Rendering is bad after 5.0.3→5.1.2 upgrade

[pmcarlton]

I have just upgraded from 5.0.3 to 5.1.2.
I had no problems starting OMERO and all the data is in place as it was before.
However, the rendering settings are not working properly:

* I am viewing a DeltaVision image imported as floating-point

* If I set the rendering scale to "Min/Max", it looks relatively fine.

* However, if I *increase* the "Min" level (slider or manually), the smallest-valued areas in the image start to become the brightest. They should be zero/black, since the "Min" level is now above those pixel values, but it looks instead is as if a LUT is being applied setting those pixels to full intensity.

* If I *continue to increase* the "Min" level, the pixel values start to go back down, and with continued increasing, the pixel intensities get stranger and eventually become noise (see attachments).

* Under "Advanced" (in the Rendering Settings in Insight), the Bit Depth is set as 255 — probably not right? Does that mean 8-bit? These should be floating-point images.

So, something strangely specific is wrong…

EDIT 02015-06-02: This scaling/rendering problem does not occur with a UInt16 image which I just imported. So this is specific to floating-point images after the 5.0.3→5.1.2 upgrade (before the upgrade, there were no problems with either floating-point or UInt16).

Thanks,
Pete Carlton

Images and `omero admin diagnostics` here: https://dl.dropboxusercontent.com/u/458 ... arlton.zip

[sbesson]

Hi Pete,

thanks for reporting this rendering issue. Would it be possible for you to upload a sample DeltaVision file allowing us to reproduce the problem at https://www.openmicroscopy.org/qa2/qa/upload/?

Best,
Sebastien

[pmcarlton]

Hi Sebastien,
Thanks for the reply; I uploaded a dataset (ID: 11111).

Some image info:

Image Size (X,Y,Sections)............... 1024 1024 204
Data Organization (NZ,NW,NT)............ 51 4 1
Image Sequence.......................... Z -> T -> W
Multi-Resolution Settings............... 1 1
Data Type............................... Floating-point
Start Pixel / Grid Spacing (X,Y,Z)...... 0 0 0 1 1 1
Pixel Size.............................. 0.06402 0.06402 0.2000

Best,
Pete

[sbesson]

Hi Pete,

many thanks for the sample file upload. I was able to duplicate the problem using OMERO 5.1.2 and the OMERO.web client. This hints towards a regression in 5.1.x for this kind of data and/or this format as I was not able to reproduce the issue with OMERO 5.0.8. I opened a ticket corresponding to this issue. Let us know if you would like to be cc'ed on it.

Best,
Sebastien

[pmcarlton]

Thanks a lot for checking it, and yes, please cc me on the ticket.
In the meantime I can work around by reimporting any needed images as UInt16.
—Pete

[jmoore]

As you've hopefully seen, I've CC'd you on the ticket, Pete.

Cheers,
~Josh.

[pmcarlton]

Thanks for fixing this in 5.1.4, everything works for us now.

[kgriffis]

Hi Pete, do you mind if we post your most recent comment on Twitter? Thanks,

Kelli
OME Social Media Assistant