THE GLASS TESTSCENE 01
LIVE-ACTION R2V LAB

GET BUG FOR REAL / MODEL STUDY 01

One scene.
Five models.
No alibis.

A controlled head-to-head test using one literal live-action master, the same glass-divided geography, and the same three formally planned dramatic chunks.

Open comparison Read the test design
David inside the literal glass nursery reference world
DAVIDSCENE-NATIVE IDENTITY
Youxin inside the literal glass nursery reference world
YOUXINSCENE-NATIVE IDENTITY
WARM BEDROOM CLINICAL OBSERVATION
03multi-shot chunks
05model lanes
01visual authority image
00completed test renders

CONTROL THE INPUT

Every supplied pixel belongs to the scene.

V3 supplies exactly one visual image. It owns both performers, wardrobe, furniture, glass, lighting, and left-to-right geography. No secondary image can compete with it.

01

One literal visual authority. Every endpoint receives the same approved live-action master. Prompt syntax changes by provider; visual evidence does not.

02

Geography survives the medium change. David’s bed is beside a single large plate-glass divide. Youxin occupies the clinical side. They initially feel co-present; the reveal makes the separation legible.

03

The glass has a dramatic function. It separates, receives both hands, then becomes Youxin’s mirror when David’s side goes black.

THE UNIT OF COMPARISON

Multi-shot chunks, not isolated shots.

Each model must solve editorial thought: a beginning, an internal transition, and a resolved end image.

MIRROR CORRECTION / REVIEW GATE

One new frame. No competing authority.

The corrected final chunk is staged from Youxin’s illuminated side so the same pane that shows David can physically become her mirror. The reference, orchestration input, endpoint settings, and terminal errors are exposed below before any paid video submission.

SYNCHRONIZED REVIEW

Five lanes. One playhead.

Every lane exposes its exact submitted prompt, ordered inputs, settings, task ID, and output asset.

00:00.0

FAIRNESS PROTOCOL

What stays locked.

  1. 01
    Same dramatic brief

    Identical story beats, spoken intent, geography, casting, aspect ratio, and requested duration.

  2. 02
    Native syntax only

    The content is locked; prompt syntax is adapted to each provider’s documented reference mechanism.

  3. 03
    First viable result

    Record the first complete render before any retries. A retry becomes a separately costed attempt.

  4. 04
    Receipts, not estimates

    Cost and render time come from actual generation receipts and timestamps, then remain editable for audit.

  5. 05
    Two distinct input semantics

    Wan, H3, and Seedance interpret the image as scene reference. FLUX.3 and LTX 2.5 treat it as exact frame zero; the page labels that difference instead of hiding it.

DECISION SURFACE

Turn impressions into production guidance.