Spacing image information in TIFF and PNG format

[fbudin69500]

It seems that the image spacing information is not saved when using SCIFIO to save these formats.
I have written a small reader and writer that use SCIFIO, for test purposes:
https://github.com/fbudin69500/scifio-spacing-test

When using the default PNG and TIFF writers, the spacing information is saved in the image.
Is this a known issue? I looked in the bioformat library and I had the impression that the spacing was not handled [1] [2].

[1] https://github.com/openmicroscopy/bioformats/blob/v5.0.0/docs/sphinx/formats/png-metadata.txt
[2] https://github.com/openmicroscopy/bioformats/blob/v5.0.0/docs/sphinx/formats/tiff-metadata.txt

Additionally, the spacing information does not seem to be passed from scifio-imageio to bioformat:
https://github.com/scifio/scifio-itk-bridge/blob/master/src/main/java/io/scif/itk/SCIFIOITKBridge.java#L538-L550

@thewtex @ctrueden
Do you think I am doing anything wrong or that this issue would be easy to address?

[ctrueden]

@fbudin69500 Really sorry for dropping the ball on this till now. I am juggling a lot of things, so you are always welcome to ping me at regular intervals if I do not answer in a timely manner.

Upon cursory inspection, I think you are right about the calibrations not being propagated. In the meantime, has anyone at Kitware looked into a fix?

If this is still an outstanding issue, fixing it might be as simple as adding code like the following after the populateMetadata invocation:

meta.setPixelsPhysicalSizeX(FormatTools.getPhysicalSizeX(pSizeX), 0);
meta.setPixelsPhysicalSizeY(FormatTools.getPhysicalSizeY(pSizeY), 0);
meta.setPixelsPhysicalSizeZ(FormatTools.getPhysicalSizeZ(pSizeZ), 0);
meta.setPixelsTimeIncrement(FormatTools.getTime(pSizeT, null), 0);
double startingWavelength = getStartingWavelength(); // what should it be??
for (int c = 0; c < dimc; c++) {
	double wave = startingWavelength + c * pSizeC;
	meta.setChannelEmissionWavelength(FormatTools.getWavelength(wave), 0, c);
}

Note that Bio-Formats now has support for units; the code above assumes the default units, since the write method of SCIFIOITKBridge does not specify them. It is also unclear what the starting wavelength should be; perhaps the bridge needs to be updated to pass that along to the write method.

Lastly, note that the dims parameter of write is currently also unused; but I did not reflect on the ramifications of this.

[fbudin69500]

Thanks for the input! As far as I know, nobody has tried to tackle this problem yet. The lines of code that you are suggesting might be very helpful! (CC @thewtex ).

[thewtex]

Thanks @fbudin69500 for the great report and @ctrueden for the great suggestions. We should have an opportunity to test further in a few months.