pyvale.render.blender.config module

Blender-specific unified renderer configuration.

class EBlenderEngine(*values)[source]

Bases: Enum

Blender render engines supported by the unified adapter.

CYCLES = 'CYCLES'
EEVEE = 'BLENDER_EEVEE'
WORKBENCH = 'BLENDER_WORKBENCH'
class EBlenderDevice(*values)[source]

Bases: Enum

Cycles compute devices supported by the unified adapter.

CPU = 'CPU'
GPU = 'GPU'
class BlenderConfig(output_dir, engine=EBlenderEngine.CYCLES, device=EBlenderDevice.CPU, samples=2, max_bounces=12, threads=1, render_deformed=False, save_images=False, save_scene=False, seed=0, use_denoising=True, use_adaptive_sampling=True)[source]

Bases: object

Stable Blender controls accepted by the unified adapter.

Parameters:

  • output_dir (pathlib.Path) – Directory used for optional TIFF and Blender-project outputs.

  • engine (EBlenderEngine, optional) – Blender render engine.

  • device (EBlenderDevice, optional) – Cycles compute device. CPU is the reproducible regression default.

  • samples (int, optional) – Per-pixel render samples.

  • max_bounces (int, optional) – Cycles maximum light-bounce count.

  • threads (int, optional) – Blender render worker count.

  • render_deformed (bool, optional) – Render each nodal-displacement frame instead of a static scene.

  • save_images (bool, optional) – Persist TIFFs and return their paths rather than retaining image arrays.

  • save_scene (bool, optional) – Persist the constructed Blender scene as a .blend project file.

  • seed (int, optional) – Fixed Cycles sampling seed used for reproducible rendering.

  • use_denoising (bool, optional) – Enable Cycles image denoising.

  • use_adaptive_sampling (bool, optional) – Enable Cycles adaptive sampling.

output_dir
engine
device
samples
max_bounces
threads
render_deformed
save_images
save_scene
seed
use_denoising
use_adaptive_sampling
__init__(output_dir, engine=EBlenderEngine.CYCLES, device=EBlenderDevice.CPU, samples=2, max_bounces=12, threads=1, render_deformed=False, save_images=False, save_scene=False, seed=0, use_denoising=True, use_adaptive_sampling=True)