Skip to main content

Hey everyone I wasn't sure where to post this as I thought it may be a bug.

I'm running FME Desktop 2016.0.0.0 Build 16167 - WIN64

I have a workbench that takes in multiple rasters, does some reclassifying and then some raster maths by adding them together.

When it gets to the writing stage, the process fails with this error message:

GEOTIFF writer: Failed to process a band with interpretation 'INT32' with the number of bits per cell set to '4'. The band bit depth must be the minimum value of (8, 16, 32) that is greater than or equal to the specified bits per cell. For example: to write 1-bit data the band interpretation should be 8-bit; to write 12-bit data the band interpretation should be 16-bit; to write 19-bit data the band interpretation should be 32-bit

A fatal error has occurred. Check the logfile above for details

I have gone into the advanced parameters of the geotiff writer and set the bits per cell to 32 but it still attempts to write at 4 bit.

I think the bug is because one of the input rasters is 4 bit and FME is maintaining that somewhere. However, that raster does get interpolated to 32 bit before any addition occurs

I have attached a zip containing my workbench and some sample data. Although, please can this data only be used for investigating this problem.

Thank you.

Hi @mattenvsys, in my observation, the GEOTIFF reader for "Opp6.tif" gives "4" to the format attribute named "geotiff_number_of_bits_per_cell", and it seems to cause the error.
If you expose and remove the attribute before writing, the error will not occur. I'm not sure whether it's a kind of bug.


Hi @mattenvsys, in my observation, the GEOTIFF reader for "Opp6.tif" gives "4" to the format attribute named "geotiff_number_of_bits_per_cell", and it seems to cause the error.
If you expose and remove the attribute before writing, the error will not occur. I'm not sure whether it's a kind of bug.

Hi Takashi.

That solution worked for me.

Thank you for your response


Reply