darwinjob / pyramidio-bioformats Goto Github PK

https://forum.image.sc/t/jpeg-invalid-scanline-stride-and-restart-interval-and-markers-invalid/63045
ome/bioformats#3773

TIFF images with these properties:

Format: TIFF (Tagged Image File Format)
Mime type: image/tiff
Class: PseudoClass
Geometry: 15184x10224+0+0
Resolution: 1.55039x1.55039
Print size: 9793.68x6594.48
Units: Undefined
Colorspace: sRGB
Type: Palette
Endianess: MSB
Depth: 8/16-bit
Channel depth:
Red: 1-bit
Green: 16-bit
Blue: 1-bit

produce corrupted pyramids. The lowest pyramid level (no scale) is ok but all the others look like this:

After reconverting with IM -define tiff:tile-geometry=256x256 -type TrueColor -depth 8 -compress LZW:

Format: TIFF (Tagged Image File Format)
Mime type: image/tiff
Class: DirectClass
Geometry: 15184x10224+0+0
Resolution: 1.55039x1.55039
Print size: 9793.68x6594.48
Units: Undefined
Colorspace: sRGB
Type: Palette
Base type: TrueColor
Endianess: LSB
Depth: 8-bit
Channel depth:
Red: 1-bit
Green: 8-bit
Blue: 1-bit

pyramids are ok.

Here:
https://github.com/darwinjob/pyramidio-bioformats/blob/master/pyramidio/src/main/java/gov/nist/isg/pyramidio/tools/ImageResizingHelper.java
This causes a number of issues with non RGB888 input images. Current workaround is -rgb parameter.

A set of files having increasing amount of dots inside can show this.

red_v1.png
green_v1.2.png
blue_v1.2.3.png
purple_v1.2.3.4.png

will produce correct folders (.png extension removed, _files appended):

red_v1_files
green_v1.2_files
blue_v1.2.3_files
purple_v1.2.3.4_files

but incorrect descriptors:

red_v1.dzi
green_v1.dzi
blue_v1.2.dzi
purple_v1.2.3.dzi

(from usnistgov#13)

The original pyramidio scaling algorithm is (presumably) the source of the several issues:
#2
#1

Can be fixed by either converting the input to RGB888 on a fly or by improving the scaling algorithm in here:
https://github.com/darwinjob/pyramidio-bioformats/blob/master/pyramidio/src/main/java/gov/nist/isg/pyramidio/tools/ImageResizingHelper.java

java.net.URISyntaxException: Illegal character in opaque part at index 2
Looks like the Windows style file path.

@PolarBean
By default icr is 1 - cache an entire image. It is possible to calculate a suitable icr value based on the heap size and an image width, height and color depth.

related to this stackoverflow issue, you may need to encode the path
https://stackoverflow.com/questions/4992317/illegal-character-in-path-at-index-16

See the below example, it doesnt load at the highest level only when zoomed in:
https://miosdv.apps-dev.hbp.eu/index.html?bucket=https://tar-svc-test.apps.hbp.eu/fakebucket/?url=https://data-proxy.ebrains.eu/api/v1/buckets/ewb-fc6b509d-8398-437e-a86e-627a41f3ed15?prefix=.nesysWorkflowFiles/zippedPyramids/test_T2star

Please see the below link for the original images:
https://data-proxy.ebrains.eu/ewb-fc6b509d-8398-437e-a86e-627a41f3ed15?prefix=.nesysWorkflowFiles%2ForiginalImages%2Ftest_T2star%2F

The default log level is too verbose

Since ome is licensed as GPL https://www.openmicroscopy.org/licensing/, would that necessitate pyramidio-bioformats to also licence under GPL?

Is it possible to run the DZI converter multithreaded on a single image?

TIFFs with alpha channel aka RGBA are not supported at the moment.