[ome-users] converting big ndpi-files

[MEYENHOFER Felix]

Hi Björn

thanks for the reply.


> On 19 Sep 2015, at 13:20 , Björn Quast <bquast at evolution.uni-bonn.de> wrote:
> 
> Hi Felix,
> 
> its only a guess, but it might be due to the kind of rgb-value storage in your 
> ndpi-files. Is this is a version of a tif container? So the rgb values might be 
> stored as Chunky format: RGBRGBRGB and so on or as Planar format: RRRRRR.. 
> GGGGG.. BBBBB…

Well if this would be the case, the attached output file should look differently if it would be planar, wouldn’t it? Since only one color is actually used when scanning slides in fluorescence mode, the other colors don’t have significant values. Hence on value would be correct, then the next tow pixels would be black, resulting in fragmented pattern. But instead the image data is somehow replicated in this 3 by 3 grid. I double-checked if I work on the correct series and anyway, if it would be the wrong series (different magnification) then the ImageReader would complain about the image size.
> 
> You might also tell the ImageWriter that the written data are grayscale? 
> Otherwise the copied header cause other programs to interpret them as RGB?

Using the MetadataTools.populateMetadata() method (see original mail) I set sizeC=1.
And most of the metadata would be the same as in the original file excep: 
sizeX
sizeY
sizeC
pixelType

Best, 

Felix