Reading from N5

[d-v-b]

I have volumes stored in N5 format, and I am serving them via a local http server (using serve). I am also running neuroglancer via npm run dev-server. In neuroglancer, I tried setting the source source to something like n5://http://localhost:5000/$CONTAINER.n5/$DATASET, where $CONTAINER and $DATASET are the names of the n5 container and the dataset I want to view, but this doesn't work. Specifically, I get Fetching "" resulted in HTTP error 0. How should I get this to work?

[jbms]

The current n5 support is rather limited. It only supports 3d volumes in the format described here:
https://github.com/saalfeldlab/n5-viewer

Your data source URL then needs to be of the form:

n5://http://localhost:5000/CONTAINER.n5/DATASET/c0

CONTAINER.n5/DATASET/attributes.json should contain something like:

{
"scales":[[1,1,1]],
"pixelResolution":{"unit":"um","dimensions":[0.097,0.097,0.18]}
}

CONTAINER.n5/DATASET/c0/attributes.json should not exist (or will be ignored if it does),

and

CONTAINER.n5/DATASET/c0/s0/attributes.json should be a normal n5 dataset attributes.json file (with raw or gzip compression).

The good news is that in a local branch I have already implemented better support for n5 that does not require such a specific directory structure, and I will be pushing it out soon. It is part of a large change that also adds support for 4-d, etc. volumes, but is not quite ready yet.

[d-v-b]

Thanks, this was very helpful -- I ultimately got it working on test data (once I disabled adblock). Re: the local branch, do you have an idea of a) how the required N5 directory structure will be different, and b) when that branch will be visible online?

[jbms]

I just pushed the work in progress to the ndims branch, so you can play with it if you like.

It is still a work in progress, and various things still don't work --- e.g. annotations don't work at the moment. However, N5 support basically works, though there may be bugs.

It should work with:

• plain single resolution datasets
• multi-resolution datasets using s0, s1, ... directories with n5-viewer-style (https://github.com/saalfeldlab/n5-viewer) attributes like "scales":[[1,1,1],[2,2,1],[4,4,2],[8,8,4],[16,16,9],[32,32,17]],
  "pixelResolution":{"unit":"um","dimensions":[0.097,0.097,0.18]},
• multi-resolution datasets using s0, s1, ... directories with bigdataviewer-style (https://github.com/bigdataviewer/bigdataviewer-core/blob/master/BDV%20N5%20format.md) attributes: "downsamplingFactors" : [[1,1,1], [2,2,1], ...], "resolution": [4, 4, 40]

Additionally you can specify attributes: "axes": ["x", "y", "z", "t"] to specify the axis names, and "units": ["um", "um", "um", "s"] to specify the units for each dimension.

Neuroglancer now aggregates attributes from all parent directories as well, so the cN directories are no longer necessary or treated specially.

[jbms]

Also here is an example that will work with a locally-running dev server:

http://localhost:8080/#!%7B%22dimensions%22:%7B%22x%22:%5B9.700000000000001e-8%2C%22m%22%5D%2C%22y%22:%5B9.700000000000001e-8%2C%22m%22%5D%2C%22z%22:%5B1.8e-7%2C%22m%22%5D%7D%2C%22position%22:%5B7534.7314453125%2C13950.4384765625%2C3323.5%5D%2C%22crossSectionScale%22:0.7672059499758557%2C%22projectionScale%22:32768%2C%22layers%22:%5B%7B%22source%22:%22n5://gs://neuroglancer-exllsm-data/m8-10_Sample12/n5/c0%22%2C%22type%22:%22image%22%2C%22blend%22:%22default%22%2C%22shader%22:%22#uicontrol%20vec3%20color%20color%28default=%5C%22red%5C%22%29%5Cn#uicontrol%20float%20brightness%20slider%28min=-10%2C%20max=10%2C%20step=0.001%29%5Cn#uicontrol%20float%20contrast%20slider%28min=-20%2C%20max=20%2C%20step=0.01%29%5Cnvoid%20main%28%29%20%7B%5Cn%20%20emitRGB%28color%20%2A%20%28toNormalized%28getDataValue%28%29%29%20+%20brightness%29%20%2A%20exp%28contrast%29%29%3B%5Cn%7D%5Cn%22%2C%22shaderControls%22:%7B%22brightness%22:-0.006%2C%22contrast%22:3.27%7D%2C%22name%22:%22c0%22%7D%2C%7B%22source%22:%22n5://gs://neuroglancer-exllsm-data/m8-10_Sample12/n5/c1%22%2C%22type%22:%22image%22%2C%22blend%22:%22default%22%2C%22shader%22:%22#uicontrol%20vec3%20color%20color%28default=%5C%22red%5C%22%29%5Cn#uicontrol%20float%20brightness%20slider%28min=-10%2C%20max=10%2C%20step=0.001%29%5Cn#uicontrol%20float%20contrast%20slider%28min=-20%2C%20max=20%2C%20step=0.01%29%5Cnvoid%20main%28%29%20%7B%5Cn%20%20emitRGB%28color%20%2A%20%28toNormalized%28getDataValue%28%29%29%20+%20brightness%29%20%2A%20exp%28contrast%29%29%3B%5Cn%7D%5Cn%22%2C%22shaderControls%22:%7B%22color%22:%22#00ff00%22%2C%22brightness%22:-0.006%2C%22contrast%22:3.4%7D%2C%22name%22:%22c1%22%7D%5D%2C%22showSlices%22:false%2C%22selectedLayer%22:%7B%22layer%22:%22c1%22%2C%22visible%22:true%2C%22size%22:290%7D%2C%22layout%22:%22xy%22%2C%22statistics%22:%7B%22size%22:372%7D%7D

[jbms]

The ndims branch has now been merged, so I believe this issue is resolved.