Model Lab

Empirical model testing, published openly.

Real measurements from real generation runs: convergence per step, quality per resolution and aspect ratio, cost, reproducibility, failure rate. Published when the method is reproducible by a stranger with the same open-source model and config. Client-derived tests and proprietary pipeline internals stay private, on principle, not as a limitation, see the disclosure note below.

Media wall — every modality, every stage

real tiles: solid border · fake tiles: dashed red + tag

Moon-01 · step 1/24 · real frame

Moon-01 · step 2/24 · real frame

Moon-01 · step 4/24 · real frame

Moon-01 · step 6/24 · real frame

Moon-01 · step 9/24 · real frame

Moon-01 · step 12/24 · real frame

Moon-01 · step 16/24 · real frame

Moon-01 · step 20/24 · real frame

Moon-01 · step 24/24 · real frame

step 1

Moon-01 · full sweep, real capture · synced to shared clock

Loading real splat data…

Real capture, freely orbitableConventional capture (RealityCapture + Postshot)

Flowers · gaussian splat · conventional capture, real

Fake

step 3

Run 2 · step 3/24 · not captured yet

Fake

step 8

Run 2 · step 8/24 · not captured yet

Fake

step 16

Run 2 · step 16/24 · not captured yet

Fake

step 24

Run 2 · step 24/24 · not captured yet

Fake

0°

Splat reprojection · 0° · not run yet

Fake

90°

Splat reprojection · 90° · not run yet

Fake

180°

Splat reprojection · 180° · not run yet

Fake

Audio · per-step sonification · no real capture yet

Fake

10%

Splat growth · 10% densified · optimizer not built yet

Fake

40%

Splat growth · 40% densified · optimizer not built yet

Fake

75%

Splat growth · 75% densified · optimizer not built yet

Fake

100%

Splat growth · 100% densified · optimizer not built yet

Finding the sweet spot

"Sweet spot" isn't a subjective call here. For any real independent variable (step count, resolution, aspect ratio) a model is swept across, a real quality or convergence signal is measured at each point, and the sweet spot is found geometrically: the point of maximum curvature on the normalized curve (the "kneedle" elbow-detection family). Marked automatically on every chart, not eyeballed.

Sweep axes

Step count: real per-step measured detail/convergence, same technique proven on the homepage's moon-01 pieces.

Resolution / aspect ratio: real per-configuration quality measurement across a model's supported range.

Spatial & temporal consistency: how much a model drifts across a batch (same prompt/style, different seeds) or across frames in a video run, real perceptual-hash distance between real outputs. This answers "will asset #1000 still be on-brand" for a production batch. Not measured yet, planned alongside the resolution sweep.

Step-count convergence

1 real run

Detail energy vs. stepLaplacian magnitude

Denoising stepReal measured detail energy

Real per-step Laplacian-filter magnitude from the moon-01 generation's 9 real captured steps. Convergence within one run's trajectory, a proxy for the sweet-spot question, not yet a true cross-run sweep. Elbow found in today's data: step 12.

Step count, in numbers

Steps captured9samples

Range1–24steps

Sweet spot12steps

Detail gain past sweet spot19%

Resolution & aspect-ratio sweet spot

Preview · fake data

Quality vs. resolutionper model, real sweep pending

Resolution / aspect ratioReal measured quality signal

Invented numbers, layout preview only. No resolution/AR sweep has run yet.

Cost & latency

Preview · fake data

Cost per asset0.014USD

Median latency8.2s

P95 latency14.6s

Failure rate0.6%

Invented numbers, layout preview only. No real Engine benchmark run yet.

Disclosure policy

Published if a stranger could reproduce it with the same open-source model and published seed/config. GENERAITR-specific pipeline architecture and any client-derived test stays private, results only where shown at all.

Start with a pipeline audit

Compare models

Preview · fake data

Spec	Model A (fake)	Model B (fake)	Model C (fake)
Steps (sweet spot)	20	28	16
Resolution	1024x1024	1024x1024	1280x1280
Cost / asset	$0.012	$0.019	$0.021
Median latency	6.4s	9.1s	10.8s
VRAM	12 GB	16 GB	24 GB
License	Apache 2.0	OpenRAIL-M	Apache 2.0

Invented specs, layout preview only. No real head-to-head has run yet.

step 1/24