Topic: HSB Color Mode

[Lee Harper]

(because of its general interest, I've added Lee Harper's post to Articles.  We welcome your additions and comments - Mike and Greg)

Hi everyone,

I'm almost at the end of the CM 101 course, and one of the things that I keep wondering about is when HSB should be my 'go to' color mode?

I really like the saturation curve, and it's usually quite obvious to me when the saturation curve would be a good choice, but I'm struggling a bit with the Hue and Brightness curves.

I've two questions. I'm not looking for definitive answers - I'm just interested to hear how everyone else approaches this:

• Why use the Brightness curve instead of the L curve in Lab? Does it have any special properties?
• Why use the Hue curve at all? For the life of me I've no idea how to even begin to approach Hue in HSB.

Greg has been kind enough to give me his opinion about HSB, and it would be great to know how others approach these two questions.

All the best,
Lee.

[sjordan93436]

(I know this is an old topic.)

HSB...  the brightness seems to be very similar to the L of LAB. 

The Saturation is interesting.  It allows you to bump or drop saturation in a range.  I wonder how good that will be.  Decreasing skin saturation might decrease some leaves with the same saturation.  ....  Seems like it needs a mask or selection.

Hue.. Potentially interesting.  You can target a certain hue to move.  Same problem as with saturation.  Other problem is the 0 line.  Red is zero.  Going from 5 to 15 easy.  Going from 5 to 355 with a curve, not so easy.

It seems to me to be specialized with good selections of masking to fix hue/ saturation problems.  Or for changing a broad range of saturation.

I have many photos with skin that is 355.  More magenta than yellow.  Unfortunately that little bump to positive 5 or 10 is not easy.  Unfortunately the hue shift affects lips and hair. 

HSB??  Not mainstream yet.

[Lee Harper]

Hi Steve,
Thanks so much for your thoughts

I'm still in a place where HSB seems interesting, but is yet to prove useful. I'm hopeful that a new Color Correction book can help me turn the corner though. It's called 'The Color Correction Handbook: Professional Techniques for Video and Cinema' by Alexis Van Hurkman (www.peachpit.com/store/product.aspx?isbn=0321713117), and it contains an entire chapter about using HSB for color correction.

From the table of contents:

• HSL Qualification in Theory
• Individual Qualifier Controls
• A Basic HSL Qualification Workflow
• Tips for Using and Optimizing HSL Qualifications
• Different Ways of Using HSL Qualifiers
• Hue Curve Adjustments
• Other Types of HSL Adjustments

I have ordered a copy into the library where I work; I'll report back to everyone anything that I pick up from it...

All the best,
Lee.

[sjordan93436]

BTW- this is so obscure, I do not consider this a bug.  ....  Pinning colors in HSB is (ahem) inconsistent at best.  Why anyone would want to pin is beyond me.  I was messing around.  I guess if you selected a sky (for example) that might make sense.   I might have been the only guy to try this.  It has no utility as far as I can see.

I think from a programming perspective, Mike needs to get the information for the hue clocks.  That information is essentially HSB.

[curvemeister]

 Pinning to a particular HSB value may not be very valuable, but a hue, saturation, or brightness pin might be useful.  Could you describe the behavior?  Maybe it is a bug.

If you like obscurities, HSB is definitely closely related to the hue clock.  The hand position is the hue, and the length of the hand is the saturation.  There is an option, under settings, to use an Lab based hue clock, where the clock hand position and length are based on the a and b values of the color.  There is also another option to offset the hue clock, for people who are used to red being at a different position on the clock.

Even more obscure, when a sample is averaged in HSB space, I actually average the values of the hue clock angle, rather than the hue values themselves.  This avoids the problem of the discontinuity at zero, which is really HSB's big Achilles heel.

HSB was invented by Alvy Ray Smith (Alvy mentions HSB is several articles on his web page), and the best article I know of on the HSB family of color spaces was written by Jacob Russ.

Mike

[curvemeister]

BTW - thanks to Lee for bringing up this interesting topic.  I missed it initially and an glad to participate at this point.

[Greg Groess]

Insert Personal Opinion Here: 
I find HSB more suited for problem solving rather than general color correction. 

The color issues I have with HSB are:
1) I find it better suited for single point color selection.  It is way too easy to introduce a wild color cast.
2) The Zero problem is amplified by showing the color space in CM as a linear line.  HSB really is a "circular" space. The range around zero is tricky at best for corrections.  I always hate having to work on both ends of the linear curve to get the red and magenta right.

I use HSB most in concert with masks for spot or area hue corrections.  It can be a powerful tool but it is on the shelf for the vast majority of my work.  I do consider the Saturation channel once in a while but LAB has always been a more approachable space for saturation work..

I have tried to make it all the way through Jacob's article...way too much information for me. I always pay attention to his writing because  there is always a ton of info in there...reading the instructions for the Jacob's Ladder taught me a few things.  I never adopted the entire process but some of the techniques were very handy...

End Personal Opinion

Greg

[curvemeister]

I agree totally about the ends of the hue curve being a headache - I've sometimes thought of rearranging it to start off as a horizontal line, and connecting the left and right ends together.  In essence the curve would be a donut shape, with the ability to distort it, or move it up and down.

I do find the saturation curve to be dynamite, particularly when you want to add saturation to less intense colors, and leave the already saturated colors alone.  Lab adds saturation to everything, and tends to blow out the colors that are already pretty saturated.

[sjordan93436]

Lee's book is about video.  I do not do video, but it must be the saturation control that is useful to them.  I think that video is sensitive to oversaturation.  HSB could boost lower saturation and (??) lowering over saturation.   It could work with a series of images (frames).  LAB is probably not available to then.

Mike, I agree that HSB mode could be improved and made more useful, BUT on your list of enhancements, ...  it must be near the bottom.   

[Lee Harper]

Hi guys,

I just stumbled upon this: http://documentation.apple.com/en/color/usermanual/index.html#chapter=10%26section=8. It is a webpage explaining how another application (again - video specific) handles HSL/B curves.

This other application's HSL user-interface is great, and the Hue curve is horizontal - as Mike mentioned, and the left-and-right sides of the curve are joined, so the problems with reds that Greg mentions would be avoided.

The reason that I pointed to the book was because HSB is used as a curving space by video professionals, and there is information about how to use these curves. In the photography world there is no information to be had - other than what is on this site. I'm sure that the techniques can be easily applied to still images; I'm just finding that I am needing to look to the video world to learn what the techniques are.

What do you all think about the user-interface for the HSB curves that this other application is using?

Lee.

P.S., Thanks for all the input so far; It will be great to push this a bit further 

[curvemeister]

Once again, a great idea.  Somehow the curve looks more interesting when stretched out horizontally.

[sjordan93436]

I did a couple of simplistic test images.  I took three swatches from CM (flesh, sky, and greens).  I copied and boosted in CM the S of HSB. 

Column 1 original
Column 2 plus 10 S of HSB
Column 3 plus 20 S of HSB
Column 4 plus saturation of Saturation adjustment PS
Column 5 Lab color boost 1.5

I have one copy with RGB numbers.  One with HSB numbers.

Consider this a learning exercise for me. 

Interesting differences.  Not all saturation systems are equal.  HSB keeps the H and B constant.  In this instance, it kept the dominant color at the same RGB value and reduced the others.  (If one color stays the same and the two others are reduced, why does brightness or luminance stay the same)

PS saturation not in favor with experts.  (For me, I ignore it because it has never helped.)  It increases the dominant color. 

LAB color boost shifted the hue of the "sky".   The R value was substantially reduced. 

[curvemeister]

Thanks for a great set of questions and observations.  Here's my quick response to your question about brightness staying the same when the channel values change, and a couple of things that might also be of interest.

HSB brightness is defined as the max value of the three RGB channels, expressed as a percentage.  So for example, RGB(0,0,255), RGB(255,255,0), and RGB(255,255,255) are all the same 100% brightness, even though to the eye they are vastly different. 

The various color spaces do indeed handle saturation increases differently, since the representation of colors as channel values determines what can be done mathematically.

A couple of other points:

Curvemeister has an option to use Lab for the hue clock.  You may want to experiment with it, but the main reason I seldom use it is that pure colors will not necessarily point to their correct letter on the hue clock.  By default, pure red will point straight up for both HSB and Lab.  You can adjust this, or even reverse the direction of the hue clock by specifying a negative hue clock offset.

HSB's saturation curve is probably the most useful feature of that color space, since it allows bumping color in less saturated objects, without affecting objects that already have plenty of color.

The Brightness curve of HSB is very similar to the composite RGB curve, with the interesting difference that dark and light objects behave slightly differently between the two.

[Lee Harper]

HSB's saturation curve is probably the most useful feature of that color space, since it allows bumping color in less saturated objects, without affecting objects that already have plenty of color.

The Vibrance slider in Adobe Camera Raw (and many other Raw Conversion utilities) and Photoshop does much the same job - with the additional benefit of its partial protection of skin tones.

From my point-of-view, the Hue curve remains the most interesting; although as has been mentioned, it is/will be difficult to take advantage of until such a time as the curve is presented horizontally, rather than diagonally - I'm hopeful that to do so would not be very difficult to code (there could perhaps be an option within the Settings window for users who prefer the diagonal UI for the Hue curve, if anyone would be likely to protest the change).

Thank you Steve and Mike for the great information.

Lee.

[Lee Harper]

Hi everyone

I discovered a color correction application app today with quite a nice HSB curves user interface (see attached); the app is called Color Finesse.

The Color Finesse UI features horizontal curves, and backgrounds that make adjustments more intuitive. I think that I prefer the style of the Apple Color UI that I linked to in a previous post - but I'm interested to know what everyone else likes. The aspect of this UI that I'm not keen on is how colorful the backgrounds are (I prefer monochrome interfaces - they are less distracting when I'm correcting); a compromise would be for the curve backgrounds to appear monochrome until the user mouses onto the UI (at which point they could light up).

Just food for thought...

All best,
Lee.