Skip to main content

When writing to Geotiff I would like Band interleaved by Pixel (BIP, interleaved RGBRGB) rather than Band Sequential (BSQ, per channel RRGGBB) interleaving but whichever option I use, the Geotiff always saves using Band Sequential interleaving.

I have checked using Photoshop and also GDALinfo and both agree that INTERLEAVE=BAND rather than PIXEL.

Is this a bug/can anyone else replicate this?

Hi @erinjayaustin, May I ask if you happen to see this in a translation where the source data is also GeoTIFF? If so, this could be due to the fact that the reader is reading in a geotiff_interleaving_type format attribute, which stores the original interleaving type of the source data. This attribute currently can not be overwritten by the writer. In the future, we might consider adding a parameter on the writer to allow this overwrite to happen.

 

In the mean time, would you mind trying the following:
  1. expose the geotiff_interleaving_type attribute on the reader Format Attributes tab
  2. use an AttributeRemover to remove said attribute

This should allow the writer parameter setting for interleaving type to be honored.

 

Hope this helps.


Hi @erinjayaustin, May I ask if you happen to see this in a translation where the source data is also GeoTIFF? If so, this could be due to the fact that the reader is reading in a geotiff_interleaving_type format attribute, which stores the original interleaving type of the source data. This attribute currently can not be overwritten by the writer. In the future, we might consider adding a parameter on the writer to allow this overwrite to happen.

 

In the mean time, would you mind trying the following:
  1. expose the geotiff_interleaving_type attribute on the reader Format Attributes tab
  2. use an AttributeRemover to remove said attribute

This should allow the writer parameter setting for interleaving type to be honored.

 

Hope this helps.

Hi @XiaomengAtSafe! Thanks for your answer, it looks like that was the case. Removing the geotiff_interleaving_type has worked - thank you!

 


I just want to point out that in FME 2024.0 this is still happening. At least with the COG writer.

Actually, my case is worst. My source data are JPEG, so I can’t apply the workaround described here.


Reply