Flesh out documentation for meshing

README.md

@@ -2,4 +2,47 @@
 
 flyemflows
 ==========
+Welcome to flyemflows, the center for distributed data processing for the FlyEM team at HHMI's Janelia Research Campus. This repo contains a suite of tools for processing large-scale volumetric image data using Dask. flyemflows operates on workflows defined in YAML documents, and those workflows can be run locally or on a cluster. Define your YAML file, launch your workflow, and you're off to the races!
+
+Installation
+==========
+The flyemflows toolkit is available via the FlyEM team's conda channel, flyem-forge. To install flyemflows and get started, you can run a conda install from within your conda environment of choice. Here, we rely on the community-supported conda-forge channel to fill in any additional dependency needs.
+
+`conda install -c flyem-forge -c conda-forge flyemflows`
+
+Workflows?
+==========
+Yes, workflows! These are defined sequences of operations that are useful for all* of your volumetric image processing needs. Do you want to see all of the beautiful workflows available to you? Of course you do! Issue the following command, leveraging the ever-so-useful `launchflow` executable. This command will take a few seconds, as launchflow currently imports its libraries each time it is called.
+
+`launchflow -w`
+
+Running a Workflow
+==========
+The command above will display the built-in workflows, and it will also describe the process for using third-party workflows. Want to use a given workflow? This can be done by pointing `launchflow` at a template directory containing a workflow.yaml document. Once you know which workflow you want to run, you can use the following command to get an template directory named `mesh_template` that will contain the scaffolding of the YAML required for running the `CreateMeshes` workflow.
+
+`mkdir mesh_template && launchflow -y CreateMeshes > mesh_template/workflow.yaml`
+
+You can also get a more detailed YAML file by running eg `launchflow -v CreateMeshes` for the verbose version. Once you specify the details required of your workflow (input data settings, output data settings, workflow-specific parameters, etc), you can run the workflow! In order to run successfully, you must at a minimum set values for parameters where the value is `{{NO_DEFAULT}}`. Note that the data sources (input/output) should only have value.
+
+To run our example workflow with 10 workers, we would simply issue the following command.
+
+`launchflow -n 10 mesh_template`
+
+This will create a run directory based on the template with a timestamp appended to the name. That run directory will contain information about that run of the workflow, and you can reuse the template directory at a later date to run the job again.
+
+Congratulations, you've run flyemflows!
+
+Running Locally
+==========
+While leveraging cluster resources can be invaluable for large-scale data processing, it is often easier to get off the ground by running locally. To use local resources, simply set the value of the `cluster-type` field to `local-cluster`. You can still pull/push data from/to remote resources, but the compute itself will be local in this case.
+
+Data Sources
+==========
+
+Downsampling
+==========
+
+Mesh Generation
+==========
+The CreateMeshes workflow provides a convenient entrypoint for mesh generation, the manufacture of approximations the surfaces of segmented objects. In a volume consisting of contiguous labels each comprising a distinct object (ie neurons in a brain), one can use the CreateMeshes workflow to generate a triangular mesh for each member of a subset of the labels. The default is all labels EXCEPT 0, which is typically designated as background.