Closed pawlowska closed 5 years ago
TeraStitcher can read TIFF multipage which I think should be fully compatible with OME TIFF. Could you please send me a couple of tiles in OME TIFF format so that I can test which is the best way to read them with TeraStitcher? Thanks
-- Giulio
Il giorno mar 15 gen 2019 alle ore 12:16 Monika Pawlowska < notifications@github.com> ha scritto:
Is it possible to directly load ome tiff files as generated by Micromanager directly into terastitcher? Currently I am converting them into sequences of 2D tif files with Fiji which takes a long time and requires double disk space during processing.
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/abria/TeraStitcher/issues/44, or mute the thread https://github.com/notifications/unsubscribe-auth/AIdv1gmBjb_tMLb0olNvEd2DWXpANdZXks5vDbiHgaJpZM4aAwPp .
--
Giulio Iannello Preside della Facolta' Dipartimentale di Ingegneria Universita' Campus Bio-Medico di Roma v. Alvaro del Portillo, 21 00128 Roma, Italy
Tel: +39-06-22541-9602 E-mail: g.iannello@unicampus.it Fax: +39-06-22541-9609 URL: https://scholar.google.it/citations?user=L-UJxIgAAAAJ
https://drive.google.com/file/d/12lRSec7KFarXDjxmmaLjj1BdW4mV11w-/view?usp=sharing this is a dataset taken with Micromanager2, consisting of 2 tiles. The metadata is embedded in the first file.
Monika, Terastitcher can read with no problem the TIFF you sent to me. You have simpy to change in the xml_import.xml file as follows.
Change the line:
Dear Giulio,
this is good news. But I do not understand your instructions. I use terastitcher to generate the xml file, using
terastitcher --import
but if I try to run the script with
terastitcher --import --volin="%CD%" --ref1=-2 --ref2=1 --ref3=3 --vxl1=1.45 --vxl2=1.45 --vxl3=5 --projout=xml_import --volin_plugin=TiledXY|3Dseries
then I get an error "Unable to find stacks in a given directory".
I also don't see any attachement in your message.
thank you for your help. I made progress. I was able to edit the XML file and stitch my two-tile dataset correctly. However, I have some questions:
-if using my own xml_import file, do I still need to use folder hierarchy and change filenames? Or can I encode my own filenames (ending with Pos00, Pos01 etc) in the XML?
-is editing xml the only way to import this data? with terastitcher --import
I'm still not managing
Il giorno lun 21 gen 2019 alle ore 10:53 Monika Pawlowska < notifications@github.com> ha scritto:
thank you for your help. I made progress. I was able to edit the XML file and stitch my two-tile dataset correctly. However, I have some questions: -if using my own xml_import file, do I still need to use folder hierarchy and change filenames? Or can I encode my own filenames (ending with Pos00, Pos01 etc) in the XML?
as it is explained in the guides you can download from GitHub, you can import datasets in two ways: either you structure the dataset into the two-level hierarchy of folders and then delegate to TeraStitcher the creation of the xml, or you generate your own xml conformant to the structure accepted by TeraStitcher and then proceed directly to the displacement computation step. Actually, for new users the latter alternative is probably simpler and even more powerful. In particular, it is the only choice to import multi-channel datasets where channels are stored in separate files (I do not know if this option is of interest for you).
-is editing xml the only way to import this data? with terastitcher --import
I'm still not managing
of course not, if you generate your own xml you have only to specify the right input plugin to be used.
If you give me an idea of your typical acquisition pipeline I can give you suggestions about how to fully exploit TeraStitcher capapibilities, including the ability to generate the stitched image is several different formats.
I do not remember if you already use parallelization to speedup the stitching process. If you have not explored these possibilities and you are interested, let me know (especially if you can use e CUDA enabled device).
Best,
-- Giulio
Giulio Iannello Preside della Facolta' Dipartimentale di Ingegneria Universita' Campus Bio-Medico di Roma v. Alvaro del Portillo, 21 00128 Roma, Italy
Tel: +39-06-22541-9602 E-mail: g.iannello@unicampus.it Fax: +39-06-22541-9609 URL: https://scholar.google.it/citations?user=L-UJxIgAAAAJ
I found the error in my batch script. I was using:
terastitcher --import --volin_plugin=TiledXY|3Dseries
and the correct and working version contains quotation marks:
terastitcher --import --volin_plugin="TiledXY|3Dseries"
. So I hope I can modify the output format in the same way.
As far as CUDA and parallelization are concerned, I still do not manage to get it to work, ie there is no speed up and Task Manager of Windows shows 0% of graphic card activity. I'll try running it with the nvidia profiler as suggested in one of your previous emails and I'll let you know.
Ok, I have progress! I am now managing to stitch ome tiff directly, but I think there might be a bug. I have a dataset with 555 slices. If I set --subvoldim
to 550, everything works. If I set it to 560, terastitcher crasches. With 2D data it was not a problem.
Monika, I have tested with a multipage tiff (non ome-tiff) dataset (500 planes, --subvoldim=550), with your small dataset (41 planes, --subvoldim=50) and with another one-tiff dataset (86 slices, --subvoldim=90). All they worked fine. I also checked the code and it seems all ok. Hence the problem is not trivial. As usual I am interested to find the problem.
Could you please:
Best.
-- Giulio
Il giorno gio 24 gen 2019 alle ore 11:25 Monika Pawlowska < notifications@github.com> ha scritto:
Ok, I have progress! I am now managing to stitch ome tiff directly, but I think there might be a bug. I have a dataset with 555 slices. If I set --subvoldim to 550, everything works. If I set it to 560, terastitcher crasches. With 2D data it was not a problem. [image: error] https://user-images.githubusercontent.com/11158300/51671961-9c70be00-1fca-11e9-932c-a09bb2edbca2.PNG
— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/abria/TeraStitcher/issues/44#issuecomment-457146326, or mute the thread https://github.com/notifications/unsubscribe-auth/AIdv1jZJm7FZDqt9GjshDDfn2Hv_Ju5Qks5vGYoUgaJpZM4aAwPp .
--
Giulio Iannello Preside della Facolta' Dipartimentale di Ingegneria Universita' Campus Bio-Medico di Roma v. Alvaro del Portillo, 21 00128 Roma, Italy
Tel: +39-06-22541-9602 E-mail: g.iannello@unicampus.it Fax: +39-06-22541-9609 URL: https://scholar.google.it/citations?user=L-UJxIgAAAAJ
I'll first make sure that I can reproduce this on another dataset. On the small test dataset I also didn't see this problem, so it is a bit confusing.
Is it possible to directly load ome tiff files as generated by Micromanager directly into terastitcher? Currently I am converting them into sequences of 2D tif files with Fiji which takes a long time and requires double disk space during processing.