Hello Graeme: This explanation is fairly clear to me, thanks a lot for your patience. I am eagerly waiting for the next release, which looks like to be a major one. Best, tmwu-----原始郵件----- From: Graeme Gill
Sent: Thursday, April 14, 2011 1:25 PM To: argyllcms@xxxxxxxxxxxxx Subject: [argyllcms] Re: 2D view of icc's boundary is not smooth tmwu wrote:
The reason I asked about the boundary smoothness of the profiler is because I thought that the boundary reflect the limit my printer can print color. So it looks weird to me that certain part of boundary looks erratic. I guess you have already explained why this might happen in you mail, but I have to admit that my knowledge of color management
Hi, the way I would explain it is that creating a model of the device behaviour from the scattered data points of the test chart is a compromise between accurately modelling the data points, and the smoothness of the result. If the device behaviour is fairly "twisted" in some areas of the gamut, then the interpolation model of what is smooth, and the underlying device model become noticeably out of alignment, and the trade-off becomes more severe. So the default stiffness gives a good fit but poor smoothness in some areas. Increasing the stiffness of the model improves smoothness, but at the cost of interpolation accuracy. There are probably ways of improving the trade-off, which I will look into when I get the chance. Graeme Gill.