{
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "\n# Proto-object in motion\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "collapsed": false
      },
      "outputs": [],
      "source": [
        "import pandas as pd\nimport torch\n\nimport matplotlib.pyplot as plt\nfrom matplotlib import animation\nfrom IPython.display import HTML\n\nfrom deadleaves import (\n    LeafGeometryGenerator,\n    LeafAppearanceSampler,\n    ImageRenderer,\n    LeafTopology,\n)\n\n\nimage_shape = (512, 731)\narea = image_shape[0] * image_shape[1]\n\nshared_params = {\n    \"leaf_shape\": \"circular\",\n    \"shape_param_distributions\": {\n        \"area\": {\"uniform\": {\"low\": area * 0.005, \"high\": area * 0.01}}\n    },\n    \"image_shape\": image_shape,\n}\n\ntable_background, _ = LeafGeometryGenerator(\n    **shared_params,\n    n_sample=200,\n).generate_segmentation()\n\ntable_target, _ = LeafGeometryGenerator(\n    **shared_params,\n    n_sample=10,\n    position_mask={\"shape\": \"circular\", \"params\": {\"area\": area * 0.05}},\n).generate_segmentation()\n\ntable_background = LeafAppearanceSampler(leaf_table=table_background).sample_color(\n    color_param_distributions={\"gray\": {\"normal\": {\"loc\": 0.5, \"scale\": 0.2}}},\n)\n\ntable_target = LeafAppearanceSampler(leaf_table=table_target).sample_color(\n    color_param_distributions={\n        \"R\": {\"uniform\": {\"low\": 0.2, \"high\": 1.0}},\n        \"G\": {\"constant\": {\"value\": 0.0}},\n        \"B\": {\"constant\": {\"value\": 0.0}},\n    },\n)\n\ntable_background[\"type\"] = \"background\"\ntable_target[\"type\"] = \"target\"\n\ntopology = LeafTopology(image_shape=image_shape)\n\ntable_merged = topology.merge_leaf_tables(table_background, table_target)\ntable_merged = LeafTopology.randomize_index(table_merged, seed=42)\n\n\ndef move_leaves_one_step(\n    leaf_table: pd.DataFrame,\n    frame_idx: int,\n    image_shape: tuple[int, int],\n    radius: float = 5.0,\n    target_velocity: tuple[float, float] = (1.0, 0.0),\n    bg_angles: torch.Tensor | None = None,\n    bg_angular_velocities: torch.Tensor | None = None,\n) -> pd.DataFrame:\n    table = leaf_table.copy()\n\n    # Identify background and target leaves\n    bg_mask = table[\"type\"] != \"target\"\n    target_mask = table[\"type\"] == \"target\"\n\n    n_bg = bg_mask.sum()\n    bg_indices = table.index[bg_mask]\n\n    # Initialize angles and angular velocities if not provided\n    if bg_angles is None:\n        bg_angles = torch.distributions.uniform.Uniform(0, 2 * torch.pi).sample((n_bg,))\n    if bg_angular_velocities is None:\n        bg_angular_velocities = torch.distributions.uniform.Uniform(0.02, 0.05).sample(\n            (n_bg,)\n        )\n\n    # Move background leaves\n    for i, idx in enumerate(bg_indices):\n        angle = bg_angles[i] + frame_idx * bg_angular_velocities[i]\n        dx = radius * torch.cos(angle)\n        dy = radius * torch.sin(angle)\n        table.loc[idx, \"x_pos\"] = table.loc[idx, \"x_pos\"] + dx\n        table.loc[idx, \"y_pos\"] = table.loc[idx, \"y_pos\"] + dy\n\n    # Move target leaves linearly\n    table.loc[target_mask, \"x_pos\"] = (\n        leaf_table.loc[target_mask, \"x_pos\"] + frame_idx * target_velocity[0]\n    ) % image_shape[1]\n    table.loc[target_mask, \"y_pos\"] = (\n        leaf_table.loc[target_mask, \"y_pos\"] + frame_idx * target_velocity[1]\n    ) % image_shape[0]\n    return table\n\n\nframes = []\nn_frames = 10\nfor t in range(n_frames):\n    table = move_leaves_one_step(\n        leaf_table=table_merged,\n        image_shape=image_shape,\n        frame_idx=t,\n        radius=5.0,\n        target_velocity=(image_shape[1] / n_frames, 0.0),\n    )\n\n    renderer = ImageRenderer(\n        leaf_table=table,\n        image_shape=image_shape,\n        background_color=torch.tensor(1.0),\n    )\n    image = renderer.render_image()\n    frames.append(image.cpu())\n\nfig, ax = plt.subplots(figsize=(7.31, 5.12))\nim = ax.imshow(frames[0])\nax.axis(\"off\")\nax.set_position((0.0, 0.0, 1.0, 1.0))\n\n\ndef update(i):\n    im.set_data(frames[i])\n    return [im]\n\n\nani = animation.FuncAnimation(\n    fig,\n    update,\n    frames=n_frames,\n    interval=100,\n    blit=True,\n)\n\nplt.close(fig)\nHTML(ani.to_jshtml())"
      ]
    }
  ],
  "metadata": {
    "kernelspec": {
      "display_name": "Python 3",
      "language": "python",
      "name": "python3"
    },
    "language_info": {
      "codemirror_mode": {
        "name": "ipython",
        "version": 3
      },
      "file_extension": ".py",
      "mimetype": "text/x-python",
      "name": "python",
      "nbconvert_exporter": "python",
      "pygments_lexer": "ipython3",
      "version": "3.12.3"
    }
  },
  "nbformat": 4,
  "nbformat_minor": 0
}