Lorenzo Ridolfi wrote:
I'm building a profile for an Epson Pro 3800 and Ultrasmooth paper. I already made one with PM 5.0 and I'd like to use this profile as an input to targen adaptive algorithm. The normal advice is use -A 0.5. But if I'm using a good profile as input, a higher value is recommended ?
No, it's not clear that it helps (but then it's not clear that the adaptive algorithm helps either.) Higher values of adaptation make the results more "fragile" - you really want some minimal density of sampling, and a value of 0.5 or so establishes that.
Another question I have is how many patches should I use? I'm thinking about 1500. Any experience in this particular scenario ?
It depends on what your doing, and what you're after. Using more points is going to have far more effect than switching from the default point distribution algorithm to adaptive. Presumably this is an output profile, and I'm guessing you're running in RGB mode. 1500 is probably not bad for RGB, but more is pretty much always better. There's no point using small increments though, you probably want to increase the number of patches by 50-100% to expect to see any noticeable difference, but the overall accuracy will always depend on the weakest link (ie. input profile, viewing conditions, etc., if the output profile is pretty good). Graeme Gill.