STD-Raytracer

{
  "camera": { },
  "scene": {
    "backgroundcolor": [],
    "shapes": [],
    "lightsources": []
  },
  "rendermode": "preview|phong|pathtracer|binary",
  "nbounces": 5,
  "frames": 1,
  "shadow samples": 1,
  "camera samples": 1,
  "animation path": "animation.json"
}

Required fields:

• type
• width
• height
• fov
• position
• lookAt
• upVector

Optional fields include:

• exposure
• gamma
• diameter
• Lw

Supported light types:

• ambient
• pointlight
• arealight

Point lights and area lights can define attenuation coefficients:

• kc
• kl
• kq

Area lights accept either a 2D size:

"size": [width, depth]

or a 3D size:

"size": [width, height, depth]

Implemented shape types:

• triangle
• sphere
• uvsphere
• cylinder
• uvcylinder
• ply
• cubemap
• sphericalmap

Common required fields:

Optional fields used by some shapes include:

• uv0
• uv1
• uv2
• alias
• euler
• texturescale

Observed material fields:

• ka
• kd
• ks
• diffusecolor
• specularcolor
• specularexponent
• reflectivity
• isreflective
• isrefractive
• refractiveindex
• roughness
• metallic
• transmittance
• texturepath

Implementation details:

• roughness >= 0 enables the BRDF path.
• Roughness is squared internally for GGX-style parameterization.
• Metallic materials use a Disney-style metallic workflow.
• Fresnel reflectance uses Schlick approximation.

A scene can reference an external animation file:

"animation path": "animscene.json"

Objects may define an alias; animation data can target aliases instead of autogenerated object IDs.

The renderer constructs a BVH automatically. Every primitive provides a bounding box and participates in acceleration.

Supported texture mappings include:

• UV sphere mapping
• UV cylinder mapping
• UV triangle mapping
• Cube maps
• Spherical environment maps

Textures are loaded as floating-point RGB values. Current texture support is for PPM files.

Area lights precompute a sample cache and reuse it during rendering to reduce runtime random-number generation.

Rendering is performed in linear colour space. Before output, the renderer applies:

1. Exposure
2. Tone mapping
3. Gamma correction

The script below exports key-framed object position/rotation data for use with the animation JSON format. It cannot guarantee correct output for every Blender scene.

import bpy
from numpy import degrees # can be replaced with math module

# Note: this only extract key-framed animations
# 1. bake all simulations
# 2. select the objects to bake
# 3. in Object menu at top-left, select [Object > Rigid Body > Bake to Keyframes]
# 4. now, run this script

# This brings up console for debug:
# bpy.ops.wm.console_toggle()

# name of objects to exclude
exclude = ["0", "836"]

# path to save the json
file = open("path\\to\\your\\animscene.json", 'w')

# load scene, objects and frame numbers
scene = bpy.context.scene
frame_num = scene.frame_end - scene.frame_start + 1
objects = [o for o in bpy.context.scene.objects if o.name not in exclude]

file.write("{\n\t\"frames\" : %d,\n\t\"animated objects\" : %d,\n\t\"animations\" : [\n"%(frame_num, len(objects)))
template = "\t\t\t{\n\t\t\t\"pos\":[%.9f, %.9f, %.9f],\n\t\t\t\"rot\":[%.9f, %.9f, %.9f]\n\t\t\t},\n"
content = ""
for obj in objects:
    object_animation = "\t{\n\t\t\"id\" : \"%s\",\n\t\t\"transform\" : [\n"%obj.name
    animated_frames = ""
    for frame in range(scene.frame_start, scene.frame_end+1):
        scene.frame_set(frame)
        loc = obj.location
        rot = degrees(obj.rotation_euler)
        # print(frame, loc)
        animated_frames += template%(loc.x, loc.y, loc.z, rot[0], rot[1], rot[2])
    object_animation += (animated_frames[:-2] + "\n\t\t]\n\t},\n")
    content += object_animation
file.write(content[:-2] + "\n\t]\n}\n")
file.close()