Open rmd13 opened 5 years ago
rmd13
commented
5 years ago I trid once using --resolutions =2, and found that file size frop from 700MB to 5MB. I originally thought that it just reduce the X-Y length to(1/2^2) * (1/2^2), i.e.1/16, and got a file size of 700/(16) = 43MB, but when I open it , I found that Z-layers drops heavily from 30 to 6 ( result of (1/2^2))
iannellog
commented
5 years ago Answers:
downsampling at the moment can be a only power of 2
if you are using teraconverter and you have an anisotropic voxel (typically x-y size remarkably smaller than z size) you can add the option --isotropic which applies downsampling to x-y only until an approximately isotropic voxel is obtained. For insance if your voxel size is 0.65x0.65x2.0 (as stored in the xml file) and you include the options: --resolutions=0123 --isotropic you get:
at resolution 0 a voxel of size 0.65x0.65x2.0
at resolution 1 a voxel of size 1.3x1.3x2.0
at resolution 2 a voxel of size 2.6x2.6x4.0
at resolution 3 a voxel of size 5.2x5.2x8.0 the idea is that resolution in x-y decreases, but it is alway higher than that in z
if you have not used --isotropic the image has been downsampled even in Z and since bottom single slices are lost when downsampling is applied you should have got floor(30/2) = 15 slices at resolution 1 and floor(15/2) = 7 slices at resolution 2 (I cannot explain why you get only 6 slices).
I think I have answered all your points.
-- Giulio
Il giorno mer 6 feb 2019 alle ore 12:12 rmd06 notifications@github.com ha scritto:
I trid once using --resolutions =2, and found that file size frop from 700MB to 5MB. I originally thought that it just reduce the X-Y length to(1/2^2) * (1/2^2), i.e.1/16, and got a file size of 700/(16) = 43MB, but when I open it , I found that Z-layers drops heavily from 30 to 6 ( result of (1/2^2))
— 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/47#issuecomment-460986415, or mute the thread https://github.com/notifications/unsubscribe-auth/AIdv1nUqxI8SH3s4QIzKXq4KS-UgbhjVks5vKriagaJpZM4alAOh .
--
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
On wiki of Teraconverter it shows that: --resolutions=
Resolutions indices ( layers l) of the image pyramid to be produced. l must be in [0,10] and 2^l is the corresponding subsampling factor. Default is 0, that means that only the original resolution (subsampling factor 2^0=1) will be produced. For instance: 0123 means pyramid layers 0, 1, 2 and 3 will be produced; 25 means pyramid layers 2 and 5 will be produced; and so on.
How about if I want to save the image as certain sampling ratio such as 1/3 ,1/5,...of original size? And also usually we only want the X-Y to be downsampled, but keep the Z unchanged.