Getting Started

Before starting with this, follow the installation guidelines to install the bpycv3d package: Installation

Hello World Script

Simple script to export diffuse properties of the scene from the scene’s default camera

import bpy
from bpycv3d import DataCapture, CameraRandomizer

dc_config = {
    "engine": "blender_eevee",
    "passes": [
        "diffuse_color", "diffuse_indirect", "diffuse_direct"
    ]
}
dc = DataCapture.DataCapture(dc_config)
dc.render(f"frame")

This would output two images - one MultiLayerEXR and another preview image of the complete render pipeline.

Camera Randomization Starter

Now, to run any camera randomization strategy, we need curves to ensure uniformly sampled views around an object.

• Active the Curve Plugin: Edit > Preferences > Add-ons

• Search for Curve and activate the add-on Curve Tools

• In the 3d Viewport add a new curve object and structure it however you would like to within the scene.

Below, in the follow_path under cfp_config set it to the name of the curve you just created.

The target similarly should be set to the name of the object that you want the camera to be focussed on. You could use an empty object here too.

Camera Randomization added to our hello world script above

import bpy
from bpycv3d import DataCapture, CameraRandomizer

dc_config = {
    "engine": "cycles",
    "passes": [
        "glossy_color", "glossy_indirect", "glossy_direct"
    ]
}
dc = DataCapture.DataCapture(dc_config)

cfp_config = {
    "follow_path": ________,
    "target": _______,
    "num_samples": 20
}
cfp = CameraRandomizer.CameraFollowPath(cfp_config)

for i in range(cfp_config["num_samples"]):
    dc.render(f"frame_{i}")
    cfp.step()

This script would now render 20 images along the path with the camera looking at the target object both as MultiLayerEXR with all the passes required as well as a preview image.