I think people forget that the renderer itself is doing a lot of interpolation of the image map and general rendering plays a part. Even a billion x billion image map is going to get interpolated on render. ;) There is interpolation, smoothing, and antialiasing that come into play. The most exquisite image map is no better than the renderer and render size in which it is used. And Anton makes a stellar point about mesh resolution as well. So, you can play to the highest end and produce uncompressed 32768x32768 images that no one wants to download because 3GB is just unbelievably unworth the effort. Or you can use the statistical information about render size and image map size and come to a compromise. Unless you are expecting pore-level renders, you should take both extremes and find an image map size and informational detail that will satisfy most conditions. The bell curve says that the guy rendering V3 from two miles away and the other one rendering the hair follicle on her left cheek (take that as you will) are on the extreme ends. Then, as many have pointed out, you have to take the detail of the image map into consideration when compressing. With JPG, the more compression, the 'blurrier' the details (as it works on a neighbor format). If you want higher details, use less compression. But a TIF/BMP is not much better (and I stress this) than a top quality JPG, but the storage/download requirements are worlds apart. What are you really gaining in 10x's the file size, but only 0.5% difference in quality?

C makes it easy to shoot yourself in the foot. C++ makes it harder, but when you do, you blow your whole leg off.

 -- Bjarne Stroustrup

Contact Me | Kuroyume's DevelopmentZone