GET BUG FOR REAL / MODEL STUDY 01

One scene.
Five models.
No alibis.

A controlled head-to-head test of element/reference-to-video generation using the same live-action cast, the same glass-divided geography, and the same multi-shot dramatic chunks.

Open comparison Read the test design

David inside the literal glass nursery reference world — **DAVID**SCENE-NATIVE IDENTITY

Youxin inside the literal glass nursery reference world — **YOUXIN**SCENE-NATIVE IDENTITY

WARM BEDROOM CLINICAL OBSERVATION

03multi-shot chunks

05model lanes

01locked reference pack

00completed test renders

CONTROL THE INPUT

Every supplied pixel belongs to the scene.

V2 removes the contradictory “style only” and illustrated references. One live-action master owns the world; every identity and chunk frame is derived from it and can be interpreted literally.

One literal scene system. Eight coherent live-action stills cost $1.848 to build and are reusable shared preparation—not charged again to individual model lanes.

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.

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.

SYNCHRONIZED REVIEW

Five lanes. One playhead.

SYNC 00:00.0 Loop Mute

FAIRNESS PROTOCOL

What stays locked.

01
Same dramatic brief
Identical story beats, spoken intent, geography, casting, aspect ratio, and requested duration.
02
Native syntax only
The content is locked; prompt syntax is adapted to each provider’s documented reference mechanism.
03
First viable result
Record the first complete render before any retries. A retry becomes a separately costed attempt.
04
Receipts, not estimates
Cost and render time come from actual generation receipts and timestamps, then remain editable for audit.
05
Two distinct tracks
True multi-reference video models rank together. FLUX.3 and LTX 2.5 remain visible but are not misrepresented as equivalent R2V APIs.

DECISION SURFACE

Turn impressions into production guidance.