
Thread
2003.12.26 09:36 "Re: [Tiff] problems with custom TIFF", by Andrey Kiselev
Pushkar,
On Thu, Dec 25, 2003 at 08:25:47PM -0600, Pushkar Pradhan wrote:
I'm working on a TIFF created by ERDAS IMAGINE 8.6, a remote sensing/GIS software.
The file info as given by tiffinfo (it warns about a few custom tags that it can't read, these are above 32000):
Image Width: 421 Image Length: 295
Resolution: 1, 1 (unitless)
Bits/Sample: 16
Sample Format: unsigned integerCompression Scheme: None
Photometric Interpretation: min-is-blackExtra Samples: 2<unspecified, unspecified>
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
Orientation: row 0 top, col 0 lhs
Samples/Pixel: 3
^^^^^^^^^^^^^^^^^^^^
These two tags means that you don't have RGB image, but single band image with the two extra channels. It is not clear how to interpret the two external channels, it should be application dependent. SamplesPerPixel tag contains total band number, and ExtraSamples tag specifies how many extra bands in the total number.
I read in the 3 bands using TIFFReadEncodedStrip and write out to a new file (after converting RGB to LHS in IEEE FP format). This is info given by tiffinfo of output file:
Image Width: 421 Image Length: 295
Bits/Sample: 32
Sample Format: IEEE floating pointCompression Scheme: None
Photometric Interpretation: min-is-blackSamples/Pixel: 3
Rows/Strip: 9Planar Configuration: single image plane
However, the software is unable to open this file. I suspect this has got something to do with the Extra Samples: 2<unspecified, unspecified> thing? I found out that this tells whether there is alpha information in the file or not. But can somebody explain to me how to read it? The man page is confusing. Also once I get it what should I do with it when writing out the output file?
Description looks fine. Have you used Imagine to read that file or other software?
Here's my code for more information:
size = TIFFReadEncodedStrip(tif, strip, bufRGB, (tsize_t)-1);
^^^^^^^^^^^^^
Why do you using -1 here? It should be a size of bufRGB in bytes.
Rgb2Lhs(bufRGB, bufRGB+size, bufRGB+(size*2), tags.dType, tags.config,
bufLHS, bufLHS+size, bufLHS+(size*2),
tags.imageLength, tags.rowsPerStrip);
I think you are trying to do th wrong thing here. With the above PalnarConfiguration the values in buffer are pixel interleaved:
RGBRGBRGB....
size = TIFFWriteEncodedStrip(wtif, strip, bufLHS, size*sizeof(float));
^^^^^^^^^^^^^^^^^^
It should be (size/TIFFDataWidth(inputType))*sizeof(float), because TIFFReadEncodedStrip() returns the number of taken bytes, not number of taken samples.
--
Andrey V. Kiselev
Home phone: +7 812 5274898 ICQ# 26871517