Rutt's First Law of Dgrin

rutt

If someone asks whether a picture of a person looks better in color or B&W, the flesh tones of the color version measure A>B in LAB.

This is a predictive empirical statement. As such it is subject to objective in/validation by observation. I am firmly convinced that I have investigated every thread I've ever seen and have yet to see a counter example. But I had difficulty finding the old examples by searching. So I'm establishing this thread to keep track of observations which conform to the hypothesis of my theory. To kick it off:

http://dgrin.com/showthread.php?t=158832

Oh, and there is another theory I'm formulating which is closely related: the photographer will tend to reject adjusted flesh tones as posted (but not as printed) in direct proportion to level of experience.

Anyway, please help me by collecting both examples, bot ones that tend to reinforce and counterexamples.

bfjr

I'll jump in by including this thread as well:
http://www.dgrin.com/showthread.php?t=158261

rutt

bfjr wrote:

I'll jump in by including this thread as well:
http://www.dgrin.com/showthread.php?t=158261

The link to the color version in this thread is broken.

bfjr

rutt wrote:

The link to the color version in this thread is broken.

Fixed it

bfjr

Little wheels turning
Color or B&W

Wil Davis

Colour has it (although a tad yellow); all eye (iris) detail has been lost in B&W…


- Wil

rainbow

rutt wrote:

If someone asks whether a picture of a person looks better in color or B&W, the flesh tones of the color version measure A>B in LAB.

Can you explain the "A>B" that you refer to? I assume these are some basic parameters used in processing...

Thanks in advance.

rutt

Not the point and you guys know it. OK, maybe the law needs an exemption for cooked up counterexamples. But really, watch what people do when they don't know they are being watched.

rutt

rainbow wrote:

Can you explain the "A>B" that you refer to? I assume these are some basic parameters used in processing...

Thanks in advance.
[/INDENT]

Basically measures more magenta than yellow. The first couple of links here will get you off the ground.

pathfinder

bfjr wrote:

Color or B&W

I don't care for the jaundiced look myself.

nickeverett

Another example to prove the rule: http://www.dgrin.com/showthread.php?t=158988

rutt

nickeverett wrote:

Another example to prove the rule: http://www.dgrin.com/showthread.php?t=158988

Barely. It's actually more of a counterexample. I'd let that first one pass if it were mine in spite of having a few B>A places, it's A<=B almost everywhere that counts.

rutt

pathfinder wrote:

I don't care for the jaundiced look myself.

Take off you good photographer hat and put on your sociologist hat. The question is whether people ask this question overwhelmingly when the subject is too magenta. I still postulate that they do.

bfjr

pathfinder wrote:

I don't care for the jaundiced look myself.

Actually the color is balanced against the knob (lower frame/left)

The parents don't like the color because they feel the bug bite on his forehead is very
distracting (I agree). They don't like the B&W they feel it's to dark but they do want the image
and love the posture of their child as it does seem to capture that, "Wheels are Turning" look.

So I'm still working on it.

rutt

bfjr wrote:

Actually the color is balanced against the knob (lower frame/left)

The parents don't like the color because they feel the bug bite on his forehead is very
distracting (I agree). They don't like the B&W they feel it's to dark but they do want the image
and love the posture of their child as it does seem to capture that, "Wheels are Turning" look.

So I'm still working on it.

OK, OK, OK. It's off my original subject, but I can't resist. I've stopped taking one-click color balance on neutral objects as gospel and always check flesh tones. Very pale skin as here is always very fragile and easy to get wrong in post. In this case, I'll go with Jim and say this measures too yellow (which is sort of rare.) So maybe this really falls into some generalization of Rutt's Frist Law of Dgrin. But I digress...

What if you do the absolutely most standard thing:

1. Healing brush on that bug bite,
2. RGB curves in a color blended layer to bring up the red just a tad in the quarter tones (where his face is.)
3. Paint onto a color-blended layer to get the purple out of the area around his eyes. I may have overdone this, but if you do it yourself, you'll have control over the opacity.

bfjr

rutt wrote:

OK, OK, OK. It's off my original subject, but I can't resist. I've stopped taking one-click color balance on neutral objects as gospel and always check flesh tones. Very pale skin as here is always very fragile and easy to get wrong in post. In this case, I'll go with Jim and say this measures too yellow (which is sort of rare.) So maybe this really falls into some generalization of Rutt's Frist Law of Dgrin. But I digress...

What if you do the absolutely most standard thing:

1. Healing brush on that bug bite,
2. RGB curves in a color blended layer to bring up the red just a tad in the quarter tones (where his face is.)
3. Paint onto a color-blended layer to get the purple out of the area around his eyes. I may have overdone this, but if you do it yourself, you'll have control over the opacity.

[img]http://www.ruttpix.com/photos/788337806_pVvCm- S.jpg[/img]

Excellent, I will proceed in that direction. Actually I had but yours still looks cleaner, so I will have
at it again when time permits.

And even more off the subject : That Yellow
well do you think it might have to do with the fact that Parents are raising Child as a, "Vegan"?
Not kidding I being seeing that "Yellow" more and more in my captures of him.

pathfinder

Ben, I started a post earlier and deleted it as I thought it me be taken as flippant.

I have made myself a rule not to shoot folks wearing yellow green shirts. They just are not that attractive and reflect green light where it is not needed.

Diets very high in carotene ( yellow or orange colored vegetables ) may very well give a more yellow cast to the skin.

bfjr

pathfinder wrote:

Ben, I started a post earlier and deleted it as I thought it me be taken as flippant.

I have made myself a rule not to shoot folks wearing yellow green shirts. They just are not that attractive and reflect green light where it is not needed.

Diets very high in carotene ( yellow or orange colored vegetables ) may very well give a more yellow cast to the skin.

THanks, good info that makes sense. No big deal to have the boy change shirts.
I just wish I could get the parents to understand the word, "Omnivore".

You know you can be as flippant as you like with me, I'd just flip it back at ya

BinaryFx

John - I personally would temper your "1st law" when it comes to very young children, even more so if blue-eyed and blonde or red hair...as such children probably should have more magenta than yellow in their skin when evaluating Lab colour values (that is unless they are known to be very tan). In these cases, the poster asking whether colour or montone may very well be asking for "mood" reasons and is not being affected by the colour skin values in a negative way. Your "1st law" may indeed hold for those with dark eyes/hair and older ages.


Stephen Marsh

http://members.ozemail.com.au/~binaryfx/
http://prepression.blogspot.com/

Seefutlung

I like the B&W (with a little work the blacked out eye issue can be easily resolved).

Gary

PS- I always go B&W when shooting vegans.
G

pathfinder

bfjr wrote:

THanks, good info that makes sense. No big deal to have the boy change shirts.
I just wish I could get the parents to understand the word, "Omnivore".

You know you can be as flippant as you like with me, I'd just flip it back at ya

It is just so easy to be misunderstood on the Web, Ben

pathfinder

Seefutlung wrote:

I like the B&W (with a little work the blacked out eye issue can be easily resolved).

Gary

PS- I always go B&W when shooting vegans.
G

I have to ask - is this a straight informative statement, or a joke? Inquiring minds need to know!

rutt

BinaryFx wrote:

John - I personally would temper your "1st law" when it comes to very young children, even more so if blue-eyed and blonde or red hair...as such children probably should have more magenta than yellow in their skin when evaluating Lab colour values (that is unless they are known to be very tan). In these cases, the poster asking whether colour or montone may very well be asking for "mood" reasons and is not being affected by the colour skin values in a negative way. Your "1st law" may indeed hold for those with dark eyes/hair and older ages.

The law is more about the dgrin population than about good post processing. It's really just an observation about what makes people on the "people" forum consider B&W. It seems that mostly it's because there is too much magenta in the flesh and they don't know how to recognize and/or fix that problem. They just know the color is ugly.

Ben's picture is an interesting example because it also has ugly color but B>>A. I've never seen this before, but it seems to me that I've seen dozens of examples with A>B. The fix there was to get the balance a little closer. So you were right. The right generalization of my rule is something like:

If someone asks whether a picture of a person looks better in color or B&W, the flesh tones of the color version are out of balance, most likely A>B but sometimes with light skinned people...

Seefutlung

pathfinder wrote:

I have to ask - is this a straight informative statement, or a joke? Inquiring minds need to know!

If you have to ask ... then my remark was an attempt at a joke.

Gary

pathfinder

Some of us are just a bit slow, Gary:D

LiquidAir

This may also be a bit off topic, but I do my white balancing in Lightroom where I don't have easy access to LAB. My guideline for skin tone is:

R-G ~ G-B

i.e. R should be about as much larger than G as G is than B.

For a properly normally exposed image with Caucasian skin I tend to adjust the Lightroom curves so the green in the (non-specular) highlights is about 80. Then I tweak the temperature and hue to get the red and blue where I want them.

R = 90, G = 80, B = 70 is about as saturated as I am willing to go.
R = 83, G = 80, B = 77 starts looking pretty pale.

Generally I know what kind of lighting I shot under so I'll slide the temperature control in range first watching the G-B gap. Then I tweak the hue to dial in the red channel. Finally I salt and pepper to taste. Any kind of mixed lighting (including bounce off a green shirt) is going to require some compromises in the white balance so there will always be judgement calls, but I do use the numbers as a guideline to keep me from getting too far off track.

jjbong

This maps nicely into what Dan Margulis and others have recommended using CMYK. High red means low Cyan, Blue less than Green means Yellow higher than Magenta.

Just an observation, as I don't want to rekindle the CMYK debate. But it's easier to accept these guidelines when they reduce to an easier application of what others have been doing.

BinaryFx

jjbong wrote:

This maps nicely into what Dan Margulis and others have recommended using CMYK. High red means low Cyan, Blue less than Green means Yellow higher than Magenta.

Just an observation, as I don't want to rekindle the CMYK debate. But it's easier to accept these guidelines when they reduce to an easier application of what others have been doing.

Indeed, whether one evaluates in Lab, RGB or CMYK - the same general ratio applies for a "healthy" tan.

Dan did not invent the ratios, they were long used by drum scanner operators long before scanners had graphical displays and they were thus "flying blind".

Attached is an image illustrating the relationships of the primary colours in each mode.


Stephen Marsh

http://members.ozemail.com.au/~binaryfx/
http://prepression.blogspot.com/

michswiss

This is an incredibly technical discussion for me, but I do have a question. Is the rule (observation) meant to apply to people with a particular skin tone? Reason why I'm asking is that most of my shots tend to be people of Asian or Chinese decent. There is even kodak film stock that has been tuned to improve skin tones.

I'm starting to play with colour images in a small way again, and you can be fairly certain they will include people.

BinaryFx

michswiss wrote:

This is an incredibly technical discussion for me, but I do have a question. Is the rule (observation) meant to apply to people with a particular skin tone? Reason why I'm asking is that most of my shots tend to be people of Asian or Chinese decent. There is even kodak film stock that has been tuned to improve skin tones.

To review, this is about Caucasian folk that need a nice tanned appearance, viewed in normal lighting and not in shadow and not measuring makeup or people exercising or in the snow etc. The traditional rule of thumb for CMYK evaluation was for the base magenta value, the cyan should be from 1/4 to 1/2 the value of magenta - with yellow being near to equal to magenta or just a little bit higher.

So very limited circumstances.

As for Asian - it all depends! However to generalise for darker skinned Asians (some have light skin)...then one may expect higher values in CMY, with the ratio of cyan to magenta being much stronger than with Caucasian, so at least half the magenta value if not stronger. Yellow would also be higher than magenta in most cases - however it is very important not to go overboard and make them jaundiced, so if in doubt have the yellow near the same as magenta (also a good rule for any race).

A profiled monitor should probably be your main guide, however reading the values in the image can also help a great deal, both methods compliment each other.

P.S. The attached image is in sRGB, avoid evaluating makeup, try for the unshadowed neck/upper chest in the image below.

Asian musician: Lab - L64 a14 b17, sRGB - r187 g146 b127, Fogra39 CMYK - c23 m42 y45 !k


Stephen Marsh

http://members.ozemail.com.au/~binaryfx/
http://prepression.blogspot.com/

BinaryFx

Another sample attached.

As there are different "flavours" of CMYK and RGB, one should note the ICC profile when mentioning the numbers in the sample...which is why Lab mode readings are usually the best as Lab does not need an ICC profile.

Asian model, under collar bone: Lab - L51 a23 b25, sRGB - r166 g105 b80, Fogra39 CMYK - c27 m59 y64 !k

I repeat: A profiled monitor should probably be your main guide, however reading the values in the image can also help a great deal, both methods compliment each other.



Stephen Marsh

http://members.ozemail.com.au/~binaryfx/
http://prepression.blogspot.com/