PixelsPyramidReader: Set canSeperateSeries for BF 6

ome / omero-romio

Gradle project for the ROMIO library of OMERO

GNU General Public License v2.0

0 stars 13 forks source link

PixelsPyramidReader: Set canSeperateSeries for BF 6 #12

Closed dgault closed 5 years ago

dgault commented 5 years ago

This PR should only be included alongside the upgrade to Bio-Formats 6. It is setting a flag which was introduced as part of changes to MinimalTiffReader in BF 6.0.0 Derived formats such as this reader were affected and updated, see https://github.com/openmicroscopy/bioformats/pull/3202

Without this PR and using BF 6:

Open a multi channel OMERO pyramid and zoom to view the highest resolution
At the highest resolution the channels will appear duplicated
This can be compared with the lower resolutions which should be unaffected

With the PR:

The pyramids should behave and display as expected

jburel commented 5 years ago

I built locally with this PR and https://github.com/ome/omero-model/pull/28

Import a 8x8k jpeg
Go to the highest resolution. Black tile. Not expected. See screenshot

dgault commented 5 years ago

Retested and was able to reproduce the issue which was due to a difference between my working branch that I was testing on and this. With the call to canSeparateSeries after setId all of the channels at the highest resolution of the 8k x 8k image appeared red, with the latest commit moving the call before setId the channels display as expected.

jburel commented 5 years ago

Still black tiles with the latest commit in one part of the image. Basically now the reader assumes that one channel only is active i.e. red channel even if the other channels are active.

sbesson commented 5 years ago

@dgault pushed an extra commit (3fff79b) to add a dependency on the BF 6 API, let Travis CI pass and have this included in the daily merge builds for additional troubleshooting. When omero-model is released with Bio-Formats 6, the additional implementation can be reverted.

jburel commented 5 years ago

Retested this morning on merge-ci The pyramid looks good now at 100% pyramid

sbesson commented 5 years ago

I performed a similar testing using 8kx8k.png and 8kx8.jpg i.e. other source datasets with the same dimensions but different file formats. Same conclusions as https://github.com/ome/omero-romio/pull/12#issuecomment-490368934 i.e. the reading of the generated pyramid seem to be working at all levels with various combination of channels activated.

jburel commented 5 years ago

Travis is green with the new version of omero-common (including BF 6.1) Merging