Metering in digital cameras and highlights .

I have read quite a few articles in the internet on camera metering , especially in digital cameras and all say that digital cameras are prone to clip highlights more often the analog slr's . That plus the fact that digital cameras have less data in the dark parts of a photograph making the inherent noise in digital photograph more obvious in the dark parts then in the highlights where there is more information which tends to mask the noise has made me think a bit on the subject . Now , here comes my question guys to see what you might think about this .

Why don't camera manufacturers design a metering system that takes into account digital photographs inherent problems , why not design a metering system that might monitors highlights and by doing so keep the highlights in check so that they don't clip ? I believe everything can be done in the digital realm and would not be all that dificult to implement. With a system like this it would be fairly easy to use the cameras full dynamic range without having to be worried about clipping the highlights.

I think the problem would be, do you always want the highlights to avoid clipping? If you want to allow some clipping, how much and in which color channel? It can become a confusing mess. Something that is probably best left to the human brain. After all, we have to have some rights to passage don't we? ;-)
--
As far as possible, without surrender,
be on good terms with all persons.
-- Max Ehrmann

RobertLaw said:

Something that is probably best left to the human brain. After all, we have to have some rights to passage don't we? ;-)

Click to expand...

Robert is exactly right. The camera is just a dumb computer. It can see some light, and it can measure it accurately, but it has absolutely zero human recognition what it means. It does not know if subject is a black cat in a coal mine, and ought to be dark, or if it is a polar bear on the snow, and ought to be bright. The human brain does immediately know, because we are smarter, we have remembered experiences, and we can look around and see the entire situation, and think it out. The stupid camera only sees what aim it at, and it cannot tell Aunt Minnie from a bookend. It has absolutely no clue what it is, or how it ought to be.

The camera computer "could" be designed to place the brightest part right at the 255 end, but often it should not be that bright, or sometimes it really should go off the bright end. Stupid camera has absolutely no clue what it is, or how it ought to be.

So what the camera does (all it can do) is to try to make every scene's average value come out about mid point, and it hopes for the averages to hold (hopes it is a typical average scene). We human photographers need to learn to watch it, and compensate it, when required. We can easily learn to immediately recognize the scene, and to know what the camera is going to do. We can also look at results, and try the second chance at it.

See http://www.scantips.com/lights/metering.html about how camera meters work.

Put the camera in manual mode, point it at the brightest thing you want to keep detail in. If the meter reads over +2 then you are clipping highlights and need to reduce aperture size, increase shutter speed or lower ISO. In A or S mode, frame our shot, hit exposure lock, and point the camera at the brightest thing you want to keep detail in and and see if the meter reads over +2.

You can also check the color channels in the histogram after taking your shot to see if anything in a channel falls off the right side.

Hi,

You already got some very good answers to your question.

Just want to add that Nikon provides ADL for the purpose to avoid clipping highlights and opening up shadows which is necessary when exposure is lowered to avoid clipping.

I dont use ADL when using flash however and when not using flash i would only enable ADL in high contrast situations ( with D5100 you can do ADL bracketing ).

When shooting RAW i prefer to disable ADL ( the matrix metering in the D5100 is realy good BTW, much better then my D40 ). I look at the colour histogram and "blinkies view" to decide if i need to retake the shot whith negative EC.

If you use matrix metering you can also use following technique if you wish :

from an old reply in following thread : http://forums.dpreview.com/forums/readflat.asp?forum=1034&message=38309863

Just another possibility....

A lot has already been written about the nikon MM ... it puts a lot of emphasis on what is below the active focuspoint and averages two meterings.

KR also mentions this behaviour in his D70 review

see http://www.kenrockwell.com/nikon/d70perf.htm#performance

QUOTE:

OBSERVATION: The meter has an odd and clever bit of programming. In normal operation (not setting AE lock to ON for the shutter button) the AE is still half locked when you press the shutter halfway! What does this mean? It means if you lock the focus by holding the shutter button halfway that the exposure only moves half as many stops as it has to when you recompose. You'll only notice this if you recompose to something many stops different. If you do you'll see the meter change its reading as you recompose, but now if you take your finger off the button you'll see the reading jump the rest of the way to the exposure at your final composition. This is also what you might call "sticky" exposure: it sort of partially sticks to where you locked focus, even when you have not intended to lock exposure

UNQUOTE

In case you are not aware of this : one can use matrix metering in a way not mentioned in the Nikon manual anywhere !!!

When using MM on my D40 is dial in -0.3 exp comp as a start and in high contrast situations i use following technique ( i shoot raw ).

One can make an average of 2 meterings , a metering on the place where you lock focus and a second metering on the brightest part ( e.g. in the sky ) , this way reducing the risk for a blown out / overexposing ( e.g. the sky ). But you will eventually have to open up shadows in camera or in PP : D-lighting or equivalent method.

QUOTE ( with thanks to Dayd3 ) :

Dayd3 wrote:
I explored a little bit more Matrix metering on D80 and found some
tip on the web how to use this metering in a way which is NOT
MENTIONED ANYWHERE IN NIKONS USER-MANUALS.

We all agree that Matrix Metering try to protect shadows which very
often finishes in blown out highlights if under our focus point is
something darker than middle gray. While sometimes we want such a
behaviour many times we don't. We also know that by default metering
doesn't lock when shutter is half pressed AE LOCK - off (CSM 19).

But what I didn'rt realize before is that MM also takes two metering
and averages them. For example if we focus on someone's face, half
press shutter and than recompose, instead of simply taking a new
reading camera will average the original exposure and the new one.

If we combine this behaviour with a AE-L/AF-L set to AE Lock Hold or
AE Lock Only (CSM 19) we can actually focus where we like by half
pressing shutter button and holding it (camera focuses and takes
first reading) than we point camera for example to the sky and press
AE-L/AF-L button (or press and hold what depends on our settings CSM
18). Than we recompose again and take picture. In this way MM can
average readings from the darkest and lightest subject and prevent
blown out skies.

This is working only in Matrix metering mode and it seems doesn't
work in AF-C mode.

Link to web page where I found this tip:
http://www.nikoncafe.com/vforums/showthread.php?t=95098

UNQUOTE

Of course on the D90 and more recent bodies you can also use ADL , which will also reduce the risk of blown highlights.

Matrix metering + ADL is IMO Nikon's answer to the "overexposure" issue with D80, D40 matrix metering ...

But i have to say the D5100 matrix metering is A LOT better then on the D40 and does not overexpose as much as my D40.

Greetings,
Marc
--
ps : some interesting tutorial websites :

http://www.ronbigelow.com/articles/articles.htm
http://www.cambridgeincolour.com/tutorials.htm
http://www.luminous-landscape.com/tutorials/
http://www.normankoren.com/sitemap.html



my photos (examples, see dpreview galleries):

I have a good feeling that Nikon is already doing this in their meter logic, this is why they consistenly get a comment such as "tends to overexpose at default settings" in their reviews.

Adjusting the meter to capture the maximum overall frame DR may not always give you the desired DR in the subject areas you want. I think the situation would be greatly improved as the sensors go to lower SNR levels that can go towards less noise in the shadow areas so that you are not exposing to the right and loosing highlights

Wouldn't it just be easier to spot meter off a mid-tone, set your exposure and then check the highlights you want to keep detail in.

I think the point is that the "mid" tone is relative to what you want to have as a subject. The ask here is the way we have auto area auto focus and focus tracking there should be be some technological means way that the camera uses something like spot meter, however it somehow decides where your spot is for different shots based on some logic (kind of like subject tracking)

I am not a fan of any auto function that is not instataneously predictible 100% of the time, it means that I am taking a chance with my picture. ie Auto ISO, Auto Scene mode, Auto Area AF, Subject tracking AF. etc... none of them work 100% of the time.

gwlaw99 said:

In A or S mode, frame our shot, hit exposure lock, and point the camera at the brightest thing you want to keep detail in and and see if the meter reads over +2.

Click to expand...

Have you actually tried that?

.

Mannypr said:

I have read quite a few articles in the internet on camera metering , especially in digital cameras and all say that digital cameras are prone to clip highlights more often the analog slr's . That plus the fact that digital cameras have less data in the dark parts of a photograph making the inherent noise in digital photograph more obvious in the dark parts then in the highlights where there is more information which tends to mask the noise has made me think a bit on the subject . Now , here comes my question guys to see what you might think about this .

Why don't camera manufacturers design a metering system that takes into account digital photographs inherent problems , why not design a metering system that might monitors highlights and by doing so keep the highlights in check so that they don't clip ? I believe everything can be done in the digital realm and would not be all that dificult to implement. With a system like this it would be fairly easy to use the cameras full dynamic range without having to be worried about clipping the highlights.

Click to expand...

They do, as best they can, but the real world exceeds the range of any recording or viewing medium, and film cameras had to deal with exactly the same issue as digital cameras. Film cameras had a bit of advantage in that they didn't clip like digital cameras do, which made any overexposure that inevitably occurred less painful. Digital cameras now have more range than film stocks did, but the limitation still exists. So part of the problem is one of artistic intent; if there's a highlight in the scene, it is one that's important, or one that's not (like the featureless intensity of the sun)? It's tough for a machine to sort this out; all the meter can do is make sure that MOST of the scene is within recordable bounds, or give you the precision to determine what you want recorded accurately. Photography is not capturing the scene as it is, but as you see it and want to tell it. Fully automatic exposure systems don't do that.

Point and Shoot cameras pretty much do what you want - they make sure that JPEGs are recorded as best they can. But they give up a lot in the process, and part of that is shadow noise. One of the advantages of digital capture is that there is a tremendous amount of information in the brightest ranges of the sensor's range, and we can exploit this by Exposing To The Right - increasing exposure to maximize recorded signal. This results in an overly bright image OOC and the potential for blown sensor channels, but it works great for RAW shooters. There have been some proposals for metering/imaging systems that ETTR for best signal/noise ratio, then back down on the rendered image in your LCD for review, storing the difference between recorded exposure and "perceptual optimum" exposure in the EXIF for PP program's benefit. I believe that pro video cams already do this, as well as some large-format still cams. But we're still waiting for Nikon and Canon to get a clue for their SLRs.

What might be useful is a dual shot mode, where you set the camera up and regardless of your settings it takes a second shot making sure the brightest part is not clipped.

That way you would have something to come back to in PP that you know cannot be clipped.

It's pretty basic zone system. Anything above zone 7 (+2) will be clipped. You need to use spot metering. Of course, this assumes that the meter works correctly and your exposure actually locks.

Mannypr said:

Why don't camera manufacturers design a metering system that takes into account digital photographs inherent problems , why not design a metering system that might monitors highlights and by doing so keep the highlights in check so that they don't clip ? I believe everything can be done in the digital realm and would not be all that dificult to implement. With a system like this it would be fairly easy to use the cameras full dynamic range without having to be worried about clipping the highlights.

Click to expand...

I think they're probably working on it but there are a few problems.

One is that the DR of many scenes is still more than that of the sensor, and that exposing for the hottest highlights may sink the shadows too much into noise. Well that's more like an excuse, at least they could give us a choice.

The resolution of the exposure (RGB) sensor may not be enough for things to work as some would expect; in live-view this is not an issue and one could for example define the max radius of specular highlights to be blown, but there may be issues related with processing speed.

Another is that they may not be able to process high DR with the current chips and algorithms. I guess this is a serious limitation currently, considering the odd sub-optimal implementation of ADL which chooses to underexpose raw data rather than do it in post-processing. HDR processing requires 32-bit or flow-point math, looks like the current in-camera processing is still 16-bit.

Another is good automatic HDR processing. If you want to show high DR you need to squeeze it into the viewable range, and by default that produces helluva flat results. There are certain recipes to squeeze DR but keep the image punchy but it is difficult to guess which one to apply to a certain scene. Heavy-handed processing is involved. I've noticed Sony's DRO seems be some way ahead of Nikon's ADL.

Yet another is the user interface that would have to make a split between raw data exposure and display exposure. That might be a tough thing for many users to chew. I think it is a serious challenge how to pull this through so that experts get what they want without complicating things for the masses and the customer support. Even the pros are accustomed to the old meter-centric concept and might need time to get used to sth different. But I think the market would reward a good solution because this is often a huge problem in practice.

gwlaw99 said:

It's pretty basic zone system. Anything above zone 7 (+2) will be clipped. You need to use spot metering. Of course, this assumes that the meter works correctly and your exposure actually locks.

Click to expand...

The procedure you described doesn't work with Nikon DSLRs. When in A, S, or P mode, the light meter display is used to indicate the amount of Exposure Compensation that's currently in effect. It will never change with luminance, even if AE Lock is active.

So you would do the opposite of what you described. You first spot meter the brightest area with detail that you wish to retain, lock exposure, and then simply apply +2 EC. Now your exposure is set to retain the highlights.

.

Mannypr said:

Why don't camera manufacturers design a metering system that takes into account digital photographs inherent problems , why not design a metering system that might monitors highlights

Click to expand...

I don't know why, probably history. I'd like a 'highlight priority' mode.
--
Erick - http://www.borealphoto.com

I thank you all for your answers and further thoughts on the matter . I admit not having thought some of the implications on such a system as many of you have mentioned . It's seems that if they do implement it there should be a way to disable it in order to have the best of both worlds depending on ones artistic intent . In any event you guys have broaden my thinking on the subject and thank you very much for your answers.

Ok thanks for correcting me. I only shoot in M so I did not know Nikon cameras worked that way.

blue_cheese said:

I think the point is that the "mid" tone is relative to what you want to have as a subject.

Click to expand...

I am not sure what you mean by this. Mid tone colors are always the same. Mid tone does not mean the middle tone in the shot. You can meter off mid tones because your camera always thinks you are metering off mid tones so by metering off them you get correct exposure. You have to familiarize yourself with what mid tone colors look like to do this. There is a color chart here that is fairly accurate. http://photo.tutsplus.com/articles/theory/understanding-using-ansel-adams-zone-system/

If you know where other tones fall on the zone scale you can meter off them as well making sure the meter reading is appropriate for that tone. For example, if you meter off snow, you want to change your aperture, shutter speed and iso until your meter reads +2 in manual mode. This is because your camera will think the snow is a mid tone and make it middle gray (a zone 5 tone) when it is really a zone 7 tone.

Clipping is likely to happen if one tries for a typical overall sense of correct exposure ooc. Meaning that in many cases, the exposure that protects HLs will look a bit dark w/o any adjustments.

Here is an example. The whites are just below clipping point. But image looks, at least to me, underexposed overall. One way to make it brighter w/o clipping HL would have been to use a softer tone curve, but then image would lack punch and people would say camera is not good.

Exposure 0.004 sec (1/250)
Aperture f/8.0
Focal Length 17 mm
ISO Speed 100



Now look at same file after some adjustments, using ACR first and CNX2 for local adjustments. In ACR I just used fill light (LLs) and recovery of HLs to compress DR a bit. Then I corrected exposure and did some local adjustments in CNX2 with u-point tool to bring exposure to what I think is like it looked in real life. Maybe it's still not ideal, but as a final print it would be better than ooc.



--
Renato.
http://www.flickr.com/photos/rhlpedrosa/
OnExposure member
http://www.onexposure.net/

Good shooting and good luck
(after Ed Murrow)

rhlpetrus said:

Here is an example. The whites are just below clipping point. But image looks, at least to me, underexposed overall. One way to make it brighter w/o clipping HL would have been to use a softer tone curve, but then image would lack punch and people would say camera is not good.

Exposure 0.004 sec (1/250)
Aperture f/8.0
Focal Length 17 mm
ISO Speed 100



Now look at same file after some adjustments, using ACR first and CNX2 for local adjustments. In ACR I just used fill light (LLs) and recovery of HLs to compress DR a bit. Then I corrected exposure and did some local adjustments in CNX2 with u-point tool to bring exposure to what I think is like it looked in real life. Maybe it's still not ideal, but as a final print it would be better than ooc.



--

Click to expand...

Renato, love your stuff and always have. But this time I'm on the other side. I much prefer the OOC pic vs the second. On my screen the 2nd looks a bit overexposed and somewhat harsh. That's why I love this hobby. It's so easy to produce what ever makes us happy. Thanks for sharing.....I just happen to like the first pic more.......But your explanation and example is excellent regards the subject.