Branch: test (v0.1.5)

Prerequisites

• I am using the latest available Docker container
• I have checked that there are not duplicate existing issues
• Documentation
• Bug
• Feature

Description

See expected/actual behaviour.

Steps to Reproduce

Run instances.csr or this much simpler one

# sphere.csr
version 0.1.5

Camera
lookfrom 10 0 0
lookat 0 0 0
vup 0 1 0
vfov 60
aspect_ratio 16/9
aperture 0.0001
focus_dist 10

Sky
top 0.5 0.7 1.0
bottom 1.0 1.0 1.0

Material[Diffuse]
id red
texture no
albedo 1.0 0.2 0.2
roughness 0.5

Sphere
id sphere1
position 0 -1 3
material red
radius 1
medium false

Instance[SpherePrimitive]
prim_id sphere1
translate 0 0 -6

Expected Behaviour

A correct Oren Nayar implementation should look like this:

Actual Behaviour

Currently even at 0 roughness the spheres look flat - as if they are always rough...

The cos_theta term in the integrator has been double checked many times including the OrenNayar pdf - if either is set to use the r_in direction instead then the grazing edge darkness as seen with sigma = 0 IS evident, but then increased roughness also doesn't have the expected output. Furthermore, the horizons when oren nayar spheres are used as grounds end up ridiculously dark.

Self-Service

• I'd be willing to fix this bug/add this feature myself
• I'd be willing to add this documentation myself
• I would like help/cannot do this feature/documentation