| AWARE [SYSTEMS] | Imaging expertise for the Delphi developer | |||||||
![]() |
TIFF and LibTiff Mailing List Archive | |||||||
LibTiff Mailing List
TIFF and LibTiff Mailing List Archive Contact
The TIFF Mailing List Homepage |
Thread2006.01.04 18:57 "Re: Strip Vs Tile", by Chris CoxOn 1/3/06 8:37 AM, "Joris" wrote: > > IMO, what Bob said makes a lot more sense. If you don't need tiles, use > strips. Saving large images (say over 4 kilopixels wide and high), you > need tiles. I should add that for certain color spaces and compression > modes, like black and white G3 or G4 compressed, it is customary to save > a single strip even if the image is somewhat bigger then 4 kilopixels. Note that for huge images, there will be more image coherency in 2D tiles than in 1D strips -- so you may get better compression using tiles. > As to the strip/tile size, these days, I would almost recommend 8 > megabyte rather then 8 kilobyte. I agree. > We've recently learned on the list that > Photoshop's new strategy is to use as large a buffer as is affordable on > the machine in the run-time circumstances (though I'm not sure how > exactly 'affordable' is defined), because it helps getting the best > possible compression ratio and saves IO calls and fragmentation. Unfortunately, "affordable" is a complex measure inside Photoshop based on installed RAM, user memory settings, code size and OS overhead, locked image tiles, OS paging activity (more paging == use less RAM for the app), etc. And even though Photoshop uses planar tiles internally, we default to interleaved strips in TIFF for greater compatibility (we offer the option of planar data in CS2, but we have seen some applications choke on it and warn the user about possible compatibility problems). Chris |
|||||||