Currently there's no clear up-to-date source for the IDIA HDF5 schema, and this is causing confusion.
There's an abandoned Python prototype of this converter, and the repository includes a wiki page documenting the schema as well as a graphviz diagram of the schema. These resources should be reproduced in this repository. I don't think that it makes sense to use a wiki; I would suggest a schema subdirectory, with a markdown file inside next to the diagram source and generated files. The main README can link to it.
I believe that the schema diagram is still up to date (although we should update it to show some 3D mipmaps when that feature is merged). The schema documentation should be updated to remove references to specifics of the prototype implementation (which for example would only generate certain datasets on request, whereas the current implementation automatically generates all datasets that are required for a particular image based on its shape). We should also mention how datasets are chunked, and replace references to non-existent example images (we should perhaps also include small generated example images in this repo, and refer to those instead).
Currently there's no clear up-to-date source for the IDIA HDF5 schema, and this is causing confusion.
There's an abandoned Python prototype of this converter, and the repository includes a wiki page documenting the schema as well as a graphviz diagram of the schema. These resources should be reproduced in this repository. I don't think that it makes sense to use a wiki; I would suggest a
schemasubdirectory, with a markdown file inside next to the diagram source and generated files. The main README can link to it.I believe that the schema diagram is still up to date (although we should update it to show some 3D mipmaps when that feature is merged). The schema documentation should be updated to remove references to specifics of the prototype implementation (which for example would only generate certain datasets on request, whereas the current implementation automatically generates all datasets that are required for a particular image based on its shape). We should also mention how datasets are chunked, and replace references to non-existent example images (we should perhaps also include small generated example images in this repo, and refer to those instead).