White Balance....white, gray, and black WB cards.

[Gavjenks]

I've seen actual controlled examples as well, they showed a reduction in saturation and an increase in granularity of the color in the real world.

Saturation: What leads you to believe that either the higher or the lower saturation represents lower or higher accuracy compared to the real world? Maybe lower saturation is more truthful... And regardless, the RAW has so much vastly more color depth than the final jpeg or print does that unless it changes by like half a dozen stops, it shouldn't matter. You could readjust as you like either way with no observable side effects.

Color granularity: not sure what you mean by this exactly. You mean like banding/posterization? Or something more like dithering? (E.g. right versus middle or left of this diagram) http://provideocoalition.com/images/uploads/Colour_banding_example01.png Or...?

 

[Scatterbrained]

First, I don't know anyone who shoots landscapes or street scenes with such cards. Landscapes and street/candid scenes as well as portrait work are often done "to taste" anyway. Processed to reflect a mood rather than render technically accurate tones and colors. Before the days of digital a scene such as you're describing would have been metered in several different areas by a reflective spot meter, then those readings would have been compared and weighted by the photographer before an exposure decision was made.

Secondly, the cards go where the subject will be in the frame, then you fill the frame with the cards. That way the histogram will only reflect the cards themselves. Then you'll know your subject will be accurately exposed in the target area. The OP is an artist, photographing his artwork. As I have pointed out before, when absolute color and tonal accuracy matter, this is the technique used to achieve it. Photographing a city scene from across a river doesn't require anything near absolute color accuracy, nor can it even be achieved in such a situation as such scenes often involve the use of monochromatic lights sources.

 

[Scatterbrained]

I've seen actual controlled examples as well, they showed a reduction in saturation and an increase in granularity of the color in the real world.

Saturation: What leads you to believe that either the higher or the lower saturation represents lower or higher accuracy compared to the real world? Maybe lower saturation is more truthful... And regardless, the RAW has so much vastly more color depth than the final jpeg or print does that unless it changes by like half a dozen stops, it shouldn't matter. You could readjust as you like either way with no observable side effects.

Color granularity: not sure what you mean by this exactly. You mean like banding/posterization? Or something more like dithering? (E.g. right versus middle or left of this diagram) http://provideocoalition.com/images/uploads/Colour_banding_example01.png Or...?

It's easy to tell if you're getting accurate colors simply by using a color card. That's what the colorchecker card is for. Or did you thing these things were just "relics" from the film era?

 

[Gavjenks]

Okay yes, for artwork, the street scene thing is not a DISadvantage. But I'm still failing to see what any advantage is. Considering cards cost money and must be carried around and remembered, etc., they need to accomplish something that the histogram cannot in order for them to be worthwhile.

The gray card is useful. I'm not saying it isn't. If you have a good reason to want accurate color (like products, as you say), it provides a reference for WB that cannot currently be precisely determined otherwise for sure. But I'm still not seeing what the white or black cards are bringing to the party.

Unlike WB, though, there is no such thing as absolute "tonal accuracy" in the first place. The relationship between tones and the world is only defined with respect to a given exposure, and the choice of 18% reflectance as halfway through modern camera histograms usually, is a purely arbitrary choice that isn't any fundamentally more or less "correct" or "accurate" than 37.37% reflectance as neutral gray would be. Or 90%. Or 2%. Or anything else.

One might hypothesize that you could call the "correct" number to be "the reflectance that the average human eye sees as middle gray." But, of course, there is no such thing, because the eye constantly adapts in all sorts of crazy ways that cameras don't. So there is no anchor to justify any one middle gray over any other.



Edit: In fact, not only is the center point of neutral gray arbitrary, but so is the compression scaling arbitrary. I.e. the difference between top and bottom of the histogram being about 5 stops or whatever the cards tell you to do = arbitrary choice. It would be no more or less "correct" to aim for 7 stops or 4 stops total range.

 

[Gavjenks]

Also, what do you do if your subject has significantly lower or higher reflectance materials in it than your cards are? For example, if your black card is at 5% reflectance, and you set it at a position where black details are still distinguishable, then your product photo will have very blocked up shadows if the product is, for example, a piece of clothing with black velvet on it, which could be more than 3 stops blacker than such a card.

 

[table1349]

Simple, it's called experience.

 

[WayneF]

When using a neutral card for setting WB, is there a particular reason for choosing one of the three. I just ordered all 3..... I watched several video on YouTube about them, but no one ever explained why they chose the variation that they were using.. at least between the white and gray (they didn't show using the black one.)

You really don't want an 18% gray card. It will work, but white is better for WB (assuming both are in fact neutral color). Any neutral color (equal RGB components) can work. Black can work, but it is too dark to work very well. An 18% gray card is quite dark too (18%). An 18% card is a film concept for exposure, not for white balance. 18% is 18%, it is NOT 50%, not in any imagination. The only reason it appears up almost near the histogram midpoint is because RGB data has gamma boost added (changing 18% to 46% in the data, but the photo tone is 18%). An 18% tone is dark, and cannot show much of a pink tint. In contrast, white is bright, and is able to show any tint much more predominately (so it can be identified and processed).

Again, 18% is an analog film concept (for printing and for exposure, not for WB). The "gray cards" for digital white balance use (like say the WhiBal card) are NOT 18%. They are much lighter than 18% (because 18% is too dark, and lighter is better). The concept is "White Balance". If you buy a set of the "three" cards, black, gray, and white, they will give three WB values. Trust the white more. Trust a good white card even more.

Here is a great WB card that is inexpensive:
Porta Brace White Balance Card WBC B&H Photo Video

For $5, it is less critically controlled than say a WhiBal card. However, not much need or concern, white is easier (no dyes to have to measure), and it is more than great for WB.

 

[Scatterbrained]

Also, what do you do if your subject has significantly lower or higher reflectance materials in it than your cards are? For example, if your black card is at 5% reflectance, and you set it at a position where black details are still distinguishable, then your product photo will have very blocked up shadows if the product is, for example, a piece of clothing with black velvet on it, which could be more than 3 stops blacker than such a card.

It's called knowing how to set up your lights. Do we need to rehash this again? All you're doing is showing that you have little to no experience shooting product in a studio environment.

 

[Scatterbrained]

Okay yes, for artwork, the street scene thing is not a DISadvantage. But I'm still failing to see what any advantage is. Considering cards cost money and must be carried around and remembered, etc., they need to accomplish something that the histogram cannot in order for them to be worthwhile.

The gray card is useful. I'm not saying it isn't. If you have a good reason to want accurate color (like products, as you say), it provides a reference for WB that cannot currently be precisely determined otherwise for sure. But I'm still not seeing what the white or black cards are bringing to the party.

Unlike WB, though, there is no such thing as absolute "tonal accuracy" in the first place. The relationship between tones and the world is only defined with respect to a given exposure, and the choice of 18% reflectance as halfway through modern camera histograms usually, is a purely arbitrary choice that isn't any fundamentally more or less "correct" or "accurate" than 37.37% reflectance as neutral gray would be. Or 90%. Or 2%. Or anything else.

One might hypothesize that you could call the "correct" number to be "the reflectance that the average human eye sees as middle gray." But, of course, there is no such thing, because the eye constantly adapts in all sorts of crazy ways that cameras don't. So there is no anchor to justify any one middle gray over any other.



Edit: In fact, not only is the center point of neutral gray arbitrary, but so is the compression scaling arbitrary. I.e. the difference between top and bottom of the histogram being about 5 stops or whatever the cards tell you to do = arbitrary choice. It would be no more or less "correct" to aim for 7 stops or 4 stops total range.

Well, it's all so arbitrary, why bother with light meters, grey cards, white cards, color cards, etc? But I guess you've stumbled onto some deep wisdom that the entire imaging industry has been missing for the past 100yrs.

Here's a quick scenario for you: A textile manufacturer contacts you and wants shots of 250 different colored swatches of fabric for their catalog. What are you going to do? Feel free to just ETTR it all day. Let me know what you come up with. How are you going to handle a scale of red that goes from almost black to blazing red over 24 different swatches? How are you going to know if the swatch your looking at on your computer is supposed to be white, or maybe that was actually the slightly off white one and you over exposed it? Was that swatch supposed to be blood red, or maybe it was deep burnt orange? How can you know?

 

[Gavjenks]

Simple, it's called experience.

It was a rhetorical question, to highlight the fact that the answer is not "plastic cards"

An 18% tone is dark

I think you are confused. 18% has nothing to do with dark or light. 18% refers to the percentage of photons that reflect off of the surface. Whether that surface is light or dark is entirely then based on how many photons there are hitting the surface, and the reflectances of other objects in the scene.

An 18% reflectance surface in a room full of 2% reflectance surfaces would be bright white.
An 18% reflectance surface in a room full of 90% refelctance surfaces would be dark black.
An 18% reflectance surface in a lush grassy field (around 20% reflectance) would be almost perfectly neutral gray.

Presumably, 18% was chosen due to some ancient guess or research or who knows that suggested that the average reflectance of "typical" scenes in the world (or perhaps the central anchor point of our eye adjustments for "typical scenes") is around 18%. However, I've never seen any data or actual scientific reasoning behind this, so in reality it's mostly just an arbitrary number that is neither dark nor light nor right or wrong particularly. Just a number.

But it's "supposed to be" about 50th percentile reflectance compared to... some sort of average of something.

 

[Gavjenks]

Well, it's all so arbitrary, why bother with light meters, grey cards, white cards, color cards, etc?

No, it's NOT all arbitrary.

Wavelength is detected in our eyes by chemicals that are identical in everybody (except for some who are completely missing one or more, i.e. colorblind). They therefore have identical response curves to wavelengths, and there is a very very very high consistency across individuals.

Brightness, on the other hand, is not even remotely controlled or consistent, not even within a person from one moment to the next. Unlike wavelength, it is detected along a smoothly sliding scale of sensititvity, due to iris size, dark adaptedness, etc. Even if you could measure some sort of "average," the variance would completely swamp the influence of any sort of minor card calibrations to the point where it makes no practical difference. Plus, we don't even have that average data anyway.

Thus, white balance has a much less arbitrary, more absolute reference point, and tonality does not.

How are you going to handle a scale of red that goes from almost black to blazing red over 24 different swatches?

By ETTR I don't mean every single shot, if you're doing a whole product line...
In this situation, what I mean would be that you'd take the lightest swacth, ETTR it, and then leave your camera at that same exposure in manual and shoot all of the rest of the swatches.
A process which gives customers exactly as much information about which swatch to pick as yours relative to one another, and is no more "right" or "wrong" (and no more or less likely to match the lighting in their living room), yet it doesn't require the purchase of additional gear, thus I'm suggesting it is superior due to cost and simplicity of stuff to remember and haul around.

 

[Scatterbrained]

Because it's a logarithmic scale, not linear.

 

[Scatterbrained]

Well, it's all so arbitrary, why bother with light meters, grey cards, white cards, color cards, etc?

No, it's NOT all arbitrary.

Wavelength is detected in our eyes by chemicals that are identical in everybody (except for some who are completely missing one or more, i.e. colorblind). They therefore have identical response curves to wavelengths, and there is a very very very high consistency across individuals.

Brightness, on the other hand, is not even remotely controlled or consistent, not even within a person from one moment to the next. Unlike wavelength, it is detected along a smoothly sliding scale of sensititvity, due to iris size, dark adaptedness, etc. etc.

Thus, white balance has a much less arbitrary, more absolute reference point, and tonality does not.

How are you going to handle a scale of red that goes from almost black to blazing red over 24 different swatches?

By ETTR I don't mean every single shot, if you're doing a whole product line...
In this situation, what I mean would be that you'd take the lightest swacth, ETTR it, and then leave your camera at that same exposure in manual and shoot all of the rest of the swatches.
A process which gives customers exactly as much information about which swatch to pick as yours relative to one another, and is no more "right" or "wrong" (and no more or less likely to match the lighting in their living room), yet it doesn't require the purchase of additional gear, thus I'm suggesting it is superior due to cost and simplicity of stuff to remember and haul around.

I think what you're missing is that when you print the image it should look just like the actual item would. That's the same concept as "matching prints to screen". Ever hear anyone print an image and complain it came out "too dark"? Again, if you overexpose something you change the tone. Then when you print this overexposed image the tone won't match the actual item. Someone orders the item and it doesn't look the same. If they ordered something expecting dark grey and got black they'd be pissed off, I know I would be. Maybe they thought they were getting pink but they actually got a bright red. The problem with ETTR is that for one you're not using an accurate exposure, so you're changing the tones. Secondly, now you have to go back and try to correct the tones afterwards, something that you may or may not be able to do. What are you going to do, hold each swatch up to the screen and adjust until they match? Are you viewing with a calibrated monitor under controlled, neutral, full spectrum illuminant with no color casts from nearby objects? Somehow I doubt it. Are you aware of, and have you corrected for, the spectral deficiencies of your shooting lights? If not, you're now shooting in the dark again.

 

[Gavjenks]

I think what you're missing is that when you print the image it should look just like the actual item would.

Yes that's the ideal. But your system does not accomplish it.

For example:
Product #1: A colored candle
Product #2: A pair of skis

The candle is designed to be viewed almost exclusively in darkened rooms. The average luminance of objects in such an environment is very low, and the wax therefore--so near the flame--will be relatively much lighter than its surroundings. If you truly wanted to depict the product as it would be seen, you'd have to make it much lighter than your standard cards suggest to compensate for the advertisement not being seen (usually) amongst pitch blackness.

The skis on the other hand, are mainly going to be used in mid day, in a blinding white environment. If you truly wanted to depict the product as it would be seen in its intended surroundings, you'd have to make it much darker than your standard cards suggest to compensate for the advertisement being seen in non-blinding white environments.

Do you make those types of adjustments when you shoot products? Of course not. It's not even possible, because we don't have any sort of data anyway to make those adjustments.




What you're actually doing is simply picking an almost random set of numbers (5%, 18% 75% or whatever) that you have no particular reason to believe actually represent anything to do with the environment your product will be used in, and standardizing all your shots to that.

Which is no more likely to be accurate (in the sense of matching how the product will actually look) than my method is, for any given product!

You are throwing your darts blindly off in one direction.
I am throwing my darts blindly off in another.
NEITHER of us know where the actual target is.

Thus, it's impossible to predict which one of us will win at darts.
However, my way requires buying and using less equipment...

If they ordered something expecting dark grey and got black they'd be pissed off, I know I would be.

Sure they might be pissed off. But since it's utterly impossible to know what the average reflectances are of each individual shopper's home decor, and their average room versus monitor lighting... it's utterly impossible to avoid this.

 

[Scatterbrained]

This, again, just shows me you don't know how this system works. Have you ever seen someone using a proofing booth? Holding an image of the product next to the product to make sure the colors match? It's likely you haven't. That's how this system works. As far as different viewing conditions. If you hold the skis next to a picture of the skis they will look the same. That's the point. I'm imagining that due to you color/tonal indifference that you don't bother calibrating your monitor either? You certainly don't print your images, I can be fairly sure of that. When you go to the store to buy a candle is it already burning, in a room with no lights? Not likely. It will likely be sitting on a shelf or on a display table, fully illuminated by room light. Shooting for absolute accuracy from the start means that the end product will give the most accurate representation when viewed by the client or potential customer. Have you never purchased from a catalog before? Aren't you glad the art directors and photographers took the time to ensure the products were accurately represented so that when they showed up at your house you got exactly what you were expecting?