Skip to content

yaoliUoA/seeing3d

 
 

Repository files navigation

seeing3d

Dependencies

Rendering models

The main script to render your models into 2D images is render.py. The usage syntax is:

render.py [-h] [--mode {rgb,depth,normals}] [--win-size WIN_SIZE]
                 [--focal-length-in FOCAL_LENGTH_IN] [--rho-in RHO_IN]
                 [--phi-deg PHI_DEG] [--theta-delta-deg THETA_DELTA_DEG]
                 [--mkdir] [--clear-cache] [--wrl]
                 MODEL_DIR OUTPUT_DIR

Render 3D models to images

positional arguments:
  MODEL_DIR             Base path of model
  OUTPUT_DIR            Model id/directory

optional arguments:
  -h, --help            show this help message and exit
  --mode {rgb,depth,normals}
  --win-size WIN_SIZE   output window size
  --focal-length-in FOCAL_LENGTH_IN
  --rho-in RHO_IN       rho (distance from camera) in inches
  --phi-deg PHI_DEG     phi/pitch in degrees
  --theta-delta-deg THETA_DELTA_DEG
                        theta/yaw increments in degrees
  --mkdir
  --clear-cache
  --wrl                 Use .wrl model. Note: this has no texture

For example,

~/render.py ~/chairs/1028b32dc1873c2afe26a3ac360dbd4 ~/tmp_renders

would use the default parameters to render 64 RGB views of the model in directory ~/chairs/1028b32dc1873c2afe26a3ac360dbd4 as a sequence of files in tmp_renders. Note that you will see an empty window pop up, which is a side-effect of the rendering process.

For a visualization of the phi, theta and rho parameters see the following figure:

Camera and model

How does rendering work?

Conceptually, we set up the scene (including camera, model and lighting), render the output into a buffer, and save this buffer into a .png file. In practice, this involves setting various obscure options in Panda and extreme care in the format of the Collada model. Unfortunately, libraries to load 3D models tend to not be very robust.

You are encouraged to examine and improve the code.

TODO

  • Warehouse 3D markup changed, breaking scraper.

About

No description, website, or topics provided.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 97.9%
  • GLSL 2.1%