| Name | visu3d JSON |
| Version |
1.5.3
JSON |
| download |
| home_page | |
| Summary | 3d geometry made easy. |
| upload_time | 2023-07-26 16:06:06 |
| maintainer | |
| docs_url | None |
| author | |
| requires_python | >=3.9 |
| license | |
| keywords |
3d
visu3d
visualization
neural rendering
nerf
|
| VCS |
 |
| bugtrack_url |
|
| requirements |
No requirements were recorded.
|
| Travis-CI |
No Travis.
|
| coveralls test coverage |
No coveralls.
|
# visu3d - 3D geometry made easy
[](https://github.com/google-research/visu3d/actions/workflows/pytest_and_autopublish.yml)
[](https://badge.fury.io/py/visu3d)
[](https://visu3d.readthedocs.io/en/latest/?badge=latest)
`visu3d` is an abstraction layer between Torch/TF/Jax/Numpy and your program.
It provides:
* Standard primitives for 3d geometry (`Ray`, `Camera`, `Transform`,...).
You can combine those standard primitives with your custom ones.
* Everything is trivially visualizable with zero boilerplate. Inspect & debug
camera poses, trajectories,...
* All primitives are
[`dataclass_array`](https://github.com/google-research/dataclass_array),
dataclasses which can be reshaped, sliced,... as if they were numpy arrays.
* Everything is extensible, you can gradually opt-in only for the features you
need, and replace any standard primitive by your custom ones.
### Core features
<section class="zippy">
Everything is a `v3d.DataclassArray`: **dataclass behave like `np.array`** (with
indexing, slicing, shape manipulation, vectorization,...).
```python
# Single ray
ray = v3d.Ray(pos=[0, 0, 0], dir=[1, 1, 1])
assert rays.shape == ()
# Multiple rays batched together
rays = v3d.Ray(pos=np.zeros((B, H, W, 3)), dir=np.ones((B, H, W, 3)))
assert rays.shape == (B, H, W)
rays = rays.reshape('b h w -> b (h w)') # Native `einops` support
top_left_ray = rays[..., 0, 0] # (b, h, w) -> (b,)
rays = rays.flatten()
rays = rays[rays.norm() > 0] # Filter invalid rays
```
</section>
<section class="zippy">
Everything is **visualizable interactively**
Every object has a `.fig` property for interactive visualization:
```python
rays.fig # Plot the rays
```
Display multiple objects together:
```python
v3d.make_fig([cam, rays, point_cloud])
```
Auto-plot figures with Colab magic:
```python
v3d.auto_plot_figs() # Once at the start of the Colab
cam, rays, point_cloud # Tuple auto-displayed without `v3d.make_fig` call
```
</section>
<section class="zippy">
Same code seamlessly **works across Torch, Jax, TensorFlow, Numpy**.
```python
rays = rays.as_jax() # .as_tf(), as_np(), .as_jax()
assert isinstance(rays.pos, jnp.ndarray)
assert rays.xnp is jnp
rays = rays.as_torch()
assert isinstance(rays.pos, torch.Tensor)
assert rays.xnp is torch
```
With native support for auto-diff, `jax.vmap`, `jax.tree_utils`,...
</section>
### Privitives
<section class="zippy">
Common primitives (`Ray`, `Camera`, `Transform`,...), so user can express
intent, instead of math.
```python
H, W = (256, 1024)
cam_spec = v3d.PinholeCamera.from_focal(
resolution=(H, W),
focal_in_px=35.,
)
cam = v3d.Camera.from_look_at(
spec=cam_spec,
pos=[5, 5, 5],
target=[0, 0, 0], # Camera looks at the scene center
)
rays = cam.rays() # Rays in world coordinates
# Project `(*shape, 3)` world coordinates into `(*shape, 2)` pixel coordinates.
px_coords = cam.px_from_world @ point_cloud
```
See [the API](https://github.com/google-research/visu3d/tree/main/visu3d/__init__.py)<!-- {.external} !-->
for a full list of primitive.
</section>
<section class="zippy">
Creating your own primitives is trivial.
Converting any dataclass to dataclass array is trivial:
* Inherit from `v3d.DataclassArray`
* Use
[`etils.array_types`](https://github.com/google/etils/blob/main/etils/array_types/README.md)
to annotate the array fields (or exlicitly use `my_field: Any =
dca.field(shape=, dtype=)` instead of `dataclasses.field`)
```python
from etils.array_types import FloatArray
class MyRay(v3d.DataclassArray):
pos: FloatArray[..., 3]
dir: FloatArray[..., 3]
rays = MyRay(pos=jnp.zeros((H, W, 3)), dir=jnp.ones((H, W, 3)))
assert rays.shape == (H, W)
```
`v3d` makes it easy to opt-in to the feature you need by implementing the
corresponding protocol.
<!-- See [the tutorial]() for more info. -->
</section>
### Documentation
The best way to get started is to try the Colab tutorials (in the
[documentation](https://visu3d.readthedocs.io/)):
* [Intro](https://visu3d.readthedocs.io/en/latest/intro.html) ([Colab](https://colab.research.google.com/github/google-research/visu3d/blob/main/docs/intro.ipynb))
<!-- {.external} !-->
* [Transform](https://visu3d.readthedocs.io/en/latest/transform.html) ([Colab](https://colab.research.google.com/github/google-research/visu3d/blob/main/docs/transform.ipynb))
<!-- {.external} !-->
* [Create your primitives](https://visu3d.readthedocs.io/en/latest/dataclass.html) ([Colab](https://colab.research.google.com/github/google-research/visu3d/blob/main/docs/dataclass.ipynb))
<!-- {.external} !-->
Installation:
```sh
pip install visu3d
```
Usage:
```python
import visu3d as v3d
```
## Installation
```sh
pip install visu3d
```
*This is not an official Google product.*
Raw data
{
"_id": null,
"home_page": "",
"name": "visu3d",
"maintainer": "",
"docs_url": null,
"requires_python": ">=3.9",
"maintainer_email": "",
"keywords": "3d,visu3d,visualization,neural rendering,nerf",
"author": "",
"author_email": "Visu3d team <visu3d@google.com>",
"download_url": "https://files.pythonhosted.org/packages/ab/48/c052e6df66e5079009871f6722ab9b0acccfa66ea5522961ba7dace9c290/visu3d-1.5.3.tar.gz",
"platform": null,
"description": "# visu3d - 3D geometry made easy\n\n[](https://github.com/google-research/visu3d/actions/workflows/pytest_and_autopublish.yml)\n[](https://badge.fury.io/py/visu3d)\n[](https://visu3d.readthedocs.io/en/latest/?badge=latest)\n\n`visu3d` is an abstraction layer between Torch/TF/Jax/Numpy and your program.\nIt provides:\n\n* Standard primitives for 3d geometry (`Ray`, `Camera`, `Transform`,...).\n You can combine those standard primitives with your custom ones.\n* Everything is trivially visualizable with zero boilerplate. Inspect & debug\n camera poses, trajectories,...\n* All primitives are\n [`dataclass_array`](https://github.com/google-research/dataclass_array),\n dataclasses which can be reshaped, sliced,... as if they were numpy arrays.\n* Everything is extensible, you can gradually opt-in only for the features you\n need, and replace any standard primitive by your custom ones.\n\n### Core features\n\n<section class=\"zippy\">\n\nEverything is a `v3d.DataclassArray`: **dataclass behave like `np.array`** (with\nindexing, slicing, shape manipulation, vectorization,...).\n\n```python\n# Single ray\nray = v3d.Ray(pos=[0, 0, 0], dir=[1, 1, 1])\nassert rays.shape == ()\n\n# Multiple rays batched together\nrays = v3d.Ray(pos=np.zeros((B, H, W, 3)), dir=np.ones((B, H, W, 3)))\nassert rays.shape == (B, H, W)\n\nrays = rays.reshape('b h w -> b (h w)') # Native `einops` support\n\ntop_left_ray = rays[..., 0, 0] # (b, h, w) -> (b,)\n\nrays = rays.flatten()\nrays = rays[rays.norm() > 0] # Filter invalid rays\n```\n\n</section>\n<section class=\"zippy\">\n\nEverything is **visualizable interactively**\n\nEvery object has a `.fig` property for interactive visualization:\n\n```python\nrays.fig # Plot the rays\n```\n\nDisplay multiple objects together:\n\n```python\nv3d.make_fig([cam, rays, point_cloud])\n```\n\nAuto-plot figures with Colab magic:\n\n```python\nv3d.auto_plot_figs() # Once at the start of the Colab\n\ncam, rays, point_cloud # Tuple auto-displayed without `v3d.make_fig` call\n```\n\n</section>\n<section class=\"zippy\">\n\nSame code seamlessly **works across Torch, Jax, TensorFlow, Numpy**.\n\n```python\nrays = rays.as_jax() # .as_tf(), as_np(), .as_jax()\nassert isinstance(rays.pos, jnp.ndarray)\nassert rays.xnp is jnp\n\nrays = rays.as_torch()\nassert isinstance(rays.pos, torch.Tensor)\nassert rays.xnp is torch\n```\n\nWith native support for auto-diff, `jax.vmap`, `jax.tree_utils`,...\n\n</section>\n\n### Privitives\n\n<section class=\"zippy\">\n\nCommon primitives (`Ray`, `Camera`, `Transform`,...), so user can express\nintent, instead of math.\n\n```python\nH, W = (256, 1024)\ncam_spec = v3d.PinholeCamera.from_focal(\n resolution=(H, W),\n focal_in_px=35.,\n)\ncam = v3d.Camera.from_look_at(\n spec=cam_spec,\n pos=[5, 5, 5],\n target=[0, 0, 0], # Camera looks at the scene center\n)\n\nrays = cam.rays() # Rays in world coordinates\n\n# Project `(*shape, 3)` world coordinates into `(*shape, 2)` pixel coordinates.\npx_coords = cam.px_from_world @ point_cloud\n```\n\nSee [the API](https://github.com/google-research/visu3d/tree/main/visu3d/__init__.py)<!-- {.external} !-->\nfor a full list of primitive.\n\n</section>\n<section class=\"zippy\">\n\nCreating your own primitives is trivial.\n\nConverting any dataclass to dataclass array is trivial:\n\n* Inherit from `v3d.DataclassArray`\n* Use\n [`etils.array_types`](https://github.com/google/etils/blob/main/etils/array_types/README.md)\n to annotate the array fields (or exlicitly use `my_field: Any =\n dca.field(shape=, dtype=)` instead of `dataclasses.field`)\n\n```python\nfrom etils.array_types import FloatArray\n\n\nclass MyRay(v3d.DataclassArray):\n pos: FloatArray[..., 3]\n dir: FloatArray[..., 3]\n\n\nrays = MyRay(pos=jnp.zeros((H, W, 3)), dir=jnp.ones((H, W, 3)))\nassert rays.shape == (H, W)\n```\n\n`v3d` makes it easy to opt-in to the feature you need by implementing the\ncorresponding protocol.\n\n<!-- See [the tutorial]() for more info. -->\n\n</section>\n\n### Documentation\n\nThe best way to get started is to try the Colab tutorials (in the\n[documentation](https://visu3d.readthedocs.io/)):\n\n* [Intro](https://visu3d.readthedocs.io/en/latest/intro.html) ([Colab](https://colab.research.google.com/github/google-research/visu3d/blob/main/docs/intro.ipynb))\n <!-- {.external} !-->\n* [Transform](https://visu3d.readthedocs.io/en/latest/transform.html) ([Colab](https://colab.research.google.com/github/google-research/visu3d/blob/main/docs/transform.ipynb))\n <!-- {.external} !-->\n* [Create your primitives](https://visu3d.readthedocs.io/en/latest/dataclass.html) ([Colab](https://colab.research.google.com/github/google-research/visu3d/blob/main/docs/dataclass.ipynb))\n <!-- {.external} !-->\n\nInstallation:\n\n```sh\npip install visu3d\n```\n\nUsage:\n\n```python\nimport visu3d as v3d\n```\n\n## Installation\n\n```sh\npip install visu3d\n```\n\n*This is not an official Google product.*\n\n",
"bugtrack_url": null,
"license": "",
"summary": "3d geometry made easy.",
"version": "1.5.3",
"project_urls": {
"changelog": "https://github.com/google-research/visu3d/blob/main/CHANGELOG.md",
"documentation": "https://visu3d.readthedocs.io",
"homepage": "https://github.com/google-research/visu3d",
"repository": "https://github.com/google-research/visu3d"
},
"split_keywords": [
"3d",
"visu3d",
"visualization",
"neural rendering",
"nerf"
],
"urls": [
{
"comment_text": "",
"digests": {
"blake2b_256": "8b33d1a3feedec734e56ebb171c5fdf3ac6d1a509751b67f962b0bf7f6ca6b32",
"md5": "65f8f615e35ce8894b8d18d1dd260654",
"sha256": "16564695d365911ffc9900c8c786d7c91c7abea1fa7907f62553136250da4289"
},
"downloads": -1,
"filename": "visu3d-1.5.3-py3-none-any.whl",
"has_sig": false,
"md5_digest": "65f8f615e35ce8894b8d18d1dd260654",
"packagetype": "bdist_wheel",
"python_version": "py3",
"requires_python": ">=3.9",
"size": 49432,
"upload_time": "2023-07-26T16:06:05",
"upload_time_iso_8601": "2023-07-26T16:06:05.340969Z",
"url": "https://files.pythonhosted.org/packages/8b/33/d1a3feedec734e56ebb171c5fdf3ac6d1a509751b67f962b0bf7f6ca6b32/visu3d-1.5.3-py3-none-any.whl",
"yanked": false,
"yanked_reason": null
},
{
"comment_text": "",
"digests": {
"blake2b_256": "ab48c052e6df66e5079009871f6722ab9b0acccfa66ea5522961ba7dace9c290",
"md5": "5783b272c9e85733f6832dc35a77562e",
"sha256": "8bcf1f935d4ca45becad745edf83840ac544476c7feed047a20db13657ea49f5"
},
"downloads": -1,
"filename": "visu3d-1.5.3.tar.gz",
"has_sig": false,
"md5_digest": "5783b272c9e85733f6832dc35a77562e",
"packagetype": "sdist",
"python_version": "source",
"requires_python": ">=3.9",
"size": 34450,
"upload_time": "2023-07-26T16:06:06",
"upload_time_iso_8601": "2023-07-26T16:06:06.831505Z",
"url": "https://files.pythonhosted.org/packages/ab/48/c052e6df66e5079009871f6722ab9b0acccfa66ea5522961ba7dace9c290/visu3d-1.5.3.tar.gz",
"yanked": false,
"yanked_reason": null
}
],
"upload_time": "2023-07-26 16:06:06",
"github": true,
"gitlab": false,
"bitbucket": false,
"codeberg": false,
"github_user": "google-research",
"github_project": "visu3d",
"travis_ci": false,
"coveralls": false,
"github_actions": true,
"lcname": "visu3d"
}
