"Meters Don't See 18% Gray"

[Fit]

This article is now seven years old... is it still generally true? Is the better metering reference 12%?

http://www.bythom.com/graycards.htm

It's an area I know little about and am just beginning to learn more about the finer points of photography.

[Colin Southern]

This article is now seven years old... is it still generally true? Is the better metering reference 12%?

http://www.bythom.com/graycards.htm

It's an area I know little about and am just beginning to learn more about the finer points of photography.

Hi Chris,

As I understand it, yes - but - I also understand that the difference is only 1/3 of a stop anyway.

[MiniChris]

I know that as well but the "given" number has always been the number 18 so rather than convolute the issue, I stayed with the norm. As Colin points out, it is splittling hairs when you are talking a 1/3 of a stop... The point to be made is the gray card sets a reasonable benchmark for white balance and what he lacked was that neutrality. As a Mini fanatic, I go to events all the time and shoot good car shots everywhere I go. I don't use a gray card everytime, but have been know to go that route on a particularly sunny day when there are lots of reflections and hot pavement. It would be a good way for him to at least achieve an even balance of light.

[Fit]

Thanks Chris, Colin. Any recommend brand/product/card? I'll just be shooting casually/learning.
Maybe this would be a good subject for a brief CiC tutorial?

[Peter Ryan]

Most camera shops would carry a 'grey card' if you ask. Brand is not so important. It is just a piece of cardboard coloured to suit.

[Fit]

Thanks Peter.

[Colin Southern]

Thanks Chris, Colin. Any recommend brand/product/card? I'll just be shooting casually/learning.
Maybe this would be a good subject for a brief CiC tutorial?

For white balancing I use a WhiBal card, for exposure I'll usually just go with whatever the camera recommends for casual shooting (and for not so casual shooting with EC and FEC dialed in) - or if it has to be right, then I'll use the light meter.

[jiro]

A couple of days ago I decided to experiment on a cheap gray card guide for photography. I decided to use a local gray spray paint (flat paint) that you can buy from Home Depot and painted a white cardboard with it. It took me 8 coats to have a decent solid gray color. After 15 minutes I tested it indoors (my room with 2 daylight balanced light bulbs as light source) and made about 5 test shots using the spot metering on my Nikon D70. For the test shot ( my working table full of books) the Matrix Metering Mode of the D70 recommended an exposure of f1.8 at 1/8 sec. When I placed the painted gray card in front of the area I wanted to shoot the spot metering recommends an exposure of f1.8 at 1/10 sec. Not bad for a cheap $2 can of paint.

I think it's nice to really nail a perfect exposure every time but if the difference is only 1/3 EV I believe your image can always be corrected on pp without much problem at all. Of course, a professionally available 18% or 12% gray card is really essential specially if taking pictures is your main occupation. But since I'm only an occasional shooter I'm already happy with my "cheap" gray cards. Pardon me for being so "cheap".

[PopsPhotos]

I have a gray cap from a film canister in my pocket when I'm out shooting. It works quite well as a gray card. Remember that you don't have to have perfect focus to get that gray card adjustment shot.

Pops

[arith]

Yes, I'm always being reminded about things I don't know. Never used a grey card for metering before, you don't even get a histogram for a film camera unless you use some special equipment. But film involves two exposures, one in an enlarger.
The only practical use of such a method of metering must be with close ups. I meter on incident light in landscapes and use GN for close flash, the only problem I get is for dark larger areas with complicated lighting.
But it also reminds me of things I just accept without question, such as WB and exactly what is going on when I simply click on a grey card in photoshop?

[shootreadyaim]

I was told by a teaching pro that a good middle gray is the color of grass that skin tone is two stops above middle gray and that certain grayish sidewalk in the usa were also good refrences for middle gray.

[joelk]

This article is now seven years old... is it still generally true? Is the better metering reference 12%?

http://www.bythom.com/graycards.htm

It's an area I know little about and am just beginning to learn more about the finer points of photography.

This old article is at least partly incorrect.

My tests indicate modern digital cameras seem to be set at 18%. This is testable with the proper gear (Light Meter, Spectrophotometer etc.). Older film cameras may have been set to 12-15%%. Probably depended on the camera. Cameras are not all set the same and there may even be variations in how two cameras of the same make behave due to manufacturing variances.

Pro users usually test their own gear to figure out what exposure compensation they should build in.

This article is much better than Thom's.
http://dpanswers.com/content/tech_kfactor.php

But I'd go so far as to say just forget 18% gray altogether when EXPOSING and start thinking in terms of holding critical highlights. Most cool modern cameras have an overexposure indicator. Turn it on! Increase your exposure until real white things start to clip and then back off by 1/3 of a stop and you'll likely have the maximum signal to noise ratio your camera can offer.

Photos have always been a combination of exposure and the printing process. Now the "printing process" is Photoshop/Printer combo and that's where the final tweaks will be made.

I wish I could take credit for that train of thought but the brilliant Tom Knoll (of Photoshop fame) lays out why that's smart here:
http://www.luminous-landscape.com/tu...se-right.shtml

Turns out - your automatic camera gets really close to that on easy shots. Exploring more creative photography with dramatic lighting often requires manual exposure control and possibly a light meter.

Finally - exposure changes can not fix bad light. Fixing bad light is what pros get paid to do.

:-)

[Colin Southern]

Hi Joe,

But I'd go so far as to say just forget 18% gray altogether when EXPOSING and start thinking in terms of holding critical highlights. Most cool modern cameras have an overexposure indicator. Turn it on! Increase your exposure until real white things start to clip and then back off by 1/3 of a stop and you'll likely have the maximum signal to noise ratio your camera can offer.

Not quite that simple I'm afraid. Anyone wanting to push things to the limit would be shooting RAW, but highlight alerts are based on in-camera generated JPEGs. Exposing to the right should - in my opinion - be considered more of an "available technique" rather than a "mantra"; when shooting high dynamic range scenes, it's mandatory (no ifs, buts, or maybes) - but if the scene is generally low dynamic range / reflective to start with, although ETTR gives you the best SNR, you also run the serious risk of getting at least one of the channels into areas of non-linear sensor response - at which point you end up with some weird colour shifts that are an absolute nightmare to try and correct in PP because the shift varies depending on the brightness of the tone.

Keep in mind also that the LL article you refer to is VERY old - modern DSLR camera typically have a DR of around 11 to 12 stops, not the 5 or 6 referred to in the article; ETTR to maximise SNR just isn't the issue it used to be (at base and low ISOs anyway).

[joelk]

Hi Joe,



Not quite that simple I'm afraid. Anyone wanting to push things to the limit would be shooting RAW, but highlight alerts are based on in-camera generated JPEGs. Exposing to the right should - in my opinion - be considered more of an "available technique" rather than a "mantra"; when shooting high dynamic range scenes, it's mandatory (no ifs, buts, or maybes) - but if the scene is generally low dynamic range / reflective to start with, although ETTR gives you the best SNR, you also run the serious risk of getting at least one of the channels into areas of non-linear sensor response - at which point you end up with some weird colour shifts that are an absolute nightmare to try and correct in PP because the shift varies depending on the brightness of the tone.

Keep in mind also that the LL article you refer to is VERY old - modern DSLR camera typically have a DR of around 11 to 12 stops, not the 5 or 6 referred to in the article; ETTR to maximise SNR just isn't the issue it used to be (at base and low ISOs anyway).

I agree there's that danger. It would greatly depend on the white you are referencing though.

(disclaimer - I've been researching the topic for a year in hopes of creating a perfect white card... well, I basically figured it out. Add grain of salt as you see fit... )

The better cameras these days have RGB histograms and will indicate if any of the channels are clipping. Is the highlight indicator triggering if any channel clips? I'd hope so, but I'm unsure on my camera and would have to test that. On lesser cameras I doubt it so there is the danger you mention.

The solution in either case would be to use a very white, diffuse card that is neutral in color - R,G & B are all reflecting 100% of the light back. Unfortunately that product is kinda rare out there... but I am testing a 99+% neutral white card and that will trigger overexposure on any channel. It works great with lesser cameras. Interestingly it yields exactly the same exposure as the program mode chooses on a 18% grey card (measured by me with a photospectrometer) on my 7D.

Thinking about the jpeg process - what's it doing? Adding dynamic range or losing some? You know it's not adding extra white information where there was none in RAW. It's more likely to be crushing whites if anything, right? So if you're shooting RAW that jpeg is possibly manufacturing a little buffer. I'd have to test it, but the jpeg process shouldn't just clip maximum white for no reason. And even if it did AND you're shooting raw then you're safe. I can't imagine the clip indicator not being a little conservative.

This is simply the zone system. Instead of placing a gray in Zone V like Ansel Adams did, you're placing a super white in Zone IX. Exact same idea - put a known value where it belongs in the zone and everything else will fall properly. These days I think you're more likely to make an overexposure mistake placing a gray too high than you are placing something extremely white too high because your histogram and clip indicators can indicate a white clip easier than showing gray is 10% too hot. Ergo... superwhite is easier and more conservative.

My real world results are that I am using less exposure than I might have when using the super white chip technique resulting in safer exposures. The reason is that white is whiter than basically everything else in existence. The sun, lightbulbs, fire, the edge of backlit clouds etc. are all brighter, of course.

I've measured an awful lot of "white" things... most are around 90% reflective and not neutral. So you're right... pushing your average "white" thing to the limit may do what you suspect - clip a color channel. The RED channel on certain woods is a good example. That ruins the color accuracy. Gray cards can easily cause the same problem if not used with care.

[Colin Southern]

Hi Joe,

The better cameras these days have RGB histograms and will indicate if any of the channels are clipping. Is the highlight indicator triggering if any channel clips? I'd hope so, but I'm unsure on my camera and would have to test that. On lesser cameras I doubt it so there is the danger you mention.

I shoot with a 1Ds3 which has the RGB histogram, so I know what you mean, but no, the highlight indicator doesn't trigger on a single blown channel.

Thinking about the jpeg process - what's it doing? Adding dynamic range or losing some? You know it's not adding extra white information where there was none in RAW. It's more likely to be crushing whites if anything, right? So if you're shooting RAW that jpeg is possibly manufacturing a little buffer. I'd have to test it, but the jpeg process shouldn't just clip maximum white for no reason. And even if it did AND you're shooting raw then you're safe. I can't imagine the clip indicator not being a little conservative.

JPEGs always discard information. JPEGs are really all geared towards small file size (for a given quality) so they use every trick under the sun to get the size down, including throwing away information we can't see (not withstanding a somewhat generous safety margin, so some editing is possible) - normalising tones that we can't differentiate between, and a compression based on an 8 x 8 pixel matrix. The thing I think it's important to remember is that most modern DSLR cameras are capturing 11 or 12 stops of DR (Off memory the Nikon D3s is up to 13 stops) - but we're only displaying around 6 on our screens and printing 4 on most papers (there are exceptions to both of those, but they're not typical). So are we throwing away 6 to 8 stops of available DR? Answer no - because what we're viewing / printing is after we've compressed and manipulated the captured DR to move it into something we can see, but the bottom line is that a typical reflective scene just doesn't need more than about 6 stops to capture it - and when we have 11 or (more typically) 12 stops available on the camera, we're operating around a full 5 or 6 stops above the noise floor with an ETTR exposure. My point is that although it's 5 or 6 stops with an ETTR exposure (at base ISO), 3 or 4 stops above the noise floor will still give perfectly clean shadows ... or put another way, with a RAW capture - of a reflected scene - at base ISO (or close to) - there's a LOT of exposure lattitude. So the better SNR of ETTR becomes something that's better in theory only; there's no practical "real-world" advantage (keeping in mind that I'm only talking about reflective low DR scenes here). For high dynamic range scenes (eg shooting into the light) things change though (and this is something I do all the time with my landscape) (you might be interested to look through my landscape gallery - just click the link at the end of the post if you are). In this case ETTR is essential (kinda, it's still my preference to gain around 3 stops of SNR "safety" by using GND filters whenever possible) - but - we still get the potential issues mentioned above (channel clipping / non linearity etc), but in practice, the (as you correctly point out) conservative nature of the in camera JPEG & highlight clipping alert tends to provide a bit of a safety margin against operating in the sensor non-linearity region - so in reality, when I'm shooting into the light I typicaly just rely on the highlight alert (and accept it's limitations); usually in conjunction with spot metering the brightest portion of the scene with a hand held lightmeter and up-shifting the exposure around 2 stops).

So in summary, what I'm saying here is that in theory ETTR has a place, but in many cases the existing mechanisms are perfectly adequate (eg highlight alert) - and in the vast majority of cases (eg purely reflective scenes) ETTR has the potential to create problems due to senser non-linearity and (in my opinion at least) is almost "a solution in search of a problem". Most of what I read about it is just good-sounding theory (much like "in theory" using a UV filter for protection degrades an image), but in reality (which is the realm I have to operate in, as a professional photographer) often the "theory" doesn't equate to a real-world advantage.

Sorry to ramble on

[PopsPhotos]

Joelk, here is a discussion of WB which fits into your post:

http://www.kenrockwell.com/tech/expodisc.htm

HISTORY and BACKGROUND (optional reading)

The phrase "White Balance" comes from color television. It is how we set a color TV camera to give accurate color in any kind of light. In Hollywood we also speak of gray and black balance and have controls for these, but not in consumer video or still cameras.

We set white balance while pointing a video camera at a white card.

In the 1950s we adjusted the gains of the three Red, Green and Blue tubes with screwdrivers until we got equal levels (balance) on a component waveform monitor or a dot in the middle of a vectorscope. Today we just press the WB button while pointed at a white card.

We use white cards, pieces of paper, backs of business cards or even white T-shirts. It's no big deal, so long as it's white or neutral. Whites are easy to find in the field; we never trust grays.

Pops

[joelk]

Sorry to ramble on

It was good ramble. :-)

Yeah, cameras these day make it a lot easier. I think we're agreeing here.

I mostly do cinematography and those guys really try to be precise with their exposures for a lot of reasons. It's not uncommon to see tweaks of 1/4 stop this way or that. When you're trying to make different shots look perfectly alike you'll find that a shot that's 1.5 stops under the previous shot is kinda tough to match just right. Also, shadow noise in a moving image is a much bigger deal than a still.

Theoretically you might have 3 stops to burn on a low dynamic range interior. Try it out with your 13 stop camera. Can you really underexpose by 3 stops and pull the underexposed image back up to match? A two stop shift is 1/4 the light. That's a lot of light to give away. Did navy blue turn to charcoal or is it still navy blue? That's what Knoll is talking about. There isn't much info in Navy Blue to start with so you don't have much room to underexpose it. OTOH You could underexpose Nicole Kidman by 3 stops and still get a nice skin tone for her.

Bad overexposure is more common in my experience. My superwhite chip has made me stop down by another 1/4 or 1/2 stop than what the camera is recommending in a fair number of cases. I still get fooled by the camera LCD. Histograms can be tricky when doing very dramatic lighting. That really leaves the overexposure indicator as the easiest thing for me personally to key on... but I'm always looking at the histogram too.

[joelk]

Joelk, here is a discussion of WB which fits into your post:
Pops

Thanks Pops. Since I've been shooting moving images for the past few years you have identified my bias. I think about white balance and holding real highlights a lot. The trick is to not get fooled by specular highlights. I've seen it happen to pros. They see a highlight warning on a camera and drop exposure too much. I saw some exterior day parking lot footage where the exposure was dropped to hold sun reflections off of glossy cars. The actors were African American. FAIL.

The sky was pretty though. LOL.

[Colin Southern]

Yeah, cameras these day make it a lot easier. I think we're agreeing here.

Agreed.

Can you really underexpose by 3 stops and pull the underexposed image back up to match? A two stop shift is 1/4 the light. That's a lot of light to give away. Did navy blue turn to charcoal or is it still navy blue? That's what Knoll is talking about. There isn't much info in Navy Blue to start with so you don't have much room to underexpose it. OTOH You could underexpose Nicole Kidman by 3 stops and still get a nice skin tone for her.

I think Guillermo posted an excellent article (with example) on this here ...

Is the end of bracketing for HDR nearing?

[Colin Southern]

Joelk, here is a discussion of WB which fits into your post:

http://www.kenrockwell.com/tech/expodisc.htm


Pops

Hi Pops,

Slightly off topic, but I felt it's worth mentioning that I have 2 Expodiscs, but don't find them particularly useful as you just can't use them easily when you're in a fill flash situation (which I usually am). Out of interest, I've been having a few frustrations in the studio recently so I got the old expodisc out and set a custom WB pointed straight into an octabox attached to one of my studio strobes. I thought "this'll be good because the colour temp in the studio is constant - and this'll mean the "as shot" setting will now yield deathly accurate skin tones that I won't ever have to adjust in PP" ... but the short answer is "I still have to adjust them". Why? Who knows, but I do.