Why no "expose to the right"?

[chris_seldon_lady]

<p>I recently read an article fom <em>Michael Reichmann:</em><br>

<a href="http://www.luminous-landscape.com/essays/leica-open-letter.shtml">An Open Letter to Leica</a><br>

He suggests new features and improvements for Leica M series. One of the proposed features caught my eye: the "expose to the right" metering mode. This is something I've been continuously lusting for, at least during the last year. It seems to me that, if your digital workflow is based on RAW, you can benefit from such a metering mode in 95% of your pictures. No more rough guessing about sensor reaction to the highlights and exposure compensation needed!<br>

Michael's article and following contribute from Mark Dubovoy let me hope that many photographers feel the same way.<br>

What do you thing about? Would you find this feature useful? Is this something we'll see in a future DSLR?</p>

[SCL]

<p>Personally I think each digital camera user should experiment with his/her gear based on their unique circumstances and workflow, to develop the appropriate settings for capturing scenes they typically shoot. I know I spent a lot of time going thru this exercise and today rarely have exposure issues. In fact, since I often shoot in manual mode and my DSLR may or may not meter with the lens of choice on it, I often can guess at the exposure as well or better than the camera can measure it. If it is really an unusuall situation I may check the histogram, but not too often. I guess I'm summarizing by saying that the mentioned feature isn't particularly worthwhile for me.</p>

[eric merrill]

<p>Seems to me like more trouble than its worth.</p>

<p>The camera would need to flip up the mirror and expose the sensor in order to determine where the white point clipping is. Hopefully, the highlights haven't been clipped, and the camera can calculate how many more stops of exposure the given scene can handle, and then make a second exposure.</p>

<p>That's if everything goes well. If the first guess clips the highlights, the camera would need to underexpose the next test and try again. Depending on the light sources in the camera, you might never be able to get an exposure with the specified white point clipping percent.</p>

<p>Then, when this actually works, imagine trying to edit these in your raw developer. Each exposure is probably different and needs a different amount of correction in order to be correctly exposed after being developed. That sounds like a royal pain.</p>

<p>While it's good in theory, I think there are many snags for this to actually work in practice.</p>

<p>In practice, if the light is constant, I use manual exposure and set my own exposure based on this idea.</p>

<p>If the light is constantly changing, I use aperture priority and an appropriate amount of compensation in order to accomplish the same.</p>

<p>I'm not sure a new exposure mode like this would be worth the effort needed to perfect it. I can't see Leica or any of the big manufacturers selling enough *additional* cameras to justify the R&D. I can't imagine many people refusing to buy a camera because it didn't have this mode.</p>

<p>Eric</p>

[bobcossar]

<p>I wouldn't find a need for it. Just one more useless thing to pay good money for...Robert</p>

[nsfbr]

<p>Eric, I think the way it could work is to end the capture on the first pixel that reaches saturation, or, more usefully, on a settable percentage of them reaching saturation. This would make every exposure an aperture priority, fixed ISO image, and have the unwelcome effect of making the camera's exposure indicators estimates only. </p>

<p>I could see practical limitations in terms of maximum shutter speed - the D70 had an electronic shutter that allowed 1/500s flash sync so I'd think doubling that would be a present day technical limitation. </p>

[robert lee]

<blockquote>

<p>I recently read an article fom <em>Michael Reichmann:</em> ... <a rel="nofollow" href="http://www.luminous-landscape.com/essays/leica-open-letter.shtml" target="_blank">An Open Letter to Leica</a> He suggests new features and improvements for Leica M series.</p>

</blockquote>

<p>Pure silliness. How does one improve upon the perfection that is an MP?</p>

[alec_myers]

<p>I think it would be a fine idea. </p>

[chris_seldon_lady]

<p>It seems that not many are interested in this feature. I don't thing it could be complex nor expensive to develop. Joel suggest a way to put in practise an "expose to the right" metering mode, but IMO it could be done simpler, exactly like any other metering mode: implementing an algorithm based on sensor response curve and TTL metering.<br>

IMO, how useful this mode could be depends on your shooting needs and applications needed. I can't check the histogram for most of my pictures, as Stephen says, since I can't take a second shot. Knowing your gear can help, of course, but it remains more a guessing game than a science. I think that a camera should be able to take care of this things. After all, there is a reason why most of us use TTL and AF!</p>

[bill_tuthill]

My prediction is that a camera company starting with P (Pentax or Panasonic) will offer ETTR before Leica. I find center-weighted (not spot!) metering completely useless, so camera manufacturers should just replace it with ETTR.

[steven_f1]

<blockquote>

<p>Eric, I think the way it could work is to end the capture on the first pixel that reaches saturation, or, more usefully, on a settable percentage of them reaching saturation. This would make every exposure an aperture priority, fixed ISO image, and have the unwelcome effect of making the camera's exposure indicators estimates only.</p>

</blockquote>

<p>Unfortunatly the camera has no way of knowing when a pixal reaches saturation. The pixal is not attached to any circuitry when the exposure is being made. The electronics in the camera cannot read the level of any pixal until the exposure is done. </p>

<p>I agree with eric. The mirror would have to flip up and down, possibly several times, to get it right.</p>

[alec_myers]

<blockquote>

<p>The mirror would have to flip up and down, possibly several times, to get it right.</p>

</blockquote>

<p>No, not at all. MR's original suggestion was for this to work in live view mode, when the mirror (if there is one) is already up and out of the way, and the sensor is already pulling in data, periodically, to update the display. Clearly each pixel is returning a value every so often, and it would be trivial to work out any exposure based on that.</p>

<p>Even outside of live-view mode, a single trial exposure (at the regular metered level) would give a sufficiently accurate basis to adjust for a second exposure (not so different from the way Canon's ETTL algorithms use a single flash to determine correct exposure for the second flash, although applied somewhat differently.) The mirror could stay up and the shutter could fire twice. Alternatively, if the sensor can be gated for an intial metering exposure then cleared down for the actual exposure even the shutter doesn't have to refire.</p>

<p>There's really no great difficulty about engineering this kind of exposure mode, if there was a will.</p>

[chris_seldon_lady]

<blockquote>

<p>My prediction is that a camera company starting with P (Pentax or Panasonic) will offer ETTR before Leica. I find center-weighted (not spot!) metering completely useless, so camera manufacturers should just replace it with ETTR.</p>

</blockquote>

<p>I agree, after all, Pentaxian TAv has been the biggest innovation in metering approaches in the last years. I agree about the center-weighted metering, too: it is an heritage of the first, older TTL systems and today sounds a little useless. At least, I never use it.<br>

Speaking of the implementation point of view, I read with much interest Alec, Eric and Steven suggestions. Unfortunately I'm more a photographer than a techie guy, and my contribution here is of no value.</p>

[janne_karila]

<blockquote>

<p>I don't thing it could be complex nor expensive to develop. Joel suggest a way to put in practise an "expose to the right" metering mode, but IMO it could be done simpler, exactly like any other metering mode: implementing an algorithm based on sensor response curve and TTL metering.</p>

</blockquote>

<p>The algorithm could be very very simple: just take the maximum reading from the matrix metering sensor and make that, say, 1 EV below clipping point.</p>

[j_stan_mason]

<p>Alex is correct. There is no mirror. Read MR's original piece. While you're at it, read the responses too. He is giving his advise to Leica on improving the M series. Rangefinders do not use mirrors, nor is MR proposing one. Agree or disagree, they are very thought provoking and give clues to how camera technology <em><strong>could</strong></em> change in the next few generations of camera.</p>