Exposing for highlights confusion

<p>The other day I wanted to practice exposing for highlights to capture enough detail to be visible but avoid blowing/clipping. I found a reasonably compliant subject, a white duck at the zoo. My understanding of this is thus: in spot meter mode a reading of a brilliantly white region should render these values at the infamous 18% gray. To obtain the level of white I want for these values I should adjust by + 1-2 stops (decreasing shutter speed in this case). What I found in reality was that I could not avoid blowing/clipping highlights unless I was shooting at with 0 or even minus exposure from what the meter was reading (shooting in manual mode). I'm not worried at this juncture about the dynamic range of the entire scene, I just want to place the whites where I want them. So, I guess I'm confused and just not understanding what I'm missing about this. Seems reasonably simple, and I've read in so many different places that this is the 'correct' procedure for exposing for highlights. I'm using a D80 by the way, did not have additional exposure compensation dialed in. My metering technique was to place the focus point over the area I wanted to hold detail in and then adjust shutter speed until the meter in the view finder indicated I was at +1 or +2 (it adjusts in 1/3 stop intervals, so 6 'clicks' of the command dial towards the plus side of the exposure scale). Ultimately I have to be an empiricist and go with what works, but it bugs me that I can't seem to replicate a procedure that seems well documented.</p><div></div>

<p>Christopher - I've not messed with this too much, but I believe to keep the highlights you should underexpose a certain amount. My D300 is set to adjust EV in 1/3 stop steps. I would try -1/3, then, -2/3, then -1 just to get a feel for things. I think the display would read -.3/-.7/-1 in this instance. This is a case where digital will give you instant feedback if you can run off 3 shots, then view on computer.</p>

<p>Hope I'm not confusing things more. Mark</p>

<p>And... were you spot metering, or matrix metering, etc?</p>

<p>yes, sorry I didn't make that more clear - spot metering. @ Mark - this is exactly the confusion that I'm having - what you suggest seems to fly in the face of the written material but seems to prove out empirically. </p>

<blockquote>

<p>but it bugs me that I can't seem to replicate a procedure that seems well documented.</p>

 

</blockquote>

<p>It appears to me you are making too many assumptions.</p>

<p>1) The image is white?<br>

2) You've achieved critical focus?<br>

3) You believe +1 or +2 exp comp will render a white object white?<br>

4) You believe your <strong>white</strong> balance is correct?<br>

5) You are spot metering from what luminance value of the target?</p>

 

<blockquote>

<p>I just want to place the whites where I want them</p>

 

</blockquote>

<p>Where would that be?</p>

 

<blockquote>

<p>I'm not worried at this juncture about the dynamic range of the entire scene</p>

 

</blockquote>

<p>Perhaps you should be concerned about the limited dynamic range for a certain area of the duck? The duck has no areas of luminance above a level of 235. (I've uploaded your image to illustrate)<br>

You are not even close to blown hi-lites; so your problem must be elsewhere; yes?</p>

<p>The +1 to +2 exp comp is something that has been thrown around to approximate the compensation (usually) required for fresh fallen snow in full high sun exposure.</p>

<p>I usually don't like to simply give answers to this type of question as I feel students understand concepts better when hints are given, forcing them to re-think the problem. The above assumptions I cited should help you solve your mystery. Hint: #2</p>

<p>..and to help a tad more.</p>

<p>A ducks feathers may indeed be white, but certainly not all over the breadth and width of the feathers.<br>

So; if you are hand holding and spot metering, you have little control at the moment of exposure to what your meter was actually seeing. You could have been looking at a level from 200 to 240. That alone will make quite a difference in camera white point settings.</p>

<p>Try this next time, use center weighted metering on what you believe to be the whitest portion of your subject. Now open up +1 to +2..be sure to get critical focus.</p><div></div>

<p>Perhaps this image will help a tad....</p>

<p>There was no way for me to capture the entire range of luminance in this image unless I shot it in a more controlled lighting environment...i.e (Studio)</p>

<p>Notice the sun/shadow positions and where I indicated blown hi-lites and not blown. Both areas are white. Hopefully you now understand the relationship of source illumination, position subject shape vs light reflection.</p><div></div>

<p>@ kevin:<br>

1) The image is white?<br>

I believe the area I was metering is white<br>

2) You've achieved critical focus?<br>

The image is soft so perhaps I can do better next time I try this to make sure focus is correct. The attached image is a 700x700 crop from the original<br>

3) You believe +1 or +2 exp comp will render a white object white?<br>

As you note about fresh fallen snow I have been taught that white subjects (perhaps the duck isn't as white as snow perhaps (another reference to #1 above?) will tend to render detail without being blown at +1 to +2 exposure comp from metered values<br>

4) You believe your <strong>white</strong> balance is correct?<br>

White balance is on Auto WB so this could have an impact - there was bright sun falling on the duck at the time but possibly one might consider that setting to be open shade too.<br>

5) You are spot metering from what luminance value of the target?<br>

since I don't know what the luminance values of the duck are, I'm not sure how to respond here - I'm metering on the whitest/brightest portion of the duck which is where you've marked it as a value of 235<br>

I think what you're suggesting is that what i am perceiving as white actually may not be as white as I think. Your point about the area that has a luminance value of 235 being the brightest part of the image is well taken. I guess what I'm questioning here is that that spot is actually where I was metering - so shouldn't that region have taken on lesser values? I'm thinking in terms of zone system concepts where this area, what I wanted to achieve, was to show at very bright levels.. perhaps my concept of what bright or high luminance is needs to shift. I'm not sure I intended to suggest that the whole of the duck should be brilliant white or the same luminance all over - thats not it at all. However, conventional 'wisdom' would suggest that i could meter the area you've indicated, add exposure to the tune of 1 or 2 stops and have a white that still holds detail (provided some of the assumptions you raise are met).<br>

With regard to your second post, the gull, I would agree that both parts you have indicated are 'white', that these reflect different quantities of light, and that taken as a whole the luminance range of the scene is too great to capture in one image. My question really is about how you meter and expose for that portion you identify as blown in that image</p>

<p>You are making your life way too complicated - the one place to look for and learn from is the histogram; it's free and right there at your fingertips. It tells you all there is to know about the exposure and which way you need to correct and by approximately how much. The one for your duck looks very nice and certainly couldn't take anymore positive exposure compensation than +1/3 stop (and that will push some areas into overexposure already).<br /> Initially, I also used spot metering for white birds - with quite mixed results. Now I am using matrix metering and have learned how to adjust the exposure to get the result I want. It depends on how large the white bird is in the frame and what brightness/darkness the background has. When in doubt and given the time, I check the histogram - or rely on the "blinking highlight warning" on the LCD.<br /> Your image has quite a green cast that takes a little effort to correct -I just went only as far as making white white:</p><div></div>

<p>I say this as a former D80 owner - I found could not to trust the D80's meter reading some of the time, especially when dealing with subject matter where there was a lot of light areas. I found the camera's meter often tended to overexpose.</p>

<p>Shooting RAW and ignoring the meter (meaning check your display for blown out highlights using whatever method you choose to insure good exposure) will yield the best results. Using programs like Photoshop, you can more easily bring out the detail of darker areas than you can retrieve blown out areas, especially with JPG files. RAW files give you more latitude in correcting both blown out areas and shadows but I still prefer to get the bright areas exposed properly and fix the darker areas during post processing.</p>

<p>Your shot looks exposed correctly so I am not sure what your issue is? You seem to have got it right.</p>

<p> </p>

<p>Couldn't give it a rest - this is the same image with a +1/3 EV exposure compensation - just at the limit were none of the pixels is blow (i.e. 255). Depending on your monitor and monitor settings, some areas might already look blown - the histogram verifies that they are not.<br>

If you really want to make this into a science - than use a gray card and a white sheet of paper under uniform lighting conditions and see what results you get as a function of exposure mode and correction. You would have a very simple histogram - ideally just a single peak that you can move around with different exposure settings to your hearts content.</p><div></div>

<p>Christopher - your original idea to measure a white spot and then expose by setting exposure to obtain an almost white is quite correct. There may be some fine tuning needed but the logic is simple and there is no problem in the logic here.</p>

<p>The problem is in the raw image conversion, assuming that you shoot raw,the only way to attack this problem properly (because the in camera conversion to jpg images is a black box and may be misleading).</p>

<p>Depending on your raw converter you can get misleading results.</p>

<p>You need to look at the "original" data in the RGB values of the pixels. There is software available to do this but this is complex and as a start it may be suffient to let us know how you convert your raw files.</p>

 

<p>The image I uploaded wasn't one with blown highlights - I knew this, and guess I didn't communicate my issue well. My confusion is about metering technique as I had believed that by accepting the meter's reading the whites around the tail feather region of this bird should have rendered an underexposure but as Dieter points out this area is very close to the point where the highlights will start to clip, much less than the +1 or 2 stops that I was expecting to need to achieve this result. The exposure for this area is what I would think of as being 'correct' - its just not what I was expecting. I understand and use the histogram and the highlight clipping warning to achieve decent exposure so its not that I don't understand or use those tools. One of the things I'm wondering about is if the spot meter is averaging more than just the area immediately under the focus point used and if that may be throwing this off a bit, behaving more like a center weighted meter for a very small area.<br>

Thanks all for your input - gives me more to think about!</p>

<blockquote>

<p>perhaps my concept of what bright or high luminance is needs to shift.</p>

 

</blockquote>

<p>..or perhaps only re-evaluated a tad.</p>

<p>If you were to circle with a black marker every area/zone that you believe should be pure white in the duck photo, I wonder how many areas you would have to draw? What about the transition zones from pure white (255) to something a little less than pure white, Maybe (235)</p>

<p>We can draw some basic truths from your question:</p>

<p>1) Photographers need to make decisions on color, brightness levels. (esp in high contrast subjects)<br>

2) We do not always have control over the light. (**Although this is a goal we all strive for**)<br>

3) Sometimes we must sacrifice one area to preserve another.<br>

4) Often the lighting variables are too numerous to come up with a formula to nail exposure.<br>

5) White point & black point are subjective based on the photographers desire in final output.</p>

<p>So; with regard to #4, I agree with one of the responders here, you are probably over complicating this..so take it easy, shoot a few frames, have a look and correct on the fly.</p>

<p>With time & practice you will find yourself thinking "Oh, that's a a duck, with my camera I need to open up 1/2..Oh, that's a deer on a snow covered field, I better use manual, spot meter the deer and open up about 1 stop.</p>

<p>These decisions are based on your metering technique as well as the way the camera meters; the latter requires intimate knowledge of your camera which requires lots of shots which requires time.</p>

<p>Darn, just when we thought photography was easy; after all, these new cameras can make anyone a photographer; right? ;)</p>

<p>Have a great day.</p>

<blockquote>

<p>I'm wondering about is if the spot meter is averaging more than just the area immediately under the focus point</p>

 

</blockquote>

<p>Quite possible.<br>

I do not own a D-80 so I do not know the angular size of it's spot.<br>

You can of course test your theory quite easily.</p>

<p>You used the word "Average." The whole idea behind spot metering is to eliminate "averaging"</p>

<p>I know my spot meter is approx 2% of the entire frame, so that's a pretty good spot.</p>

 

<p>Walter wrote:</p>

<blockquote>

<p>The problem is in the raw image conversion, assuming that you shoot raw,the only way to attack this problem properly (because the in camera conversion to jpg images is a black box and may be misleading).<br>

Depending on your raw converter you can get misleading results.</p>

</blockquote>

<p>This is a really good point. The fact that you are shooting raw doesn't mean that the image hasn't been tampered with by the time you view it on your computer screen. Is your raw processing software calculating settings (or deriving them from the EXIF data supplied by the camera) for a "correct" exposure and automatically applying them to the image? I don't know what software you're using, but please make sure that all exposure controls in that software are set to neutral values.</p>

<p>Well, maybe this will shed some light on the issue of 18% gray (or shall I say "myth"?): http://www.bythom.com/graycards.htm</p>

<p> </p>

<p>Dieter, I don't think that really means much in this context. Whatever exact standard is being used, it is still the case that meters are aiming for a medium gray value. Whether it's 18% or 12% or some ANSI-defined value that isn't publicly published doesn't really make much difference. It still should be the case that if you spot-meter a sunlit white object, you should need to add a stop or two of exposure to get it to come out bright white in the image. That it doesn't seem to be working that way in Christopher's case probably comes down to the angle of view of his camera's spot metering mode or something his raw processing software is doing automatically to "correct" his exposure.</p>

<p>Christopher,</p>

<p>From a pure theory aspect, you are absolutely correct. Everyone else is just confusing the issue. But in the real world, there are deviations from the theory, as you are seeing.</p>

<p>The most likely culprit is really simple: your meter probably read in more than just the area you wanted. In-camera meters are notorious for being slightly off and/or not having true 1degree spot meters. To really use the technique you are describing, you should first make sure what your meter is doing. How big is the spot? Is it truly in the center? You can test this by spot-metering a black/white checkerboard pattern and verifying exactly what parts your meter thinks are "white" versus "black", and whether it is too big to really get individual parts of the patches. Once you know this, you will be much more confident of what you are metering and you will be able to apply the exposure compensations more accurately.</p>

<p>Another possible culprit is the limitations of your sensor. I don't know the D80 very well, but some sensors have more exposure latitude than others. Again, the best way to figure this out is to test your camera's sensor and find out how far above (and below) middle grey it can actually go. There are procedures on the web you can use to figure out just how much you can get away with.</p>

<p>Finally, as others have pointed out, what you see on the screen as a blown highlight is not necessarily truly a blown highlight. Monitors make a difference. Also, there is an element of subjectivity to what is considered a "blown highlight" or not that varies between people, even if the RGB values do not indicate 255/255/255. And, if something is especially reflective (which a duck's feather can be if it is oily or wet), then it will be extremely difficult to capture details in the highlights the way that you want.</p>

<p>Chris</p>