Target crowd size 1,000

A 1,000-person crowd over 2 shoot days creates 2,000 background actor-days before catering, staging, and AD wrangling.

Shoot days 2 days

Physical crowd costs scale with each day. The hybrid path keeps only 50 paid extras on set, then shifts the expansion work into post.

SAG-AFTRA Zone? Union

Union zones model a fully loaded $540 background rate. Non-union scenarios use $350, but lose the compliance safety of covered minimums and documented consent.

Outputs · 1,000 bodies on camera

Projected crowd cost & delivery burden

Physical Background Army

On-set logistics

$0 0 weeks

Every visible person is hired and managed physically. Model: target crowd × shoot days × loaded daily rate, plus $35k in basecamp, AD, sanitation, and holding overhead.

Pure AI · Vendor Claim

Marketing narrative

$0 0 weeks

The self-serve SaaS pitch collapses the estimate to a flat $500. Useful as a hype floor, but it excludes consent, directability, QC, temporal stabilization, and VFX revision labor.

Controlled 50-Extra Hybrid

Recommended

$0 0 weeks

Shoot 50 directable extras for plate/tile passes, then spend a fixed $115k on roto, matchmove, crowd simulation, compositing, and revision cycles.

Risk Dashboard

Why synthetic crowds still need production discipline

Risk 01

Anti-Circumvention Minimums

Severe

In covered SAG-AFTRA zones, digital replicas cannot be used to dodge required background coverage. Productions must hire the physical minimums first, then augment the scene digitally.

Covered background minimums must be satisfied before synthetic augmentation.
Replica scans need clear, conspicuous, separately documented consent.
Scan-only days can still trigger full background-actor wages.
Making a replica speak or perform principal action can trigger upgrades and residual exposure.

Risk 02

AI Slop in Close-Ups

High

Generative crowds can pass in deep background, but telephoto compression, shallow depth of field, and network QC expose distorted faces, jittering fabric, and unstable silhouettes.

Faces morph frame to frame instead of holding identity.
Hands, hats, and props collapse into unreadable shapes.
Lighting and shadows drift during camera moves.
Rejected shots turn into manual paint and comp overages.

Risk 03

Lack of Directability

High

Physical extras and deterministic crowd tools can hit a director's note: stand on frame 48, swap red jacket for blue, cheer in section 104. Prompted video AI is probabilistic.

Specific notes can regenerate the entire background plate.
Timing, wardrobe, and eyelines are hard to isolate cleanly.
VFX bids become speculative when outputs are not repeatable.
Hybrid plate passes preserve a directable human performance source.

Methodology & source formulas

Physical baseline

cost = (crowdSize × shootDays × dailyRate) + $35,000

The fixed $35k covers simplified basecamp, AD, sanitation, and holding overhead. Union scenarios use a $540 loaded daily rate; non-union scenarios use $350.

Vendor claim

cost = $500

The flat SaaS number is intentionally optimistic. It represents the marketing claim, not a broadcast-ready premium-TV delivery plan.

Realistic hybrid

cost = (50 × shootDays × dailyRate) + $115,000

The fixed $115k models VFX post: extraction, clean plates, matchmove, 2.5D tiling or simulation, compositing, QC, and standard revisions.

Minutes of screen time 30

30 min of 24fps footage ≈ 43,200 frames ≈ 514 distinct VFX shots (at a standard 3.5s average shot duration).

Production complexity

Tier determines vendor per-minute rate ($1–2k / $2–5k / $20–100k+) and traditional days-per-shot.

Hybrid artist day rate $850

Fully-loaded studio rate. LA freelance averages $850/day; IATSE Local 800/700 Level 1 runs ~$1,997–2,160/week (~$400–432/day), but studio billing is higher.

Outputs · 514 shots modelled

Projected delivery cost & schedule

Traditional Baseline

Manual craftsmanship

$0 0 weeks

Frame-by-frame rotoscoping, 2D/3D tracking, frequency separation, grain-matched integration. Industry benchmark: 4–8 hours per cosmetic shot; 1–5 days per de-aging shot.

Pure AI · Vendor Claim

Marketing narrative

$0 0 weeks

MARZ Vanity AI claims 300× speed over manual pipelines: ~1 min per cosmetic shot, ~3 min per de-aging shot. Generative video compute: $0.50–$30 per finished minute.

Realistic Hybrid · MSR Estimate

Recommended

$0 0 weeks

AI handles tracking, roto, and initial smoothing (the 80%). Skilled compositors spend 2–4 hours per shot fixing artifacts, rebuilding skin texture, and colour-matching into ACES delivery.

Risk Dashboard

Why the "Pure AI" number is a trap

Risk 01

Talent Consent & Labour Relations

Severe

The 2023 SAG-AFTRA TV/Theatrical agreement splits AI likeness work into Employment-Based Digital Replicas, Independently Created Digital Replicas, and Digital Alterations — each with distinct consent, bargaining, and compensation rules.

48-hour informed-consent notice required before any replica scan.
"Clear and conspicuous" written consent — not buried fine print.
Reuse in a different project = separate consent, separately bargained.
Background actors made to appear to speak must be upgraded to Day Performer.

Risk 02

Quality & the Uncanny Valley

High

Neural-rendering models handle the first 80% of a shot well but routinely fail the cinematic last 20%. Occlusion, motion blur, and dynamic lighting surface the classic AI failure modes.

Over-smoothing the low-frequency layer → "plastic skin".
Frame-to-frame drift → digital makeup swims, boils, flickers.
Eyes drift off-axis; gravitas and subtext erased from the performance.
Skin de-ages correctly but BMI and skeletal mechanics don't follow.

Risk 03

Workflow & Data Provenance

Moderate

Most AI vendors ship black-box SaaS tools. Premium TV runs on Nuke node graphs, 16-bit EXR sequences, and ACES/OCIO colour pipelines — where every mask, matte, and tracking point must be editable.

Flattened outputs can't be surgically adjusted for director notes.
SDR-optimised tooling bakes in colour shifts on log-encoded plates.
Cloud processing of raw footage = unreleased IP on third-party servers.
8K facial photogrammetry is biometric data — on-prem storage required.

Methodology & source formulas

Traditional baseline

cost = shots × daysPerShot × dayRate × 1.5× days = shots × daysPerShot

Days-per-shot: Basic 0.5 · Standard 1.5 · Premium 3. The 1.5× overhead covers facility costs, render farm, supervisors, and QC that studios bundle above the bare artist day rate. Shot count assumes 3.5s average duration.

Vendor claim

cost = $8,000 base + (minutes × $8/min) duration = shots × aiMinPerShot / 60

$8,000 reflects a realistic enterprise SaaS pilot/licensing floor. Compute charge modelled at $8/min (doc range: $0.50–$30). Per-shot time from MARZ Vanity AI: 1 min (basic), 1.5 min (standard), 3 min (de-aging).

Realistic hybrid (MSR)

cost = (shots × cleanupHrs / 8) × dayRate cleanupHrs: 1.5 / 3 / 6 per tier

AI handles tracking + initial roto at negligible incremental cost. Human cleanup is the binding constraint: 2–4 hours per shot at standard complexity, scaling up for de-aging finish.

Premium Television

AI vs. Traditional Establishing Shots

Model a slate of 4K establishing shots for premium scripted television. Compare rights-managed stock or practical 2nd unit acquisition against the sub-$50 AI vendor claim and the realistic hybrid path: prompt generation plus VFX cleanup, upscaling, QC, and legal conformity.

Data: Filmsupply/Getty-style licensing · Runway/Luma/Kling pricing · VFX cleanup · E&O and guild guardrails

Number of required shots 10

10 shots requiring true 4K delivery must cut cleanly against principal photography: lens feel, grain, color space, motion cadence, and automated QC all matter.

Traditional acquisition baseline

Current baseline: premium rights-managed stock. Stock is faster but archive-constrained; 2nd unit capture is slower, expensive, and more directable.

Outputs · 10 establishing shots

Projected acquisition cost & delivery burden

Traditional Acquisition

Cleared baseline

$0 0 weeks

Premium stock licenses high-bitrate, rights-managed footage with faster turnaround but limited creative fit. Traditional paths preserve chain of title, broadcast quality, and protectable footage.

Pure AI · Vendor Claim

Marketing narrative

$0 0 weeks

A $30 SaaS subscription plus $0.50 per shot of compute makes the raw-generation line look microscopic. It excludes prompt failures, 4K upscaling, QC rejection, copyright clearance, and distributor review.

Realistic Hybrid · MSR Estimate

Final-pixel caution

$0 0 weeks

AI is treated as volatile raw material: $150 per shot for prompting and curation, plus $1,150 per shot for VFX cleanup, conformity, color, grain, stabilization, and delivery checks.

Risk Dashboard

Why final-pixel AI establishing shots are legally volatile

Risk 01

Copyright Un-protectability & IP Infringement

Severe

The report treats copyright as the existential risk. Pure prompt-generated video may not be protectable because human authorship remains the gating requirement, while training-data similarity creates infringement exposure.

Autonomous AI video can leave the underlying pixels outside copyright protection.
Heavy human VFX work may create only a thin copyright in the additions.
Training-data resemblance can invite injunctions, statutory damages, and reputational damage.
A rival can reuse an unprotectable synthetic establishing shot without the studio controlling the asset.

Risk 02

E&O Insurance Exclusions & Distributor Mandates

Severe

Networks and streamers need clean chain of title. Generative video breaks provenance because the production cannot prove what sources shaped a clip, making E&O coverage and distributor approval fragile.

Insurers may exclude claims arising directly from generative AI assets.
Without E&O coverage, a show can be blocked from delivery or broadcast.
Distributor rules require enterprise-secured tools and written legal review for visible AI use.
Outputs must not substantially recreate unowned copyrighted material or train on production data.

Risk 03

Union Guardrails & Labor Compliance

High

Establishing shots may seem actor-free, but AI-generated streets, crowds, and synthetic figures can cross WGA, SAG-AFTRA, DGA, crew, and VFX labor boundaries.

Writers must be told when provided material contains AI-generated content.
Synthetic performers or digital replicas can trigger consent and compensation rules.
Generated background action can look like avoided background or crew-covered work.
New York synthetic-performer disclosure rules take effect in June 2026.

Methodology & source formulas

Traditional baseline

stock cost = brollShots × $2,350 shoot cost = brollShots × $4,250

Premium stock models rights-managed broadcast licensing at $23.5k per 10 shots. Practical 2nd unit models crew, gear, permits, travel, and weather friction at $42.5k per 10 shots.

Vendor claim

cost = $30 + (brollShots × $0.50) days = brollShots × 0.05

The SaaS line reflects a subscription and per-shot compute only. It intentionally excludes iteration, upscaling, QC, legal review, and clearance.

Realistic hybrid

cost = brollShots × $1,300 days = brollShots × 1.05

$150 per shot covers prompting labor; $1,150 covers VFX cleanup, 4K conformity, grain/color matching, stabilization, and delivery checks.

Content tier

Current tier: unscripted/factual. The report places factual and library content in the current AI sweet spot; premium scripted remains the danger zone.

Total runtime (mins) 60 min

60 minutes multiplied across languages is the core cost driver for all three localization paths.

Target languages 5

Traditional dubbing spins up a new production cycle for each market. AI workflows scale faster once transcripts, glossaries, and voice models are established.

Cast complexity 8

Casts over 5 speaking roles trigger the model's traditional union/studio logistics penalty. Hybrid AI does not add that cast penalty, assuming voice models or guide performances exist.

Audio mix

Current mix: Stereo/5.1. Atmos adds a 20% mix penalty because AI cannot reliably author object-based vector spaces without a human mix engineer.

Add visual sync (vubbing) Off

Visual dubbing is a premium add-on, not a basic savings lever. This model assumes 2 altered shots per finished minute at $875 per worldwide-streaming token.

Outputs · 5 target languages

Projected localization cost & delivery burden

Traditional Human Dubbing

Studio & talent chain

$0 0 weeks

Voice casting, script adaptation, directors, recording studios, engineers, and final mix scale linearly by language. Cast complexity and Atmos delivery add traditional workflow friction.

Pure AI · Vendor Claim

Zero-touch risk

$0 0 weeks

The $5/min/language line captures raw ASR, translation, and voice synthesis compute only. It excludes cultural adaptation, QA liability, actor consent workflows, and final mix review.

Human-Supervised Hybrid

Enterprise standard

$0 0 weeks

AI does the heavy lift on transcription, translation, and base synthesis; human linguists, directors, and audio QC restore cultural fit, timing, emotional intent, and delivery compliance.

Risk Dashboard

Why zero-touch AI dubbing is not premium-TV ready

Risk 01

SAG-AFTRA Consent & Replicas

Severe

AI dubbing that clones or preserves performer voices enters digital-replica territory. SAG-AFTRA requires clear, conspicuous consent, not buried boilerplate, before voice cloning or replica use.

Producers must give at least 48 hours of notice before scanning or cloning.
Consent is limited to the specific project and use case that was approved.
Spinoffs, games, new seasons, and different properties require fresh consent and compensation.
Vendor systems need auditable consent trails before localized voice replicas ship.

Risk 02

Immersive Audio & Dolby Atmos

High

Stereo and 5.1 dubs can be automated more easily. Dolby Atmos is different: it depends on object-based panning, creative placement, and precise vector spaces AI systems do not yet understand reliably.

AI auto-mixing can smear vocals or place dialogue in the wrong 3D position.
Automated tools miss creative intent, such as off-screen voices moving through surround space.
ADM BWF and immersive deliverables still require expert human mix engineers.
The report treats Atmos as an unavoidable final bottleneck, not a solved AI step.

Risk 03

Emotional Flattening in Narrative

High

The report flags premium scripted drama and comedy as the failure zone. Amazon Prime Video's AI-dubbed Korean dramas were criticized as flat, robotic, and emotionally hollow.

My Man is Cupid and True to Love triggered social backlash and voice-actor criticism.
Audiences noticed missing subtext, timing, breath, and emotional dynamics.
The disputed dubs were reportedly removed quietly after the backlash.
Premium narrative needs voice direction, cultural adaptation, and emotional QA.

Methodology & source formulas

Traditional baseline

cost = baseRate × runtime × languages days = (14 + (2 × runtime / 60)) × languages

Base rates: scripted $500/min/language, unscripted $300, catalog $200. Casts above 5 roles add 15%; Dolby Atmos adds 20%.

Vendor claim

cost = $5 × runtime × languages days = (0.125 + (0.0125 × runtime / 60)) × languages

The pure-AI line represents raw compute and subscription economics. It excludes cultural rewrite, voice direction, consent tracking, and final delivery QA.

Realistic hybrid

cost = hybridBase × runtime × languages days = (2 + (0.5 × runtime / 60)) × languages

Hybrid rates: scripted $45/min/language, unscripted $30, catalog $15. Atmos adds 20%. Visual sync adds $875 × 2 altered shots per minute and 10 days.

01

Percentage

Find a percentage of any value.

Value Percentage

Result 250 25% of 1,000

Value + percent1,250

Value − percent750

02

Percentage Change

Compare two values to see the percent shift.

Original New value

Change 25% Increase from 80 to 100

Difference20

DirectionIncrease

03

Margin & Markup

Compare cost to selling price for either lens.

Cost Selling price

Gross margin 40% Gross profit $40

Markup66.67%

Gross profit$40

04

Break-even

Units required to recover fixed costs.

Fixed costs Price per unit Variable cost / unit

Break-even 3,334 units Unit contribution $15.00

Contribution / unit$15.00

Break-even revenue$83,350

05

CAGR

Compound annual growth rate between two points.

Start value End value Years

CAGR 20.11% Annualized over 5 years

Total growth150%

06

Weighted Average

Average values weighted by their importance.

Value Weight

Weighted average 83.33 Across 6 total weight

07

Ratio

Simplify two numbers into a clean ratio.

First value Second value

Simplified ratio 3:5 Simplified from 15:25

Share of total37.5% · 62.5%

08

Compound Growth

Project a value forward at a compounding rate.

Starting value Rate per period (%) Number of periods

Final value $19,671.51 7% rate over 10 periods

Absolute growth$9,671.51

Total growth96.72%

09

ROI

Return on an investment, as a percent and a multiple.

Investment cost Return value

ROI 50% Net profit $5,000

Net profit$5,000

Return multiple1.5×

10

Unit Cost

Per-unit cost from a total and a count.

Total cost Number of units

Cost per unit $20.00 From $5,000 over 250 units

10 units$200

100 units$2,000

1,000 units$20,000

In development

This tool is on the roadmap

The underlying research is underway. Check back for the full model.