Kort, Eric
2004-Jan-12 04:42 UTC
[R] Wrote new tiff package, would like big-endian tiffs for testing
Over the years I have noted a couple of people interested in an R package that will read tiff files. There were some valid arguments against (e.g. R is not an image editting suite, libtiff does not support all tiff files, etc.). True enough, when I want to work with family photos, I use Gimp. But in my research I use tiffs as data sets in the context of computational biology and molecular epidemiology. The results of my analyses in this case will not be mailed to family members, but used in subsequent statistical analyses. For this reason, I like to be able to read my images from within R. So, with that long-winded introduction, I am writing to r-help today because I have implemented a package that reads tiffs and returns the image data as a pixmap object (with additional slots for the original raw data that has NOT be rescaled to 0-1 occurs in the pixmap data slots for plotting). I based the package on libtiff, not because it is perfect, but because it is efficient and will allow access to a broad, even if not comprehensive, range of tiff images. (By the way, one problem with libtiff is that its TIFFReadRGBAImage function truncates 16 bit images to 8 bits...my library circumvents this path for 16bit images, thus maintaining the full 16bit data--at least for non-tiled images). What I am in need in, however, are "big-endian" (byte order=MM), 16-bit tiff files to test my package against. Working exclusively with intel processors, I only have little endian tiffs (and the test images that come with libtiff all seem to be little endian as well). Does anyone have a couple such images you can send me? If any of you are interested in testing out my (still pretty alpha) package, please feel free to contact me. While at the moment it only reads tiffs, I plan to add tiff writing functionality soon (writing a tiff is, after all, much simpler than reading one because of all the variations you must be prepared for in reading). Sorry for the length. Thanks for the help. Eric Kort eric.kort at vai.org This email message, including any attachments, is for the sole use of the intended recipient(s) and may contain confidential information. Any unauthorized review, use, disclosure or distribution is prohibited. If you are not the intended recipient(s) please contact the sender by reply email and destroy all copies of the original message. Thank you.
Hisaji Ono
2004-Jan-12 05:01 UTC
[R] Wrote new tiff package, would like big-endian tiffs for testing
Hi. Do you know RGDAL(http://rgdal.sourceforge.net/) which can read Tiff files?> So, with that long-winded introduction, I am writing to r-help todaybecause I have implemented a package that reads tiffs and returns the image data as a pixmap object (with additional slots for the original raw data that has NOT be rescaled to 0-1 occurs in the pixmap data slots for plotting). I based the package on libtiff, not because it is perfect, but because it is efficient and will allow access to a broad, even if not comprehensive, range of tiff images. (By the way, one problem with libtiff is that its TIFFReadRGBAImage function truncates 16 bit images to 8 bits...my library circumvents this path for 16bit images, thus maintaining the full 16bit data--at least for non-tiled images).
Kort, Eric
2004-Jan-12 12:45 UTC
[R] Wrote new tiff package, would like big-endian tiffs for testing
> -----Original Message----- > From: Hisaji Ono [mailto:hi_ono2001 at ybb.ne.jp] > > > Hi. > > Do you know RGDAL(http://rgdal.sourceforge.net/) which can > read Tiff files?Yes. Thanks for mentioning this excellent alternative. This may be a fine solution for some/many/most/nearly all people. However, I still think it would be elegant for those not doing geospatial research to have a stand alone tiff package, and one that does not require GDAL libraries, etc, to compile and install. For one thing, since my embryonic package is focused solely on TIFFs, it allows close attention to be paid to things like bit preservation (at least in theory). I would bet $1 U.S. that this is an attractive solution for at least 3 people in the known universe. I only hope all three of those people aren't me. But, obviously, to each his/her own. -Eric> > > So, with that long-winded introduction, I am writing to r-help today > because I have implemented a package that reads tiffs and > returns the image > data as a pixmap object (with additional slots for the > original raw data > that has NOT be rescaled to 0-1 occurs in the pixmap data slots for > plotting). I based the package on libtiff, not because it is > perfect, but > because it is efficient and will allow access to a broad, even if not > comprehensive, range of tiff images. (By the way, one problem > with libtiff > is that its TIFFReadRGBAImage function truncates 16 bit images to 8 > bits...my library circumvents this path for 16bit images, > thus maintaining > the full 16bit data--at least for non-tiled images). > >This email message, including any attachments, is for the so...{{dropped}}