why L a b? why not L u v , L c h

[PUNEETRAZDAN]

why do we use only "L a b" values for measuring color when we have other values also like "L u v","L c h" etc. ?

 

[gordo]

why do we use only "L a b" values for measuring color when we have other values also like "L u v","L c h" etc. ?

Putting on my cynic's hat, in the case of CIEL*a*b* being used instead of CIEL*c*h* I can only assume that it was to make the understanding of color models and color management less intuitive and understandable for people who were not engineers, color scientists, or color consultants. The use of terminology to make things obscure has a long history in the graphic arts, perhaps because it usually provides the appearance of competency.

My Monday rant.

Best, gordon p

my print blog here: Quality In Print current topic: Can't go to Print 09?

 

[claude72]

when we have other values also like "L u v","L c h" etc.?

"L u v" is used in television:

"L" = luminance (the grayscale picture)

"u" and "v" are the two components of color in PAL and NTSC color-coding system.

 

[meddington]

why do we use only "L a b" values for measuring color when we have other values also like "L u v","L c h" etc. ?

I would say probably because ISO 13655 (Graphic technology – Spectral measurement and colorimetric
computation for graphic arts images) mandates CIElab, per the CIE (CIE S 014-4/E:2007 / ISO 11664-4), and instrument manufacturers follow this standard. But rather than say its "used because that's what they said we should use", consider color difference equations as well.

I find it easier to interpret individual delta a* and b* values rather than delta C and/or delta h (or uv for that matter). For example, If I get a negative delta b* value, I know the image is too "blue". I often refer to the delta a* and b* value when editing say, a spot color, as it shows me which direction to make a correction. A delta C or h just informs me of the difference, large or small, but not in which direction its off. Also consider that the improved and developing color difference formulae (dE94 and dE2000) are based on CIELab.

Putting on my cynic's hat, in the case of CIEL*a*b* being used instead of CIEL*c*h* I can only assume that it was to make the understanding of color models and color management less intuitive and understandable for people who were not engineers, color scientists, or color consultants. The use of terminology to make things obscure has a long history in the graphic arts, perhaps because it usually provides the appearance of competency?

I agree with Gordo's overlying point that LCH is a much more intuitive metric to work with for visualizing and editing (still crave LCH from Phtoshop, but alas). However, LCH is not directly calculated from the XYZ tri-stimulus values, as is CIELab. In fact it is calculated from the CIELab values themselves. This itself isn't problematic, but note that LCH can also be calculated from CIELuv, and these values likely will not align perfectly with LCH values derived from CIELab, hence two flavors: LCH(ab) and LCH(uv).

Note that most handheld spectrophotometers allow for data to be reported in CIELab, CIELuv or LCH, and that its up to the user to define how they want to measure color, though exchanging it becomes more complex when one deviates from CIELab. Not to say that CIElLab is perfect, or that alternatives aren't being considered, but its currently the standard for Graphic Arts, both officially (Via ISO 13655) and as a de facto.

 

[dimitri]

Very interesting topic! The difference is traced back to the development of color science models.

CIELUV scales its coordinates subtractively, and I think it was the first model that tried to move up from the chromaticity coordinates. CIELAB scales it's coordinates proportionately, or through 'multiplicative normalization' of the XYZ coordinates.

There are studies that show that one is more accurate and other studies that show that the other is more accurate... accurate in terms of matching to human perception. However, "the subtractive transform adaptation transformation incorporated in CIELUV is even further from physiological reality than the wrong von Kreis transform" used in CIELAB (Color Appearance Models, by M. Fairchild). This means that it is less accurate in predicting visual data or calculating color differences, and that changes in luminance do not translate in constant changes in chromaticity; as a result, all the color difference formula are derived from CIELAB.

If CIELUV is indeed preferred for additive, self-luminous color sources, like televisions or CRT monitors, it could be because it is based on a subtractive model itself, so the plus-minus might work better. Not sure on this though since I do not deal with monitors.

Anyhow... CIELAB apparently is better than CIELUV

Luckily, google books has the section in question online: Color Appearance Models - Google Books