Steve Thai

This is the discussion thread for the September 2010 Class.
sjordan93436
Posts: 462
Joined: Sun Jul 11, 2010 4:23 am
Contact:

Postby sjordan93436 » Tue Oct 05, 2010 4:36 am

Just for chuckles  I tried the Thai image.

I assigned a profile with a 1.0 gamma.  Then, LAB for lightness and color.  I assume the sky is yellow because the faces were dark.  I can try to fix that later.

First, jpg is the result of that.  The second jpg is a conversion to lab, duplicate layer, multiply, and apply the L channel to the mask. 

It still needs work.  When I have time, I should go back and do this without the gamma.
Attachments
thai-a-ss-jpg
thai-a-ss-jpg (999 Bytes) Viewed 7518 times
thai-a-jpg
thai-a-jpg (999 Bytes) Viewed 7518 times
thai-a2-jpg
thai-a2-jpg (999 Bytes) Viewed 7518 times

ggroess
Posts: 5342
Joined: Wed May 24, 2006 2:15 am
Contact:

Postby ggroess » Tue Oct 05, 2010 5:05 pm

It is better than It was...
I would make a small list of things you want to fix and see if you can tackle them. 
This is one image that can be many different things but you have a lot of choices...

So far you are no worse off than anyone else who has attempted this one...Keep going when you have time...

Greg

sjordan93436
Posts: 462
Joined: Sun Jul 11, 2010 4:23 am
Contact:

Postby sjordan93436 » Tue Oct 05, 2010 5:28 pm

Thanks.  I can see contrast issues and noise issues.  This may be common for this kind of photograph.  If you brighten as normal, you lose some "atmosphere".  Like a flash portrait of a subject in front of a sunset.  There is some artistic decisions.

Perhaps, this should be a separate topic...

Hue clocks.  Let me repeat (?) what the docs say.  The angle of the hue clock is "hue".  The length is "saturation"  and Lightness or brightness has no effect on the hands of the clock.  I guess if I blow out an area with brightness that can affect, but within reason.


Therefore to color correct by the hands of the clock...  in rgb or cmyk, you have usually two colors (one if the hand points at that color) that pull the hand of the clock.  The hand on the opposite side pulls the hand, but also reduces the length.  It is geometry. 

All three colors can push or pull the hand in the right direction unless it is pointing directly at or away. 

Sometimes, the hand is in the right direction, but too long or too saturated.  Specifically, skin tones can easily be too red/ yellow.  The hand is at 12:30 but too long.  In that case, reduce the red and green or increase the blue.

If the color is too dark, then another pass in LAB. 

Is that sort of right?

BTW- I did an interesting experiment or demonstration.  I took an image and put in hue clocks.  I went into HSL color space and watched the hands-  cool.  Just as you would expect.  the hands were not totally uniform but it was very interesting.

My misconception was that pinning the skin would solve the problem.  If it was too far gone, that is not correct.  My problem was not fixing the saturation.  (in general)

ggroess
Posts: 5342
Joined: Wed May 24, 2006 2:15 am
Contact:

Postby ggroess » Tue Oct 05, 2010 6:30 pm


Thanks.  I can see contrast issues and noise issues.  This may be common for this kind of photograph.  If you brighten as normal, you lose some "atmosphere".  Like a flash portrait of a subject in front of a sunset.  There is some artistic decisions.

Perhaps, this should be a separate topic...

Hue clocks.  Let me repeat (?) what the docs say.  The angle of the hue clock is "hue".  The length is "saturation"  and Lightness or brightness has no effect on the hands of the clock.  I guess if I blow out an area with brightness that can affect, but within reason.

Those would be correct statments the touble is that the hue clock is very angle dependent but not very accurate.  You would have to adjust the hue channel of HSB to a specific Hue angle to be accurate and very few people think about color that way...I guess it could be done but ...Ouch...


Therefore to color correct by the hands of the clock...  in rgb or cmyk, you have usually two colors (one if the hand points at that color) that pull the hand of the clock.  The hand on the opposite side pulls the hand, but also reduces the length.  It is geometry. 


It's about that simple... yes...If you look at a color wheel the color opposites apply Red is opposite Cyan Green is opposite Magenta and Yellow is opposite Blue.  Visually Blue actually looks more purple. We combine adjustments to be rid of color casts.  That is why in RGB if you have a magenta cast you have to adjust red and blue or...green but very rarely all three.  Adjusting all three generates neutral density.    It is just like the CMYK process.  You can reduce the CMY values by adding K


All three colors can push or pull the hand in the right direction unless it is pointing directly at or away. 


See explanation above about all three colors...


Sometimes, the hand is in the right direction, but too long or too saturated.  Specifically, skin tones can easily be too red/ yellow.  The hand is at 12:30 but too long.  In that case, reduce the red and green or increase the blue.


If the Saturation is driving the color to be off then it may have been wrong in the first place.  Remember that the "12:30 Rule"  is a "Big Thumb Rule..."  It will get you in the ball park but it is not exact.  Funny thing about Skin tones...We can have vastly different tones in the same image..that is why we here at CM do not give you an exact skin tone to hit but a range...there is a ton of "normal" out there...


If the color is too dark, then another pass in LAB. 

Is that sort of right?

Lab can help but sometimes you can try other things for brightness issues.  For instance..Take the Thai picture and copy the background layer and use Screen as the layer mode rather than doing a Gamma change.  You might get more out of the image...Or try shadow and highlight adjustment...Lab can be effective but it is not the only tool in the box...


BTW- I did an interesting experiment or demonstration.  I took an image and put in hue clocks.  I went into HSL color space and watched the hands-  cool.  Just as you would expect.  the hands were not totally uniform but it was very interesting.


The calculation difference is certainly something to keep in mind...you will sometimes get that in LAB when you go back to RGB if you have over saturated the image.


My misconception was that pinning the skin would solve the problem.  If it was too far gone, that is not correct.  My problem was not fixing the saturation.  (in general)


I would say that If you want to pin skin in RGB you need to have a reasonable correction to the image for color cast first otherwise you are betting the image on a highly variable piece of the image.  Skin Pins can be held to a smaller adjustment in RGB after you have a reasonable correction.  Kind of like the image of the baby...Correct the image for gross color casts first then fix the skin tones...that is how I approach it anyway...

Greg

sjordan93436
Posts: 462
Joined: Sun Jul 11, 2010 4:23 am
Contact:

Postby sjordan93436 » Tue Oct 05, 2010 8:09 pm

Thanks, I was trying to restate what I have learned.  I have more to learn.

Re: hue and the clock- that was just an analogy.  The hands of the clock are not accurate for that.  I had some trouble of grasping it until....  I figured out (wow, this is obvious) that a hue clock (hand length and direction) are the same for 0,0,0 and 255,255,255 and all grays in between.  Solving the clock is often the hard part.  Dark and light are easier.

The docs state that wg CMYK (when you have it) is complimentary to RGB with the black added.  Some photos have corrections that work better in CMYK than RGB.  (?) 

ggroess
Posts: 5342
Joined: Wed May 24, 2006 2:15 am
Contact:

Postby ggroess » Tue Oct 05, 2010 8:17 pm

CMYK is used for Offset printing and other commercial work.  We teach it but if you need it you really need to go into it much more than the brief brush up we provide.

CMYK is a subtractive process.  The colors printed on the paper produce the colors and the more ink you allow...Higher numbers...the darker the image.

RGB is additive...the colors are lit from behind.  The more light you give the color the whiter it becomes.  Higher numbers...Lighter image. 

Ok...so that effects the price of tea in china how?? 

CMYK is like a feather duster when compared to the hammer that is LAB and the simple brush that is RGB. 
Each has it's uses but the most control is offered in CMYK. 

So why not use CMYK all the time?

CMYK can have hundreds of variations and profiles.  It is dependent on Paper, Ink and Press.  You can use a standard profile but as soon as you do the printer might not use it and they would convert your CMYK profile to another one.  When that occurs the image color gets smacked around and really poor results can come out of it...Different GCR settings, different color balance.  RGB is fairly forgiving in that regard and most printers use some version of sRGB or Adobe RGB.

Greg

sjordan93436
Posts: 462
Joined: Sun Jul 11, 2010 4:23 am
Contact:

Postby sjordan93436 » Thu Oct 07, 2010 6:19 am

Oops, I lost my screen shot, but another feeble attempt.  I lost a nice sky. I added some mosaic (not good).  Added color- good.  Added contrast.



Attachments
thai-b-jpg
thai-b-jpg (71.77 KiB) Viewed 7518 times

ggroess
Posts: 5342
Joined: Wed May 24, 2006 2:15 am
Contact:

Postby ggroess » Thu Oct 07, 2010 12:14 pm

Orange?? 


Return to “Curvemeister 101 September 2010”

Who is online

Users browsing this forum: No registered users and 23 guests