docs(README): lead with TestPyPI quantem.widget + Colab; document npy_bin variants

README.md

@@ -10,48 +10,82 @@ pretty_name: quantem-data
 
 # quantem-data
 
-Reference electron-microscopy datasets for [quantem.live](https://github.com/bobleesj/quantem.live).
+Reference electron-microscopy datasets for browsing and learning. Open them in your browser via [quantem.widget](https://github.com/bobleesj/quantem.widget) — **no quantem.live install needed**.
 
-Two buckets:
+Two buckets, that's it:
 
-- **`4dstem/`** — 4D-STEM acquisitions (one folder per dataset: Arina master + data chunks)
-- **`haadf/`** — HAADF survey images
+- **`4dstem/`** — 4D-STEM acquisitions. `_npy_bin*` variants are pre-binned NumPy files for fast workshop / Colab demos; the originals are full Arina h5 bundles.
+- **`haadf/`** — HAADF survey images.
 
-## Download with quantem.live
+## Quick start — Show4DSTEM in Google Colab (or any Jupyter)
+
+Workshop path: just `quantem.widget` from TestPyPI + `huggingface_hub`. No CUDA, no h5, no quantem.live.
+
+```python
+!pip install -q --pre --extra-index-url https://test.pypi.org/simple/ quantem.widget huggingface_hub
+
+import os, json, numpy as np, torch
+from huggingface_hub import snapshot_download
+from quantem.widget import Show4DSTEM
+
+# Pre-binned NumPy bundle — fast download, no decompression needed.
+folder = snapshot_download(
+    "bobleesj/quantem-data",
+    repo_type="dataset",
+    allow_patterns=["4dstem/gold_512_npy_bin8/*"],
+)
+asset = os.path.join(folder, "4dstem", "gold_512_npy_bin8")
+data = np.load(os.path.join(asset, "data.npy"))          # (512, 512, 24, 24) uint16, ~300 MB
+meta = json.load(open(os.path.join(asset, "meta.json"))) # sampling + units
+
+Show4DSTEM(torch.from_numpy(data), sampling=meta["sampling"], units=meta["units"])
+```
+
+Drag the scan cursor in the real-space image; the CBED updates in real time. Chrome works best (WebGPU).
+
+For HAADF use `Show2D` with `quantem.live.io.read_image` (or load the `.emd` directly with `hyperspy`).
+
+## Datasets
+
+| name | kind | shape | dtype | size | use |
+|---|---|---|---|---|---|
+| `4dstem/gold_512_npy_bin8/` | NumPy bundle | (512, 512, 24, 24) | uint16 | ~302 MB | workshop / Colab demo |
+| `4dstem/gold_512_npy_bin4/` | NumPy bundle | (512, 512, 48, 48) | uint16 | ~1.2 GB | sharper version |
+| `4dstem/gold_512/` | Arina h5 (full) | (512, 512, 192, 192) | uint16 | ~5 GB | power user, full data |
+| `4dstem/gold_30mrad1.3mx04` …`09` | Arina h5 (full) | smaller scans | uint16 | ~5 GB each | series demo |
+| `haadf/gold_haadf.emd` | Velox EMD | (4096, 4096) | float32 | a few MB | Show2D |
+
+Each `_npy_*` bundle ships a `meta.json` next to `data.npy`: `shape`, `dtype`, `sampling`, `units`, `bin_factor`, original source.
+
+## Power-user path (full data, GPU decompression)
+
+If you have an NVIDIA GPU and want the full uncooked h5 path, install `quantem.live` (private):
 
 ```python
 from quantem.live import io
-from quantem.widget import Show4DSTEM, Show2D
+from quantem.widget import Show4DSTEM
 import torch
 
-# 4D-STEM: download (cached after first time), bin to fit your GPU, browse
-folder = io.download("gold_512")                       # -> 4dstem/gold_512/
-data = io.load(io.discover_masters(folder)[0], det_bin=2)
-Show4DSTEM(torch.from_dlpack(data.data))
-
-# HAADF image
-Show2D(io.read_image(io.download("gold_haadf")))       # -> haadf/gold_haadf.emd
+folder = io.download("gold_512")                              # → 4dstem/gold_512/
+result = io.load(io.discover_masters(folder)[0], det_bin=2)   # GPU decompression
+Show4DSTEM(torch.from_dlpack(result.data))
 ```
 
-From a terminal: `live data list` · `live data download gold_512`.
+Same data, full Arina h5, decompressed on the GPU. Faster + full resolution.
 
-## Memory (VRAM)
+## Memory (VRAM) for the full h5
 
-The 4D-STEM sets here are 512x512x192x192. Peak VRAM on load is about **1.4x the
-loaded array**, set by `det_bin` (bins the detector on the fly):
+The Arina h5 sets are 512×512×192×192. Peak VRAM is about 1.4× the loaded array; control with `det_bin`:
 
-| det_bin | detector | loaded | peak VRAM | fits 16 GB? |
+| `det_bin` | detector | loaded | peak VRAM | fits 16 GB? |
 |---|---|---|---|---|
-| 1 | 192x192 | 18 GB | ~25 GB | no |
-| **2** | 96x96 | 4.5 GB | **~6.9 GB** | **yes** |
-| 4 | 48x48 | 1.1 GB | ~2.2 GB | yes |
-| 8 | 24x24 | 0.3 GB | ~0.5 GB | yes (good for flipping through many) |
+| 1 | 192×192 | 18 GB | ~25 GB | no |
+| **2** | 96×96 | 4.5 GB | **~6.9 GB** | **yes** |
+| 4 | 48×48 | 1.1 GB | ~2.2 GB | yes |
+| 8 | 24×24 | 0.3 GB | ~0.5 GB | yes (good for flipping through many) |
 
 Show4DSTEM keeps the data on the GPU; cursor interaction is ~0.03 ms/frame.
 
-## Current datasets
-
-- `4dstem/`: gold_512, gold_30mrad1.3mx04, gold_30mrad1.3mx06, gold_30mrad1.3mx07, gold_30mrad1.3mx09
-- `haadf/`: gold_haadf
+## Licence
 
-Run `live data list` for the live set.
+CC-BY-4.0. Cite quantem.live / quantem.widget if you use these in a publication.