[argyllcms] Re: ImageTarget version 0.1.1 released

Hi :)

Klaus Karcher wrote:
- "Dust and Scratches" had only slight effects on gamut mapping (slightly increased overall saturation, no objectionable clipping). 
Similar results can be obtained with downscaling to approx. 60%.


No real surprise there, given the intention of that particular filter :)
- "Reduce Noise" had nearly no (at least no positive) influence on gamut mapping. The VRML file suggests that "Reduce Noise" deforms the gamut rather than reducing it. viewgam -i confirm this impression: 

Hmmm - interesting.
- Gaussian Blur had enormous influence on gamut. Even very small radii reduce the gamut volume drastically (see attachment "GB_V(r).pdf").
Yup, that confirms the results I've seen in my own experiments - though I've only been judging by eye, not making any formal measurements.
Therefore it's difficult to find the right "dose". Too large radii cause serious clipping (and even r = 1 can be too large). The scaling of the slider in ImageTarget's UI should account for the exponential relation between r an V.
That can be arranged - if indeed we don't decide in the long term to ditch the Gaussian blur in favour of downsampling.
- Median filtering gives reasonable results, but the integer increments are somewhat coarse and not perceptually uniform.
True - though a new filter could be created which has an "Amount" parameter, used in a weighted average between the original and median-filtered pixel.
- Bicubic downsampling gives very good results. It is easy to dose (the scale is roughly proportional to V^(1/3) and therefore perceptually nearly uniform). Moreover tiffgamut benefits from the smaller size of the downscaled images. IMHO it is the best of the tested methods to control the gamut mapping. 

Yup, the speed-increase side-effect is certainly worth pursuing.
- Bicubic downsampling is strongly TRC dependent e.g. the effect of Downscaling to 20% with gamma 1 corresponds roughly to downscaling to 60% with L* TRC.
That's not really a surprise, since neither interpolation nor averaging of gamma-encoded values is strictly correct.
I suppose the "correct" way to approach this would be to use a proper sub-pixel box filter downsampling, operating in Lab space. (I'm a little nervous, though, about the prospect of generating the gamut from Lab-transformed data and then expecting it to "line up" with the original image/colourspace/profile for the icclink call - in theory it should work. Once Graeme's next beta is out (which I think he said includes changes to tiffgamut to allow it to support Lab TIFFs) I shall try it :) ) 


- Gaussian blur is much less TRC dependent.
And that's interesting too - I wonder whether a proper box-filter downsampling would behave more like Gaussian Blur in this regard? 

All the best,
--
Alastair M. Robinson

Other related posts: