Measurements Scene gallery Records

Implementation output / retained evidence

Light, rendered
and measured.

Native scenes, exact settings, and the performance–accuracy tradeoff of a rasterized photon-sheet viewfinder.

20captured scenes
66native films
6matched studies
3recorded binaries
01 / Preview

The cost of a faster view

Reference gathering and a biased rasterized preview share the same scene, light convention and angular geometry in each pair. The preview reduces camera work, interpolates shading and filters narrow folds. A ratio above 1 means the preview costs less per measured pass.

Time to the first native film
Vector chart ↗PNG ↗
Steady rendering cost
Vector chart ↗PNG ↗

Matched native study · 84

Utah teapot

fast-preview-final-teapot

Finite sheet preview
480 × 300 output240 × 150 preview raster96 rays × 8 sheets12 glass interfacesRefracted fog onlyBlur σ = 1.0 px
First renderer pass24.88 → 20.40 ms1.22× reference / preview cost
Steady render pass22.81 → 15.71 ms1.45× reference / preview cost

First film

One native pass on each path. Geometry setup is included in the measured cost.

Reference · first film
480 × 300 · 1 passes
24.88 ms / measured pass
Mean RGB 0.001176 · peak component 0.521
Configuration Render timing Film measurements
Preview · first film
480 × 300 · 1 passes
20.40 ms / measured pass
Mean RGB 0.001332 · peak component 0.05978
Configuration Render timing Film measurements

Approximately equal render time

Pass counts come from steady timing estimates; the recorded total below includes initialization, so actual times can differ.

Reference · approximately equal render time
480 × 300 · 22 passes
25.67 ms / measured pass
Mean RGB 0.001263 · peak component 0.217
Configuration Render timing Film measurements
Preview · approximately equal render time
480 × 300 · 32 passes
15.40 ms / measured pass
Mean RGB 0.001428 · peak component 0.04848
Configuration Render timing Film measurements

Reference: 564.8 ms total measured render time · Preview: 492.9 ms total measured render time

Equal pass count · retained films

Equal pass count

The same angular sample count per pass. Screen filtering and omitted specular camera paths still change the image.

Reference · equal pass count
480 × 300 · 16 passes
22.81 ms / measured pass
Mean RGB 0.001257 · peak component 0.2874
Configuration Render timing Film measurements
Preview · equal pass count
480 × 300 · 16 passes
15.71 ms / measured pass
Mean RGB 0.001507 · peak component 0.04736
Configuration Render timing Film measurements
Reference preservation · All checked films byte identical

Scene 48 · strategy 20 · byte identical

Checkpoint · preview disabled
480 × 300 · 4 passes
3.75 ms / measured pass
Mean RGB 0.03379 · peak component 10.61
Configuration Render timing Film measurements
Current build · preview disabled
480 × 300 · 4 passes
4.26 ms / measured pass
Mean RGB 0.03379 · peak component 10.61
Configuration Render timing Film measurements

Scene 84 · strategy 32 · byte identical

Checkpoint · preview disabled
480 × 300 · 4 passes
24.97 ms / measured pass
Mean RGB 0.001416 · peak component 0.2082
Configuration Render timing Film measurements
Current build · preview disabled
480 × 300 · 4 passes
25.22 ms / measured pass
Mean RGB 0.001416 · peak component 0.2082
Configuration Render timing Film measurements

Matched native study · 87

Glass chain

fast-preview-final-chain

Finite sheet preview
480 × 300 output240 × 150 preview raster96 rays × 8 sheets12 glass interfacesRefracted fog onlyBlur σ = 1.0 px
First renderer pass22.59 → 20.36 ms1.11× reference / preview cost
Steady render pass15.92 → 14.55 ms1.09× reference / preview cost

First film

One native pass on each path. Geometry setup is included in the measured cost.

Reference · first film
480 × 300 · 1 passes
22.59 ms / measured pass
Mean RGB 0.0006919 · peak component 3.991
Configuration Render timing Film measurements
Preview · first film
480 × 300 · 1 passes
20.36 ms / measured pass
Mean RGB 0.000702 · peak component 0.1952
Configuration Render timing Film measurements

Approximately equal render time

Pass counts come from steady timing estimates; the recorded total below includes initialization, so actual times can differ.

Reference · approximately equal render time
480 × 300 · 31 passes
15.94 ms / measured pass
Mean RGB 0.0006643 · peak component 0.5673
Configuration Render timing Film measurements
Preview · approximately equal render time
480 × 300 · 34 passes
14.67 ms / measured pass
Mean RGB 0.0007254 · peak component 0.1908
Configuration Render timing Film measurements

Reference: 494.2 ms total measured render time · Preview: 498.8 ms total measured render time

Equal pass count · retained films

Equal pass count

The same angular sample count per pass. Screen filtering and omitted specular camera paths still change the image.

Reference · equal pass count
480 × 300 · 16 passes
15.92 ms / measured pass
Mean RGB 0.000669 · peak component 0.937
Configuration Render timing Film measurements
Preview · equal pass count
480 × 300 · 16 passes
14.55 ms / measured pass
Mean RGB 0.0007182 · peak component 0.1875
Configuration Render timing Film measurements

Matched native study · 85

Glass spheres

fast-preview-final-spheres

Finite sheet preview
480 × 300 output240 × 150 preview raster96 rays × 8 sheets12 glass interfacesRefracted fog onlyBlur σ = 1.0 px
First renderer pass20.95 → 22.39 ms0.94× reference / preview cost
Steady render pass16.54 → 17.77 ms0.93× reference / preview cost

At this resolution, the preview costs 7.5% more per steady pass. This result is retained alongside the faster cases.

First film

One native pass on each path. Geometry setup is included in the measured cost.

Reference · first film
480 × 300 · 1 passes
20.95 ms / measured pass
Mean RGB 0.001907 · peak component 4.439
Configuration Render timing Film measurements
Preview · first film
480 × 300 · 1 passes
22.39 ms / measured pass
Mean RGB 0.002645 · peak component 0.1923
Configuration Render timing Film measurements

Approximately equal render time

Pass counts come from steady timing estimates; the recorded total below includes initialization, so actual times can differ.

Reference · approximately equal render time
480 × 300 · 30 passes
15.98 ms / measured pass
Mean RGB 0.001468 · peak component 0.9491
Configuration Render timing Film measurements
Preview · approximately equal render time
480 × 300 · 28 passes
17.21 ms / measured pass
Mean RGB 0.001834 · peak component 0.09797
Configuration Render timing Film measurements

Reference: 479.3 ms total measured render time · Preview: 481.9 ms total measured render time

Equal pass count · retained films

Equal pass count

The same angular sample count per pass. Screen filtering and omitted specular camera paths still change the image.

Reference · equal pass count
480 × 300 · 16 passes
16.54 ms / measured pass
Mean RGB 0.001565 · peak component 0.6704
Configuration Render timing Film measurements
Preview · equal pass count
480 × 300 · 16 passes
17.77 ms / measured pass
Mean RGB 0.001919 · peak component 0.1019
Configuration Render timing Film measurements

Matched native study · 86

Glass torus

fast-preview-final-torus

Finite sheet preview
480 × 300 output240 × 150 preview raster96 rays × 8 sheets12 glass interfacesRefracted fog onlyBlur σ = 1.0 px
First renderer pass25.47 → 24.37 ms1.05× reference / preview cost
Steady render pass27.51 → 23.12 ms1.19× reference / preview cost

First film

One native pass on each path. Geometry setup is included in the measured cost.

Reference · first film
480 × 300 · 1 passes
25.47 ms / measured pass
Mean RGB 0.001042 · peak component 0.7706
Configuration Render timing Film measurements
Preview · first film
480 × 300 · 1 passes
24.37 ms / measured pass
Mean RGB 0.001308 · peak component 0.03733
Configuration Render timing Film measurements

Approximately equal render time

Pass counts come from steady timing estimates; the recorded total below includes initialization, so actual times can differ.

Reference · approximately equal render time
480 × 300 · 18 passes
26.52 ms / measured pass
Mean RGB 0.001084 · peak component 0.1336
Configuration Render timing Film measurements
Preview · approximately equal render time
480 × 300 · 22 passes
23.19 ms / measured pass
Mean RGB 0.001315 · peak component 0.02382
Configuration Render timing Film measurements

Reference: 477.3 ms total measured render time · Preview: 510.3 ms total measured render time

Equal pass count · retained films

Equal pass count

The same angular sample count per pass. Screen filtering and omitted specular camera paths still change the image.

Reference · equal pass count
480 × 300 · 16 passes
27.51 ms / measured pass
Mean RGB 0.001061 · peak component 0.1501
Configuration Render timing Film measurements
Preview · equal pass count
480 × 300 · 16 passes
23.12 ms / measured pass
Mean RGB 0.001333 · peak component 0.02434
Configuration Render timing Film measurements

Matched native study · 84

Utah teapot

fast-preview-final-teapot-hd

Finite sheet preview
1280 × 800 output640 × 400 preview raster96 rays × 8 sheets12 glass interfacesRefracted fog onlyBlur σ = 1.0 px
First renderer pass157.07 → 49.94 ms3.15× reference / preview cost
Steady render pass138.88 → 46.83 ms2.97× reference / preview cost

First film

One native pass on each path. Geometry setup is included in the measured cost.

Reference · first film
1280 × 800 · 1 passes
157.07 ms / measured pass
Mean RGB 0.001178 · peak component 0.581
Configuration Render timing Film measurements
Preview · first film
1280 × 800 · 1 passes
49.94 ms / measured pass
Mean RGB 0.001334 · peak component 0.08204
Configuration Render timing Film measurements

Approximately equal render time

Pass counts come from steady timing estimates; the recorded total below includes initialization, so actual times can differ.

Reference · approximately equal render time
1280 × 800 · 4 passes
149.61 ms / measured pass
Mean RGB 0.001423 · peak component 0.4139
Configuration Render timing Film measurements
Preview · approximately equal render time
1280 × 800 · 11 passes
46.37 ms / measured pass
Mean RGB 0.001545 · peak component 0.07168
Configuration Render timing Film measurements

Reference: 598.4 ms total measured render time · Preview: 510.1 ms total measured render time

Equal pass count · retained films

Equal pass count

The same angular sample count per pass. Screen filtering and omitted specular camera paths still change the image.

Reference · equal pass count
1280 × 800 · 16 passes
138.88 ms / measured pass
Mean RGB 0.001257 · peak component 0.3621
Configuration Render timing Film measurements
Preview · equal pass count
1280 × 800 · 16 passes
46.83 ms / measured pass
Mean RGB 0.001508 · peak component 0.06897
Configuration Render timing Film measurements

Matched native study · 87

Glass chain

fast-preview-final-chain-hd

Finite sheet preview
1280 × 800 output640 × 400 preview raster96 rays × 8 sheets12 glass interfacesStage visibleBlur σ = 1.0 px
First renderer pass134.26 → 56.84 ms2.36× reference / preview cost
Steady render pass126.42 → 58.40 ms2.16× reference / preview cost

First film

One native pass on each path. Geometry setup is included in the measured cost.

Reference · first film
1280 × 800 · 1 passes
134.26 ms / measured pass
Mean RGB 0.000956 · peak component 4.937
Configuration Render timing Film measurements
Preview · first film
1280 × 800 · 1 passes
56.84 ms / measured pass
Mean RGB 0.001008 · peak component 0.3866
Configuration Render timing Film measurements

Approximately equal render time

Pass counts come from steady timing estimates; the recorded total below includes initialization, so actual times can differ.

Reference · approximately equal render time
1280 × 800 · 4 passes
139.80 ms / measured pass
Mean RGB 0.001018 · peak component 4.451
Configuration Render timing Film measurements
Preview · approximately equal render time
1280 × 800 · 9 passes
57.16 ms / measured pass
Mean RGB 0.001079 · peak component 0.3432
Configuration Render timing Film measurements

Reference: 559.2 ms total measured render time · Preview: 514.5 ms total measured render time

Equal pass count · retained films

Equal pass count

The same angular sample count per pass. Screen filtering and omitted specular camera paths still change the image.

Reference · equal pass count
1280 × 800 · 16 passes
126.42 ms / measured pass
Mean RGB 0.0009665 · peak component 1.229
Configuration Render timing Film measurements
Preview · equal pass count
1280 × 800 · 16 passes
58.40 ms / measured pass
Mean RGB 0.0118 · peak component 1877
Configuration Render timing Film measurements
02 / Native images

Scenes rendered by the implementation

These are native image outputs, with their original aspect ratios and capture counts. Integrators have different capabilities. Launch-to-film time includes setup and is not a steady frame-rate measurement.

Luminous atlas · volume light, rough metal and dispersive glass

GPU path tracer
91 · luminous atlas
996 × 700 · 1024 GPU samples/pixel
31.59 s launch → film · includes startup
Mean RGB 0.1855 · peak component 9.006
Configuration Film measurements

Material and atmosphere studios

GPU path tracer
91 · luminous atlas
996 × 700 · 1024 GPU samples/pixel
29.74 s launch → film · includes startup
Mean RGB 0.1839 · peak component 9.006
Configuration Film measurements
GPU path tracer
92 · cloud theatre
996 × 700 · 1024 GPU samples/pixel
30.91 s launch → film · includes startup
Mean RGB 0.05709 · peak component 6.381
Configuration Film measurements
GPU path tracer
93 · tinted atmospheres
996 × 700 · 1024 GPU samples/pixel
47.97 s launch → film · includes startup
Mean RGB 0.1153 · peak component 2.028
Configuration Film measurements
GPU path tracer
94 · brushed observatory
996 × 700 · 1024 GPU samples/pixel
11.08 s launch → film · includes startup
Mean RGB 0.1059 · peak component 8.764
Configuration Film measurements
GPU path tracer
95 · prismatic still life
996 × 700 · 1024 GPU samples/pixel
14.29 s launch → film · includes startup
Mean RGB 0.3452 · peak component 13.57
Configuration Film measurements
GPU path tracer
96 · lacquer constellation
996 × 700 · 1024 GPU samples/pixel
9.67 s launch → film · includes startup
Mean RGB 0.08494 · peak component 56.91
Configuration Film measurements
GPU path tracer
97 · amber hour
996 × 700 · 1024 GPU samples/pixel
11.29 s launch → film · includes startup
Mean RGB 0.1787 · peak component 12.71
Configuration Film measurements

All fifteen paired scenes · initial native captures

Photon primitives
88 · chromatic light loom
800 × 500 · 64 native passes
3.09 s launch → film · includes startup
Mean RGB 0.009665 · peak component 9.397
Configuration Primitive measurements Film measurements
Photon primitives
89 · luminous solid
800 × 500 · 64 native passes
1.61 s launch → film · includes startup
Mean RGB 0.0699 · peak component 69.2
Configuration Primitive measurements Film measurements
Photon primitives
90 · luminous solid
800 × 500 · 64 native passes
1.63 s launch → film · includes startup
Mean RGB 0.07284 · peak component 2192
Configuration Primitive measurements Film measurements
GPU path tracer
91 · luminous atlas
996 × 700 · 64 GPU samples/pixel
2.26 s launch → film · includes startup
Mean RGB 0.184 · peak component 10.91
Configuration Film measurements
GPU path tracer
92 · cloud theatre
996 × 700 · 64 GPU samples/pixel
2.44 s launch → film · includes startup
Mean RGB 0.05727 · peak component 15.31
Configuration Film measurements
GPU path tracer
93 · tinted atmospheres
996 × 700 · 64 GPU samples/pixel
3.47 s launch → film · includes startup
Mean RGB 0.1154 · peak component 2.764
Configuration Film measurements
GPU path tracer
94 · brushed observatory
996 × 700 · 64 GPU samples/pixel
1.04 s launch → film · includes startup
Mean RGB 0.1059 · peak component 10.02
Configuration Film measurements
GPU path tracer
95 · prismatic still life
996 × 700 · 64 GPU samples/pixel
1.24 s launch → film · includes startup
Mean RGB 0.3452 · peak component 46.95
Configuration Film measurements
GPU path tracer
96 · lacquer constellation
996 × 700 · 64 GPU samples/pixel
1.04 s launch → film · includes startup
Mean RGB 0.08501 · peak component 58.52
Configuration Film measurements
GPU path tracer
97 · amber hour
996 × 700 · 64 GPU samples/pixel
1.04 s launch → film · includes startup
Mean RGB 0.1786 · peak component 14.71
Configuration Film measurements
Photon primitives
98 · fast glass procession
800 × 500 · 64 native passes
3.46 s launch → film · includes startup
Mean RGB 0.001031 · peak component 0.2591
Configuration Primitive measurements Film measurements
Path guiding
99 · learn the doorway
498 × 350 · 64 native passes
11.46 s launch → film · includes startup
Mean RGB 0.1548 · peak component 6.446
Configuration Film measurements
MLT
100 · find the bright path
498 × 350 · 64 native passes
2.04 s launch → film · includes startup
Mean RGB 0.4062 · peak component 76.24
Configuration Film measurements
Photon mapper
101 · deposit the light
996 × 700 · 64 photon-map display frames
1.05 s launch → film · includes startup
Mean RGB 0.1721 · peak component 0.9458
Configuration Film measurements
VCM
102 · connect and merge
498 × 350 · 64 native passes
11.27 s launch → film · includes startup
Mean RGB 0.5647 · peak component 92.14
Configuration Film measurements

The same bright-path scene through PSSMLT

PSSMLT
100 · find the bright path
498 × 350 · 512 native passes
13.10 s launch → film · includes startup
Mean RGB 0.4 · peak component 26.59
Configuration Film measurements

High-dimensional emitter sweep · cube

Photon primitives
89 · luminous solid
800 × 500 · 512 native passes
9.29 s launch → film · includes startup
Mean RGB 0.07079 · peak component 2.395
Configuration Primitive measurements Film measurements

High-dimensional emitter sweep · ball

Photon primitives
90 · luminous solid
800 × 500 · 512 native passes
9.28 s launch → film · includes startup
Mean RGB 0.0708 · peak component 2.46
Configuration Primitive measurements Film measurements
03 / Reproducibility

Every image has a record

The compact report contains original PNGs plus settings, timing and native-result JSON. Larger float films and native logs remain at the source paths recorded in the summary. PNGs are copied byte for byte; no local image adjustment is applied here.

Capture summary

Charts show CPU render-pass wall time. First-pass charts include geometry construction; steady charts exclude the recorded warmup. Paired runs share one executable and settings, while different studies may come from different binaries or capture sizes. The native timing record can name the OpenGL display device; that does not make the CPU preview GPU accelerated.

Native input reports & hashes
  • benchmark · fast-preview-final-teapotC:\CC_RadianceLab\build\fast-preview-final-teapot\fast-preview-report.jsonReport SHA-256: 4df49108a684110564937210e59d38fed3298c65ec534d08c1b990785bb3db70
  • benchmark · fast-preview-final-chainC:\CC_RadianceLab\build\fast-preview-final-chain\fast-preview-report.jsonReport SHA-256: 69d46bbdf98f5ce5a7f76f73b84d4a88e58e003a4ad51e83686411a892484025
  • benchmark · fast-preview-final-spheresC:\CC_RadianceLab\build\fast-preview-final-spheres\fast-preview-report.jsonReport SHA-256: e01f2d4bda977b6001f934026c20ae435f89ad171bc64f4e45ef2fa5cae28816
  • benchmark · fast-preview-final-torusC:\CC_RadianceLab\build\fast-preview-final-torus\fast-preview-report.jsonReport SHA-256: f80560ae317206952221572c474edb2b2d6730c7d9654ee575d3e6e248425f18
  • benchmark · fast-preview-final-teapot-hdC:\CC_RadianceLab\build\fast-preview-final-teapot-hd\fast-preview-report.jsonReport SHA-256: 782f21f38d1094f181e23140279b2a6ce0598661da308b65b6f65a5050f15c34
  • benchmark · fast-preview-final-chain-hdC:\CC_RadianceLab\build\fast-preview-final-chain-hd\fast-preview-report.jsonReport SHA-256: 89b63ddbb9eb41f1bb21a10d4373f3289598a3b0b395c9377ec2f10123482320
  • captures · showcases-final-atlasC:\CC_RadianceLab\build\showcases-final-atlas\results.jsonReport SHA-256: 92aa7b8b660f28985630fb171f25129372a9871f88ccc5c436fa7d31227dca27
  • captures · showcases-final-beautyC:\CC_RadianceLab\build\showcases-final-beauty\results.jsonReport SHA-256: 8a2f74d8935a31853a917977cfcd486cd0caef2e003e132fbceb29a54101cd09
  • captures · showcases-final-smokeC:\CC_RadianceLab\build\showcases-final-smoke\results.jsonReport SHA-256: 33f343de1ed18c7bae2c56987f397f709b0f227e80d407bcce1eb0928176a7c7
  • captures · showcases-final-pssmltC:\CC_RadianceLab\build\showcases-final-pssmlt\results.jsonReport SHA-256: 8ce10caf27141de7bfb04ac099ddb422d09b3008f51c97506975aab2625a303f
  • captures · showcases-final-cube-sweepC:\CC_RadianceLab\build\showcases-final-cube-sweep\results.jsonReport SHA-256: bc9fab33effb5e0cc022bf1466aa858643c6ca88c8023025d2903edc91d4e34f
  • captures · showcases-final-ball-sweepC:\CC_RadianceLab\build\showcases-final-ball-sweep\results.jsonReport SHA-256: 1b959e9606b6e531c72d7c7ae347c6cdea7b5f7c2b217a0ce127494f36bef147