These are chat archives for fiji/fiji

13th
May 2016
Curtis Rueden
@ctrueden
May 13 2016 02:05
@iarganda We presented TWS at AQLM today and people were blown away, super excited about it now. It is a gamechanger for them -- thanks very much for developing it. Expect more questions on the forum soon. ;-)
Ulrik Günther
@skalarproduktraum
May 13 2016 10:56
@ctrueden @hinerm : i’m currently trying to use scifio to read some tiff data into clearvolume, but apparently, i’m running into endianness issues… i tried both reading the whole file with ReaderFilter.stream.read(len) and accessing the bytes via ReaderFilter.openPlane(imgIndex, plane).bytes … is there a better way or am i doing something completely wrong?
Curtis Rueden
@ctrueden
May 13 2016 11:19
@skalarproduktraum Sounds like you are calling it at a really low level...? Do you need that?
The very toplevel API is the open methods of SciJava Common's IOService. But that returns Object. Next level down is open methods of DatasetIOService which return ImgLib2 data structures.
The Dataset you get back can be interrogated the same way as any other Img. You can also unwrap its ImgPlus if needed.
If your data is large you can open it as a cell image using the SCIFIOConfig settings, and it will page in cells on demand.
Curtis Rueden
@ctrueden
May 13 2016 11:36
@skalarproduktraum If you use the toplevel interface and the endianness still seems wrong, let us know—then I'll concede it's probably a bug.
Ulrik Günther
@skalarproduktraum
May 13 2016 11:44
@ctrueden: yes, i actually need it quite low level, primarily for performance reasons
so i’d rather get around the imglib2 data structures
Curtis Rueden
@ctrueden
May 13 2016 11:44
@skalarproduktraum But if the high level API is not performant, we need to fix that.
Fair enough.
You could:
Ulrik Günther
@skalarproduktraum
May 13 2016 11:44
just fyi, i have the same code in c++, using libtiff, there it works
Curtis Rueden
@ctrueden
May 13 2016 11:44
  • try high level anyway momentarily just to see if the endianness bug still occurs—if not, then dig in to find out how the calls are different than what you are doing
  • try Bio-Formats's TiffParser and see if it suffers the same.
Ulrik Günther
@skalarproduktraum
May 13 2016 11:46
do you have a snippet somewhere how to use the inservice/datasetioservice? i got the openPlane, etc code from the scifio tutorials on github
Curtis Rueden
@ctrueden
May 13 2016 11:46
Check this.
In your case, you could either A) reuse the existing SciJava context, if you have one already; or B) use new SCIFIO() instead of new ImageJ() since you are SCIFIO-focused here.
And lastly: I apologize that SCIFIO still sucks so much. The fact is that we have not had time to work on it much over the past 2 years. But it is a critical piece of ImageJ2, so we do need to work on it (eventually ;-)).
Ulrik Günther
@skalarproduktraum
May 13 2016 11:49
no problem, big thanks for the hints :)
Jean-Yves Tinevez
@tinevez
May 13 2016 14:24
@thorstenwagner just contributed a new detector to TrackMate:
https://github.com/thorstenwagner/ij-trackmate-findmaxima
Hadrien Mary
@hadim
May 13 2016 14:29
Nice !