.czi files (Zen) do not open as they should

Hi there
During our course all students with images acquired on a Zeiss microscope had problems opening their images with Fiji.
I have updated fiji and bioformat. I tried to open using the import/bioformat option. But the images just look weird. An image that has 2 channels opens as if it had 6 and the console gives me tons of this message: ‘[WARN] Unknown IlluminationType value ‘Fluorescence’ will be stored as “Other”’.
I see that there was a thread open in August 2019 but the error they get seems different so I opened another topic.
Here is an example of a .czi image. (password LCI)


1 Like

I might be wrong but I think there was a filter applied on the microsope output that is set to be viewed on different channels. I checked the meta data to find that a filterset of 6 filters is applied to the channel as follows,

I do not think so because we had the same problem with images from students working on different Zeiss microscopes.

Hi @Sylvielg,

I will check. In ZEN obviously it has 2 channels and looks OK, but somehow BioFormats “concludes” SizeC=6.

Sebi (from ZEISS)

What ZEN Version and using which camera did you use?

Hi @Sylvielg,

At our intitutue we have a Zeiss AxioObserver D1 equipped with an AxioCam 506 monochrome camera, with an additional Apotome.2 structured illumination module to be optionally included for imaging. We regularly get this warning when opening the images we take, with no additional issues. It is quite posible that this message is unrelated to the issue you report, but rather a minor import issue of bioformats (which would be great if fixed, but is not critical).


Haha, wow. That file causes QuPath to have a screaming fit. Interesting.

I have a suspicion that part of it is due to this, as I have never seen a bit depth of 42 used before:

EDIT: Oh wait, no, I see, all three filters are listed for each channel. Which nicely adds up to 6.

Still, the bit depth of 42 is… interesting.

However, when I used Change Pixel Type in Zen to make it 16 bit, it became openable.

Not 100% sure on how this is all working, just throwing out options and ideas. Time for more caffeine.

Not sure about the Zen Version, but the camera is included in the file.

This indicates 14 Bit RGB Pixel type (3x14=42). Basically 16 Bit RGB. And in this case 2 Channels with 3 components equals 6 channels for some reason (needs some debugging maybe).

1 Like

That’s… a bit disturbing for a 2 channel fluorescent image. Looking at it in Zen, though, you are right. It is saving each fluorescent channel as an RGB image. In fact, it is defaulting to brightfield gamma correction for me.
With one channel
With two

1 Like

In the OME model, SizeC is the total number of all components in all channels. Importing Snap-907.czi with BioFormats correctly reports the two Bgr48 channels (there are two Channel elements with three SamplesPerPixel in the OME-XML) but Fiji apparently does not use that information.


So you acquired 2 channels with an RGB color camera that has a 14 bit range.
And since BioFormat interprets RGB from a CZI as 3 distinct channels it shows 6 channels.

So this is technically correct. What is the application to acquire 2 RGB channels?

1 Like

A colorimetric stain plus a fluorescent counterstain would be a good example of needing a colour camera to capture fluorescent data as well as transmission. Eg beta galactosidase and dapi? Or maybe they can’t stretch to a mono camera?

1 Like

Agreed, but that can be dealt with in processing, at least by changing the pixel type. I guess the main question is can you force the RGB camera to capture as mono.

And after a quick test on an Observer, it looks like you should be able to!

Though I think you would still have to do post processing if you wanted to mix and match, as this setting is at the camera level. It should help for the initial use case presented by @Sylvielg, though, of a dual channel IF image

If one needs to do some postprocessing on a just acquired image directly within an experiment activate the Automate option in Zen Blue on the acquisition tab.
This allows you to plug in a python script directly as an integral part of the experiment to Automate the processing.

1 Like

Just seeing this thread today. As mentioned above by @cgohlke Bio-Formats reads the CZI file as having 2 BGR48 channels. In the case of a BGR pixel type being found it will display each as 3 separate channels which is what is being seen. As such this looks to be as expected for this scenario.

1 Like