Spatially varying albedo in volume

[maxie1059]

Hi,

I am trying and haven't been able to simulate a rainbow hologram volume (i.e. a cube of spatially varying albedo with gradient of colors).

The nearest tutorials I knew are "Volume_optimization" and "Participating media" bunny and perhaps, discussion #999. This is my code and results. I am guessing I need an appropiate .vol file.

import matplotlib.pyplot as plt
import drjit as dr
import mitsuba as mi

mi.set_variant('cuda_ad_spectral')

from mitsuba import ScalarTransform4f as T

to_world_sensor = T().look_at(
    target=[0,0,0],
    origin=[12,5,0],
    up=[0,0,1]
)

integrator = {
    'type': 'prbvolpath',
}
emitter1 = {
    'type': 'constant'
}
sensor = {
    'type': 'perspective',
    "fov_axis": "x",
    "fov": 34.0,
    "to_world": to_world_sensor,
    "film": {
        "type": "hdrfilm",
        "width": 1024,
        "height": 1024,
        "pixel_format": "rgba"
}
}
v_res = 16
teapot = {
    'type': 'ply',
    'filename': '/mitsuba-tutorials/scenes/meshes/teapot.ply',
    'to_world': T().translate([0, 0, 0]),
    'bsdf': {'type': 'null'},
    'interior': {
        'type': 'homogeneous',
        'sigma_t': {
            'type': 'srgb',
            'color': [0.1,0.8,0.2],
        },
        'albedo': {
            'type': 'gridvolume',
            'grid': mi.VolumeGrid(dr.full(mi.TensorXf, 0.5, (v_res, v_res, v_res, 3))),
            'to_world': T().rotate([1, 0, 0], -90).scale(5).translate(-0.5),
            'color':[0.1,0.5,0.2],
            'use_grid_bbox': True,
        },
        'phase': {'type': 'hg','g':0.7},
        'sample_emitters':False,
    },
}

scene_dict = {
    'type': 'scene',
    'integrator': integrator,
    'sensor': sensor,
    'teapot': teapot,
    'light1': emitter1,
}
scene = mi.load_dict(scene_dict)
image = mi.render(scene, spp=128)
plt.imshow(mi.util.convert_to_bitmap(image))
plt.axis('off')
plt.show()

[njroussel]

Hi @maxie1059

I'm not familiar with how these holograms actually work. The real-life process might actually need dispersion, or a heterogeneous medium -- I have no idea.

As for your current code, you'd indeed the correct "*.vol" file for your albedo parameter. The shape that encloses the media, i.e the teapot here, also defines the bounds for your medium - you might want to modify that.

[maxie1059]

Thank you for the reply! May I ask if there's sources/websites I could find for *.vol file? Or perhaps, how I could make one myself?

[merlinND]

Hello @maxie1059,

I think the easiest way would be to find OpenVDB files online and convert them using this utility from @dvicini: https://github.com/mitsuba-renderer/mitsuba2-vdb-converter
You can also of course create your own .vol files.