retain the amount of highlight detail that you want. Another way to say this is to overexpose your film as much as you need, and under develop it as much as you need.

How this works with digital cameras differs from film cameras only in the way you will process/develop the image file. When converting the RAW format you have the opportunity to bring detail back into the highlights just like you can when processing film.

  —h. eugene foster, ICP instructor

Isn't this saying "expose to the left", contrary to the usual advice, or am I misunderstanding?

[DarkPenguin]

Sounds like expose to the right to me.

This being the key line ...

"Another way to say this is to overexpose your film as much as you need, and under develop it as much as you need."

So expose to the right and hope you have enough headroom in the raw file to recover your highlights.

[digitaldog]

Isn't this saying "expose to the left", contrary to the usual advice, or am I misunderstanding?

That's how I read it too so I'd agree with you, he's way off base. Film isn't linear capture like Raw, it has a H&D curve. They are not really remotely similar.

[PeterAit]

I found the following in the "Photography Tips" section of the website of the International Centre of Photography in New York:

Shadows and Highlights

This method applies to digital cameras, when capturing in the RAW format, as much as it does to film cameras. When shooting negative film, color, or b&w, the rule of thumb is to shoot for the shadows and develop for the highlights. What this means is: you should set your exposure meter so that you will capture all of the shadow detail that you want and process/develop your film to retain the amount of highlight detail that you want. Another way to say this is to overexpose your film as much as you need, and under develop it as much as you need.

How this works with digital cameras differs from film cameras only in the way you will process/develop the image file. When converting the RAW format you have the opportunity to bring detail back into the highlights just like you can when processing film.

  —h. eugene foster, ICP instructor

Isn't this saying "expose to the left", contrary to the usual advice, or am I misunderstanding?

No, it is expose to the right - meaning that the histogram of the image will have more peaks at the right end (higher exposure) than at the left end (lower exposure.

Peter

[digitaldog]

No, it is expose to the right - meaning that the histogram of the image will have more peaks at the right end (higher exposure) than at the left end (lower exposure.

Peter

IF that's what he means, there are at least two of us misunderstanding his writing.

Film: Expose for the shadows, develop for the highlights (we've heard this for eons).

Digital: Expose for the highlights (ETTR, don't blow them out), develop for the highlights (render so you have the detail that's there).

I don't see how he can say the are similar....

[JeffKohn]

I don't degree that it's ETTR. With ETTR you expose as much to the right as possible without clipping important highlights.  What they're suggesting is don't worry about the highlights when shooting and just expose for the shadows. Their assertion that you can bring back the highglights in RAW conversion just like you can when developing negative film is not really true. You can sometimes bring back partially blown highlights, but if all three channels are clipped there's nothing to bring back. Given that we generally tend to find blown highlights more distasteful than blocked up shadows when looking at an image, I would say their advice is not so good.

[MarkL]

This works with devloping b&w film where using the assuption that exposure determines the shadows and developement time controls the highlights works well.

With digital, if the scene brightness range exceeds what your camera is capable of recording you are totally screwed regardless how much you try and pull the exposure slider in your software.

[Ken Bennett]

The original tip from the ICP is consistent with "expose to the right." He's writing it in a way that makes sense to photographers working with b+w film.

While it is true that digital is much more sensitive to highlight clipping than film, the basic idea of ETTR still works. One just has to be more careful about highlight clipping when making the exposure.

[dwdallam]

Film: Expose for the shadows, develop for the highlights (we've heard this for eons).

Digital: Expose for the highlights (ETTR, don't blow them out), develop for the highlights (render so you have the detail that's there).

I don't see how he can say the are similar....

Well, they are similar really. When you expose film for the shadows, you are exposing to the right for digital; that is, making the highlights brighter. In both instances you would want the highlights as bright as possible w/o blowing them.

[digitaldog]

Well, they are similar really. When you expose film for the shadows, you are exposing to the right for digital; that is, making the highlights brighter. In both instances you would want the highlights as bright as possible w/o blowing them.

There's no question that indeed ETTR is about exposing to get highlights you wish to record be as bright as possible without blowing them out. You don't expose for shadows with digital in that ETTR is about placing the most possible tones in the last stop which has, in a linear capture, the fewest bits with ETTR.

You've got a tonal scale that the sensor may or (most likely can't) fully capture. You either expose to get the shadow detail or the highlight detail if the range is farther than the capture device can record. In either case, one hopes to put the most bits in that last (shadow) stop. You might find that you'll lose highlight data in order to get that shadow detail but you must still calculate (expose) for the highlights in order get the bits in those shadows you wish to record. That doesn't mean you disregard shadows because in this example, you want that shadow detail. But you have to expose to the right up to the point that you can obtain the shadow detail as going the other direction reduces the number of bits in the area you want to record. Maybe that's where this fellow is suggesting "expose for the shadows" but its damn confusing and doesn't help in aiding the photographer from understanding that they have to expose as far to the right as possible in order to obtain those bits in the last stop of the shadows. His intentions might be in the right place but the communication isn't very good, two of us read it the opposite way.

Where digital and film are the same is, a photographer has to understand the scene range and the range the sensor or film is capable of capturing. But where they differ largely is where you place the weight on exposure. In digital, since its a linear capture, you weight on the opposite end where the shadows reside since in linear capture, the least amount of data is contained there and any "under exposure" results in more noise and less real data. Again, film's got an H&D curve. How's that affect shadows in terms of exposure?

So exposure weighting if you will, is different in film versus digital. And with film or digital, a photographer must know the scene range they have, the range they can capture and then decided which end they wish to keep or lose if its too large.

[Eric Myrvaagnes]

His intentions might be in the right place but the communication isn't very good, two of us read it the opposite way.

Make that three of us (at least) that read it the opposite way.

In the Zone system (for film) one of course takes meter readings of all important areas, but only the shadow readings determine the exposure you should give. The high-zone readings are used only to determine the development (N, N+1, N-1, or something more extreme.)


"When converting the RAW format you have the opportunity to bring detail back into the highlights just like you can when processing film." This is patently false, if you "shoot for the shadows" as he suggests.

The rule of thumb, he says, is "shoot for the shadows and develop for the highlights. What this means is: you should set your exposure meter so that you will capture all of the shadow detail that you want and process/develop your film to retain the amount of highlight detail that you want."

And he concludes "How this works with digital cameras differs from film cameras only in the way you will process/develop the image file. When converting the RAW format you have the opportunity to bring detail back into the highlights just like you can when processing film."

My question to those of you that say this describes ETTR is: "What part of 'shoot for the shadows' don't you understand?"

[DarkPenguin]

Well it certainly isn't expose to the left.

When he says "When converting the RAW format you have the opportunity to bring detail back into the highlights just like you can when processing film." which direction do you think he's adjusting the exposure?

I do not think of ETTR in terms of highlights.  I think of it as giving me my shadows with less noise.  So when I shoot ETTR I 'shoot for the shadows' but then I'm dumb.

edit: removed a word that bugged me.

[PeterAit]

There's no question that indeed ETTR is about exposing to get highlights you wish to record be as bright as possible without blowing them out. You don't expose for shadows with digital in that ETTR is about placing the most possible tones in the last stop which has, in a linear capture, the fewest bits with ETTR.

You have it backwards - the last stop has the MOST bits, which is precisely the reason for ETTR. You want to place the entire tonal range of the image as high as possible on the sensor's response curve in order to preserve as much of the subtle tonal gradation as possible.

Peter

[digitaldog]

When he says "When converting the RAW format you have the opportunity to bring detail back into the highlights just like you can when processing film." which direction do you think he's adjusting the exposure?

Adjusting exposure in the Raw converter? Perhaps. But its useless if you don't actually implement ETTR, there is nothing to "bring back". When you shoot using ETTR, the "normal" or default rendering in most converters produces an image that appears "over exposed" so you adjust the exposure to get the nominal rendering. "Over expose" too much, that's not going to happen of course. I don't see how this has anything to do with exposing for shadows or altering the rendering settings for such. Again, its possible he's talking about ETTR but its not clear and I don't think its a good idea to lump non linear film with linear Raw as being similar.

I do not think of ETTR in terms of highlights.  I think of it as giving me my shadows with less noise.  So when I shoot ETTR I 'shoot for the shadows' but then I'm clearly dumb.

That's a valid way to think of it. I do too but think of ETTR as exposing as close to clipping but not clipping highlight detail I wish to render. IOW, if I have a 8 stop scene and my camera can capture 7 stops, AND I want all the detail in the shadows, I have to forgo capturing that last stop of highlights via exposure settings. But I would use ETTR to get as much highlight exposure in the 7th stop in order to produce the most data in that first stop of shadows.

I see two issues here. One is targeting exposure for capturing the scene range assuming its greater than the sensor can record. 2nd is now targeting the exposure so that I implement ETTR based on this range.

[digitaldog]

You have it backwards - the last stop has the MOST bits, which is precisely the reason for ETTR. You want to place the entire tonal range of the image as high as possible on the sensor's response curve in order to preserve as much of the subtle tonal gradation as possible.

Unclear semantics on my part. Last stop being shadows. When you expose to the right, you place as much data there as possible, the results being the most data one can record in the last stop (shadows) or if you prefer, first stop (shadows):
http://www.digitalphotopro.com/technique/c...ng-for-raw.html

[Schewe]

You have it backwards - the last stop has the MOST bits, which is precisely the reason for ETTR.

Actually, ETTR really has nothing to do with "bits" and everything to do with more light....the reason ETTR works is NOT because of linear capture and data-dense packed highlights but because you are simply capturing more photons when you increase the exposure. More light = more photon...more photons = less noise. Simple as that...

[John Camp]

I think everybody is explaining the same technique, but looking at it from different ends. Some people are taking great care to get exactly the amount of exposure that will get the brightest highlights without clipping, which automatically gets you the most shadow detail (without clipping the highlights.) Other people are arguing that you should expose for shadow detail, but would also presumably not wish to clip highlights at the other end. Same thing, basically, badly phrased, perhaps. This is also assuming that nobody would want to completely blow the highlights in either film or digital, which can be done in either medium.

It's also possible to ETTL, and let the highlights go. This works in dark churches, where the blown highlights may give you a "spiritual-type" glow, while still letting you look at some of the details in the statuary, frescos, et. But you should know when you're going to let the highlights be blown...

I'm interested in the comment about the linearity of digital sensors, because it brings up the question of why somebody doesn't design a firmware tone curve for *exposure* that simulates a film curve (bring back my Kodachrome!) Shoulders are pretty damn useful sometimes.

Somebody's going to tell me it's been done, aren't they?

JC

[digitaldog]

I'm interested in the comment about the linearity of digital sensors, because it brings up the question of why somebody doesn't design a firmware tone curve for *exposure* that simulates a film curve (bring back my Kodachrome!) Shoulders are pretty damn useful sometimes.

Somebody's going to tell me it's been done, aren't they?

When the image is shown in your Raw converter (unless you've asked for linear which may or may not be available), the curve results from the current rendering based on the current color space. IOW, you apply that tone curve when the image is output referred.

[JeffKohn]

I do not think of ETTR in terms of highlights. I think of it as giving me my shadows with less noise. So when I shoot ETTR I 'shoot for the shadows' but then I'm dumb

I think what Eric and I are saying is that ETTR is about placing your highlights as far right as possible without clipping highlights (highlights that you care about, anyway). While this does give you cleaner shadows, it is not exposing for the shadows. Placing the highlights is what determines your exposure, and the shadows fall whereever they fall. WIth digital cameras there's more latitude to bring up the shadows than there is to bring back clipped highlights, so digital exposure shares more in common with slide film than negative film.

Taken literally, the authors recommendation is to use a meter reading from the shadows to determine your exposure, without any regard to highlights at all. That is not exposing to the right;  it is exposing for the shadows. The two are not the same.

[ErikKaffehr]

Hi,

To my understanding ETTR is about keeping noise down by keeping Signal/Noise Ration (SNR) as high as possible. As far I understand you need something like 100 photons captured to have a reasonable "photonic" noise. With to few photons we get som noise from statistics. So we want each sensor cell as filled as possible to keep sensor noise down. Lets assume that each sensor pixel can hold 50000 electron and that each photon yields one electron. So with the sensor cell near saturation we would have something like an SNR of 500 (50000/100). If we would be one stop below saturation Signal/Noise woul be 250, (50000/2/100) and so on.

So what we want to achieve is essentially to have the sensor nearly saturated in the highlights to get little noise in the shadows.

A couple of observations:

- In the silver halide days the expose for the shadows rule did only apply to negative film. Slide film should always be expsoed for the highlights, but with an eye always kept on mid tones.

- ETTR says essentially, don't care about middle tones or shadows. Lets's expose for minimum noise and maximum density range and fix the shadows and mid tones after the fact.

Now, if we happen to have a scene with a very wide dynamic range, like sunlighted clouds over a canyon in shade, we made need to expose for the clouds. That may cause mid tones to to dark which we need to fix in "development". The image will be darker than one exposed for correct mid tones, but it is the only way to keep good clouds, at least if we don't try with HDR/EDR.

Best regards
Erik

I think what Eric and I are saying is that ETTR is about placing your highlights as far right as possible without clipping highlights (highlights that you care about, anyway). While this does give you cleaner shadows, it is not exposing for the shadows. Placing the highlights is what determines your exposure, and the shadows fall whereever they fall. WIth digital cameras there's more latitude to bring up the shadows than there is to bring back clipped highlights, so digital exposure shares more in common with slide film than negative film.

Taken literally, the authors recommendation is to use a meter reading from the shadows to determine your exposure, without any regard to highlights at all. That is not exposing to the right;  it is exposing for the shadows. The two are not the same.