Highlight Metering mode on Nikon D750

Discussion in 'Nikon' started by sue_block|1, Sep 6, 2015.

  1. Today I read about the joy of using the camera Highlight metering mode on the Nikon D750
    However, no matter how much I searched I am unable to find out the steps to set this mode on the camera..
    Be grateful for information from a Nikon D750 user.
    thank you
    Sue
     
  2. ShunCheung

    ShunCheung Administrator

    The highlight metering mode is a special type of spot metering: http://www.nikonusa.com/en/Learn-And-Explore/Article/i4pecefz/using-the-highlight-weighted-metering-mode.html
    On most Nikon DSLRs, if you hold down the metering mode selection button and then rotate the main command dial, you can choose among matrix, center weighted and spot metering modes. The new highlight metering mode is a fourth choice. Its icon is a dot in the center, similar to the spot mode, but it has an additional asterisk outside of the rectangle symbol.
     
  3. The highlight metering mode is a special type of spot metering:​
    As much as I understand it is actually a variant of the matrix meter. But unlike the usual MM it finds the brightest highlights and exposes so that they're not blown (in wide-DR scenes, the MM goes for the midtones, with some bias for what's under the active AF points, mostly resulting in brighter images and blown highlights).
    I haven't tried it myself, but a number of exposure-sensitive people say that it doesn't do what they'd like it to do most, that is automatic ETTR for raw images - that the resulting raw files are often exposed lower than to the right edge of the raw histogram (RawDigger etc). I guess it is probably made for use with JPGs.
     
  4. "with some bias for what's under the active AF points. . ."
    I've played around with it on D810 and I agree that Sem's observation is an important one. Once you get the hang of this "bias" HHM can be handy in some situations.
     
  5. When you have it on the metering mode looks like spot metering with a little asterisk next to it. I have only used it a few times but did not find it particularly useful. Seems to be only useful for stage photography which is something I do not do.
     
  6. This was new to me. Looks like it's useful when shooting in primarily dark scenes. For instance portraits when it's dark outside, people in nightclubs, performers in stage lighting, during christmas, halloween etc. I imagine it could be useful in some sports scenarios as well.
    IMHO, there are many scenes that matrix can't handle without a serious amount of exposure correction. When you need to manually correct something it defeats the entire purpose of having something done automatically. So I imagine it would be a good tool to have for those that shoot with automatic exposure modes like program, aperture priority or shutter priority.
    PS. Actually I think it might be useful in scenarios like portraits in full sun where some parts are exposed by direct sunlight. Usually those parts are blown out because the camera exposes for the main parts of the subject. Weddings and white dresses comes to mind. With this highlight metering mode the subject would be under exposed but the parts not blown out. This is perfect because it's easy to raise the exposure in raw and at low ISO there is basically no penalty for doing it.
    So instead of trying to recover blown highlights in the raw converter, you would shoot in the highlight mode and have the highlight exposed properly and then raise the exposure in raw. Come to think of it, it should be useful in some other scenarios as well, like sunsets.
     
  7. I wonder if this Nikon's answer to Canon's highlight tone priority mode.
     
  8. ".....that the resulting raw files are often exposed lower than to the right edge of the raw histogram (RawDigger etc)."​
    Sounds good to me. It would be foolish to use up all the RAW headroom by effectively overexposing by 1 - 1.5 stops. Not many users need a 12 stop dynamic range for their everyday shots, but they do need a metering mode that doesn't blow highlights irrevocably.
    So where's my ETTR?
    And now Nikon, all you need to do is retrofit it your other pro and semi-pro cameras as a firmware update. We can probably live without the stupid little asterisk added to the top LCD display. Just lose the central black dot out of the MM indication, or lose everything except the frame round the metering icon. [_] That'd do for me. Seems crazy to offer a feature like that on the D810 and D750 while the likes of the D4 and even the more recently-introduced D7200 go without. Don't D4 and DX owners deserve a properly useful metering mode too? (At long last!)
     
  9. Rodeo Joe,
    There is actually not a lot of real headroom in the raw files.
    Using the saturation based ISO (ISO 12232) full sensor saturation is at 141% reflectance. Comparing to a gray card with 18% reflectance we are clipping highlights 2.97 stops over 18% gray.
    If we say that we are using a matte white paper as the brightest white we want to capture and it has a reflectance of 90% then raw headroom is 0.65 stop. More commonly 98% reflection is considered max and in that case we have 0.52 stop headroom.
    Newer cameras don't use saturation based ISO as their ISO settings but rather recommended exposure index which could be anything that looks good. Saturation based ISO is the only ISO definition that is based on the sensitivity of the sensor. As luck have it dxomark measures saturation based ISO on all tested cameras.
    If we look at a D750 the real saturation based ISO is 73 when the camera is set to ISO 100. That means that ISO 100 is less sensitive than a real ISO 100 would be. The difference is 0.45 stop.
    If we look at our headroom example above we have now an additional 0.45 stop of headroom. So a total of 0.45+0.52 = 0.97 stops of headroom. So one stop headroom for all practical purposes.
    What confuses a lot of photographer and making them believe they have lots of exposure headroom in the raw file is the LCD preview on the back of the camera. After shooting raw, the camera converts it into a jpeg file using the settings in the camera. That is used for previews on the back of the camera as well as histograms.
    The default settings produces a punchy and high contrast jpeg image and you get blinking highlights very easily. That is despite the fact that the raw file is actually not anywhere near being overexposed. Also the standard brightness setting of the LCD on the back of camera is too bright and makes images look overexposed.
    So when a photographer has shot an image with blinking highlights he notices after the fact that he can rescue the image in the raw converter. He mistakenly believes that there is a lot of headroom in the raw file when in fact he was fooled by the jpeg settings in the camera. The solution is to set the camera to a neutral low contrast jpeg and lower the LCD brightness. That way the LCD preview and histogram more truly reflects the actual raw file.
    I usually shoot manual and use the spot meter. Placing the brightest white at +3 stops will be very close to the maximum exposure possible without clipping the raw file.
    Most serious proponents of ETTR today just recommends using the blinking highlights as a guide and not go crazy. The advantages of perhaps being able to squeak in another 1/3 or possibly 2/3rds of a stop exposure brings very little additional image quality with todays cameras.
     
  10. 141% / 18% = 7.8 = 2.97 stops as you rightly say. 100% / 18% = 5.55555 or 2.47 stops. That's a half stop extra headroom over a JPEG....... in theory. However, my own experience is that you can drag down "blown" highlights, according to the camera's histogram, by at least one stop from a RAW file. And it's that practical >1 stop extra headroom that I was talking about. In short I don't believe that figure of 141%, and I'm not sure where that comes comes from (reference source please?).
    Anyhow whether it's 0.5 stops or 1.5 stops. It's better to have something in hand, since even an automated ETTR system isn't going to be 100% successful. Some headroom is better than none. It doesn't affect my original proposition that a move in the direction of a true ETTR metering system is better than Nikon's current flakey matrix metering.
    PS. I've used a handheld light meter and taken 18% incident readings using the camera's ISO setting on the meter. I can confirm that clipping on the camera histogram takes place with a reflectance close to 100% (e.g. white paper, whitewash, white clouds, etc.). That's a full 2.5 stops up from 18% reflectance. And I can also confirm that there's more than 0.5 stops headroom on the RAW file over the top of that. So practise doesn't match with theory here I'm afraid. BTW, white copier paper meters at close to a 100% Lambertian reflectance. 90% reflectance would look far from "white". And to top that, Nikon's applied JPEG gamma isn't anywhere near to the sRGB standard of 2.2. That's easily shown by photographing an accurate density wedge and measuring the rendering from the camera - it's not even close!
     
  11. Joe,
    The 141% comes from the ISO standard. You can derive it from the formulas.
    You can find it on imatest: http://www.imatest.com/docs/sensitivity_ei/
    You can find it mentioned on wikipedia: https://en.wikipedia.org/wiki/Film_speed#Saturation-based_speed
    It's also on dxomark pages somewhere.
    You can't look at what you can drag down in the raw converter. The reason being that most raw converters will fabricate data from two channel when one channel is clipped. If you want to look at it for real you have to look at the real raw values in the raw file. You need rawdigger or something similar for that.
    I agree on you with a true ETTR metering system. What might be difficult is that there is always some clipping going on if you have true specular reflections in the image. Which you have in a lot of images - edge of a glass, metal etc.
    I think you are misunderstanding me. I agree that there is more than 1 stop of headroom - if you use the histogram on the camera. But that's because of the contrast and tone curve from the jpeg rendition. However if you look into the actual raw file (using a real tool, not lightroom or such) you'll see that there is no clipping of the highlights yet. So what I'm saying is that there is at best 1 stop of raw headroom on newer cameras.
    I've done plenty of real world tests in the past and also recalibrated my light meters to set them up to get the max exposure without clipping in the raw file (not jpeg preview). I've also tested out different tone curves (can be downloaded to the camera) as well as tried out every possible permutation of settings to get the most accurate jpeg preview and blinking highlights.
    As I'm sure you know, but don't know if you accounted for, is also the variance of lens transmission, the light meter variations, camera body variations etc. It's extremely time consuming to make these kinds of test and try to eliminate as many variables as possible. For example in your test did you actually determine if it was your lightmeter that was off one stop or something else. In the end just finding out a method that gets you the result you want is fine. Eventually one has to decide if he's going to be an image scientist or just visual artists taking pictures.
    Regarding white paper I don't know. I just used the brightness/whiteness numbers of a few Xerox and HP printer and copier paper. They were around 90. I know that paper with optical brighteners are brighter, like a lot of inkjet paper, but that was not what I had in mind. In either case the difference between 90% and a 100% is small in stops.
    I haven't checked the gamma but you have a tone curve as well. Are you sure you are not confusing the two? On my older cameras the histogram was much larger and I had a piece of white tape and marked every stop on the histogram with mid gray as starting point. It looks very different depending on what settings you have selected in your camera. I'll see if I can find a shot of it just for reference.
    00dTmF-558358084.jpg
     
  12. thanks everyone for your input - much appreciated. I shoot raw and I will use this option when I am taking photos in high glare conditions.
    Thought I would provide a tip for some who maybe unaware of this option of easily accessing the ISO button which is in awkward position on the D750. The method I saw on a tube" Nikon D750 "Quick Tip - Button Assignment". This method is assigning the iso to the record button. Menu f9 - record button set to iso sensitivity.
    Sue
     
  13. Joe, I don't know what cameras you have but regarding the tone curves have a look at these for the D800:
    http://www.dpreview.com/reviews/nikon-d800-d800e/19
     
  14. Joe,
    Sounds good to me. It would be foolish to use up all the RAW headroom by effectively overexposing by 1 - 1.5 stops. Not many users need a 12 stop dynamic range for their everyday shots, but they do need a metering mode that doesn't blow highlights irrevocably.​
    I think it is fair to have a HWM mode for JPG shooters, but it would be even better to have an additional one for raw shooters that wouldn't waste highlights headroom.
    Pete,
    So one stop headroom for all practical purposes. ...
    I'd agree that one stop is a likely result. But it may be different with some PCs like Vivid and in extreme WB conditions, with ADL...
    I think an auto-raw-ETTR option would be welcome for many, because I think this is something useful and relatively straightforward for automatic implementation. I'm not saying that it is trivial, because it appears that it would require a certain degree of reengineering of Nikon's metering implementation.
    I agree that small specular highlights are a problem here, because they make the right edge of the histogram ill-defined. With the OVF, the number of metering points is restricted, so one might still get blown pixels. This would appear not to be a problem in Live View where metering is done using the main sensor; however I assume only a subset of pixels or a half-processed image is used for metering for the sake of speed. But this should be considered by reasonable thresholding. (Further you get questions what to do with the sun and stage-lights in the frame...)
    Of course, the spot meter is one alternative, but not a fast one.
    I wonder if this Nikon's answer to Canon's highlight tone priority mode.​
    AFAIK HTP is a variant of ADL, DRO, ALO, etc, which underexposes the raw data by a small amount (compared to the default midtones-centric meter) like all the rest, then applies instant processing which emphasises the highlights range rather than lifts the shadows like most of the rest. I think this is mainly a rendering decision. In "harsh light", the scene DR is essentially wider than what the output DR (prints and current displays) can show. Therefore, tonemapping which involves substantial DR compression is used. Typically one can make both highlight and shadow details better visible at once, but contrast is killed along the way... there are compromises involved, and the computer doesn't know what is deemed important in the image. A capable raw converter (and PS further on) has several sliders avalilable to bring out the desired result which may be very different. ADL only has one manual "intensity" adjustment (there is more done under the hood). HTP offers a different approach which doesn't compress the highlights as much.
    It is not clear to me why all these instant DR "enhancements" underexpose, except ADL used with the HWM recently, tend to underexpose raw data - in the sense that when the resulting JPG histogram looks right, the raw histogram tends to show two or more stops unused highlights headroom. Meaning that a substantial amount of the sensor DR gets wasted, and with earlier Nikons there was an obvious penalty in the noisy shadows. I suspect the underexposure was made to make room in the highlights for the instant processing - notice that the fill-light ("shadows protection" in NX) operation causes just about as much lift in the highlights as ADL underexposes the raw data. The unlogical bit is that the in-camera raw conversion in the D90 already has an exposure compensation parameter which should do the trick without a noise penalty. Unless there are other processing restrictions...
     
  15. Of course, the spot meter is one alternative, but not a fast one.​
    Sem, if it's fast or not depends only on the skill of the photographer. If you only shoot manual and use the spot meter all the time you become pretty fast eventually. And you only need to change the exposure when the light changes. The automatic modes are prone to change the exposure every time the subject reflectance changes. Of course, if you're just learning and need to think which way to spin the dials or which way the exposure meter goes for brighter or darker then using the spot meter becomes a very slow process.
    It's interesting what you write about the in-camera processing restrictions. As you say there might be restrictions based on pure processing power requirement or even memory requirements. It's of course impossible for us to know and it might explain some odd behavior.
     
  16. Sem, if it's fast or not depends only on the skill of the photographer. If you only shoot manual and use the spot meter all the time you become pretty fast eventually. And you only need to change the exposure when the light changes. The automatic modes are prone to change the exposure every time the subject reflectance changes. Of course, if you're just learning and need to think which way to spin the dials or which way the exposure meter goes for brighter or darker then using the spot meter becomes a very slow process.​
    If one is shooting landscapes, spot metering is surely viable. But for certain sports or when chasing kids, it makes no sense, with the microprocessors in current cameras. I mean, I don't think my exposure routine is something particularly creative or something that couldn't be handled by a computer; it is just that the current meters don't do what I want.
    It's interesting what you write about the in-camera processing restrictions. As you say there might be restrictions based on pure processing power requirement or even memory requirements. It's of course impossible for us to know and it might explain some odd behavior.​
    Well, yes, we see all sorts of computing restrictions in gadgets. Mainly because, due to economy and power consumption, the processors are barely capable enough to do what they are supposed to do. I guess some of the restrictions may as well be due to using limited precision (bit-length) in arithmetics, possibly fixed-point, which may be seen as a limitation of the dynamic range, etc.
     
  17. Nikon has a concise summary of this feature for the D810 (at http://nps.nikonimaging.com/technical_solutions/d810_tips/highlight/), which clarifies that the camera finds the highlights (so this mode is NOT like spot metering where you have to pick the spot), and that a single bright spot (like a stage light) in the frame can throw the algorithm off:
    "Bright light in frame:
    The camera may treat the light as a highlight, leaving the main subject underexposed. Compose the shot with the light out of frame or use matrix metering."
     

Share This Page

1111