Editorial Board
Series-level production
Manage categories and topics, batch-run autopilot, and organize series output under a single management surface built on the existing render pipeline.
Inside AI Studio • category operations • series workflows
Launch series by category and topic.
Storyboard works one reel at a time. Even in autopilot, it is still generating a single reel from a single brief. Editorial Board takes that further: start with one category premise, generate 5, 10, or 20 distinct topics underneath it, then execute a whole series of videos instead of just one. Each topic becomes its own reel with its own brief, assets, and output, all organized under the same category.
Built for repeatable editorial video formats.
Teams can shape news reel videos, explainers, documentary-style shorts, faceless news channels, and visual essays by hand first, then carry the same structure into autopilot.
Release 3.0
Five major updates in one release.
Release 3 ships Editorial Board, Avatar, an expanded Typography system, Analysis Mode, and Ideation Seeder together. The video below demos the core features in one reel.
Avatar
On-screen presenter
Add an AI-generated avatar host to any reel. Set position, choose from cast, and layer over scene media directly inside the compose flow.
Typography
Advanced title effects
New title card templates covering motion glow, chrome glass, neon, metallic, and more — all rendered client-side for zero AI token spend.
Analysis Mode
Reverse-engineer any video
Extract the script, audio, key frames, style, and narrative arc from any reference video — then seed the render pipeline with the full recreation brief.
Ideation Seeder
Seed and iterate before rendering
A new front door before storyboard. Start from an analysis, a reference image, or a blank prompt — iterate on story and style without spending tokens on full renders.
Overview
Management first. Rendering second.
This is a management surface for teams producing repeatable editorial video formats: news reel videos, explainers, documentary-style shorts, faceless news channels, and visual essays. The difference from storyboard is scale: storyboard handles one reel at a time, while Editorial Board can take one category premise and expand it into many topic-based videos in the same run.
Categories
Organize by series
Browse categories like Fashion, News, College, or Political News, then drill into nested sub-categories when a topic grows into a series.
Topics
Ideate in bulk
Create topics manually or ask AI for 5, 10, or 20 ideas with one-sentence blurbs, then turn that single category premise into a full queue of separate video jobs.
Compose
Autopilot inherits. Manual mode specifies.
Autopilot picks up category defaults automatically, while manual mode lets the operator specify each component directly: summary, script, image, title card, and media choices.
Creation
Map mode before render
Once a topic is composed, the downstream work stays in the proven slideshow, storyboard, audio, captions, and render pipeline, with `editorial-creation` acting as the map mode before final render.
Avatar walkthrough
Editorial Board in under a minute
An on-screen presenter walks through category management, topic ideation, and the series workflow.
Use case
Set up a category once, then let topics inherit the structure.
A team creates a category like College Admissions or Fashion, assigns a default title card, duration, model, aspect ratio, and other defaults, then lets autopilot inherit that setup for each topic job. For a single manual run, the operator can override each component individually and choose the exact summary, script, hero image, and scene media.
Category and topic title cards
Both category-level jobs and individual topics can set a title card. If a category defines a default title card, autopilot reuses it across inherited topic jobs unless the topic overrides it.
Token savings through burn-in
A blank title card can reserve the top two-thirds for imagery and the bottom third for dynamic text. The browser can burn titles into the card with client-side canvas, with a server-side fallback when needed.
Avatar walkthrough
How autopilot works
The presenter explains how autopilot inherits category defaults and batch-generates a full topic queue without manual compose steps.
State model
A topic moves through a clean operational state machine.
Editorial topics are tracked as working objects, not loose prompts. For now, Generated is the real done state. Published stays as a forward-looking toggle for future social distribution.
Ideation
Idea with a blurb
A one-sentence angle exists, but there is no script or scene structure yet.
Compose
Script and sourcing
Summary, reference text, format, title card, uploads, and reusable assets are all selected here.
Creation
Authoring and workflow
The topic is handed into the existing scene, narration, captions, and compile path.
Generated
Finished video
The final MP4 exists. This is the complete state for the current milestone.
Published
Future distribution
Later this can push to social platforms and generate descriptions, tags, and publish metadata.
Faceless video
Done manually before Editorial. Now Editorial does it for you.
These reels were produced by hand before Storyboard mode existed — no automation, no queue, just direct scene assembly. Editorial Board exists to make that process repeatable: set a category, generate topics, and let autopilot produce the same faceless informational format at scale.
Manual · pre-Storyboard
Fashion faceless reel · render job 74
Hand-assembled before any automation existed. The format, pacing, and structure are exactly what Editorial Board now produces through autopilot.
Manual · pre-Storyboard
Fashion faceless reel · render job 75
A second manually produced reel from the same period — two videos that would now be queued and generated as a single autopilot batch.
Current build
The build already shows the operating model.
The admin UI screenshots below are the current Editorial Board build and walk a single topic, Top 5 California dining halls, through landing, category management, summary, script, asset sourcing, scenes, and creation review.
Autopilot output
Top 5 college series · reel 96
One of the generated videos from the `Top 5 college` category series, produced through autopilot and surfaced directly in the CRUD list.
Autopilot output
Top 5 college series · reel 97
A second autopilot-generated reel from the same category series, matching the generated entries shown in the topic queue.
Autopilot output
Top 5 college series · reel 98
A third autopilot output from the same `Top 5 college` category series, showing the generated-video result that the queue can open and play.
Landing
Series-level entry point
Category cards, topline counts, and direct actions for manual category creation or AI-assisted topic generation.
Category detail
Nested categories and reusable assets
Sub-categories, default tag lines, shared artwork, and a topic table all live together before execution.
Topic queue
State-aware topic management
Search, filter by state, trigger autopilot, open generated reels, and keep topic progress visible instead of burying it in renders.
Summary
The human script is first-class
The editor can paste a ranked outline directly, keep it as-is, or ask AI to refine it later. Autopilot is optional.
Script
Readable narration before authoring
The actual scene-by-scene narration is reviewed before creation starts, so operators can catch tone and ordering issues early.
Compose
Title cards, uploads, library, cast
AI generation is just one source option. In manual mode, the operator can specify summary, script, hero image, title card, library assets, uploads, and persistent cast objects one component at a time.
Category defaults
Default projections per category
Voice, aspect ratio, duration, model, style, scene count, title-card behavior, and media defaults can be passed down to every topic so autopilot inherits them automatically and manual runs can override them per job.
Burn-in
Title text burned into the hero image
The operator can flatten title text directly onto the selected hero card, reserve the lower third for dynamic text, render it client-side in browser canvas, and fall back to server-side generation when needed, all without spending AI tokens on a fresh title frame.
Costs
Track spend by board, category, and topic
Editorial carries cost tracking through the whole system: overall spend across the board, rollups at the category level, and per-topic or per-storyboard detail so operators can see what autopilot kept, what was wasted, and which models or media types drove spend.
Scenes
Each scene declares its source
Library image, upload, clip, Pexels pull, or AI-generated reference can all sit side-by-side in a single topic build.
Creation
Editorial creation is the map mode
The system shows the exact scene order, media coverage, narration readiness, and render handoff using the existing production path, with `editorial-creation` functioning as the map mode before final render.
Build view
Map and queue before final render
The build view shows the ordered cards, title frame, topic stats, and render controls together so the operator can inspect the package before committing.
Reuse map
Editorial Board extends haus video. It does not rebuild it.
The management layer is new, but the expensive mechanical work already exists. That is the leverage point: keep categories and topics independent while sending composed topics into the same proven authoring and render services.
What stays the same underneath
Storyboard script generation, cost tracking, asset upload approval, Pexels browsing, cast roster, templates, reels effects, captions, and the BullMQ render queue.
Why this fits
One-shot autopilot later depends on better management now.
The destination is still automated production: headline to script, voiceover, 10 to 20 images, captions, and a finished vertical video. But the bottleneck is operational structure. Editorial Board gives AI Studio the missing surface for categories, reusable assets, topic states, and manual review before a sheet-driven poller starts firing jobs.
Uploads and stock matter
News reels, rankings, and documentary-style shorts often rely on stills, logos, screenshots, and stock footage rather than fully generated scenes.
Templates reduce cost
Hero cards, centered text boxes, and reusable reel formats should be generated locally and reused across topics instead of paying for avoidable AI shots.
Series memory compounds
Persistent cast, reusable library assets, and searchable captions or descriptions let later topics build on earlier work instead of starting from zero each time.
Avatar
Layer a presenter into any reel.
The avatar module integrates directly into the compose workflow. Pick from cast, set position and scale, preview against scene media, and commit to the render queue without leaving AI Studio.
Overview
Select, position, and preview
Browse cast, set position and scale, and preview the avatar composite against actual scene media before committing to the render queue.
Cast
Choose from available presenters
The cast roster surfaces all available avatar options so the right presenter can be assigned per topic or inherited across an entire category series.
Typography
Production-ready title effects, rendered client-side.
Release 3 ships an expanded set of title card templates built entirely in HTML and CSS — motion glow, chrome glass, neon, metallic, and more. No AI token spend, no server round-trip, rendered in the browser.
Effect library
Nineteen effect templates
Each effect is a standalone HTML/CSS card that maps directly to a title card template in the compose and burn-in workflow.
Avatar walkthrough
Typography effects in under a minute
An on-screen presenter introduces the expanded title card library and how effects are applied inside the compose and burn-in workflow.
motion glow
HYPE
pastel gradient
MARCH
retro offset stack
LINE
Analysis Mode
Deconstruct a competitor commercial into a production brief.
Analysis Mode is built for commercials — ads up to three minutes and 100 MB. Drop one in and it extracts everything inside: the script, transcribed audio, generated key frames, visual style, lighting and lens specs, narrative arc, and a full recreation brief. The output seeds the render pipeline directly, no manual brief writing required.
Output · reverse-engineered replica
Our recreation of the competitor ad
Analysis Mode extracted the script, transcribed the audio, and generated a full production brief from the original. This is the video we produced from that brief — a recreation of the competitor's CGI work sock ad, built inside haus video's render pipeline.
Extraction
Style, audio, and narrative — extracted
A video goes in. A structured brief comes out: plain-language description, visual style notes, lighting and lens specs, audio direction, voice tone, and a scene-by-scene narrative arc — ready to pass into a recreation brief or Ideation seed.
Board view
Characters and environments as boards
The analysis-seeded storyboard surfaces every character and environment as a discrete board. Cast, props, lighting rigs, and scene descriptions are all reviewable and editable before any AI generation runs.
Replication
Full board set from one analysis
The extracted brief seeds a storyboard automatically. Each scene, character, and environment maps into boards that match the analyzed video's visual language and can be rendered, rescripted, or handed to Ideation as a style seed.
Rescripting
Edit the script before any render
After analysis, the script can be rewritten before committing to a render. Change the hook, adjust the CTA, or swap the narrative arc — the visual style stays intact while the story is updated.
Extracted style — sock ad analysis output
Style: Hybrid CGI animation + live-action product footage — clay/Pixar-style 3D characters and anatomical renders combined with photoreal product close-ups and composited graphic elements — Pixar short film meets medical explainer animation; clay-sculpture aesthetic with warm studio product photography.
Ideation Seeder
A new front door. Seed a brief, then iterate until it's ready.
Ideation is a creative development layer that sits before storyboard. Start from an Analysis result, a style reference, or a blank prompt. Seed a bare-bones brief, then refine the story, swap environments, and change characters — without committing any AI spend to a full render. When the brief is locked, launch it directly into the storyboard flow.
Multiple entrypoints
Start from an Analysis Mode result, a reference image, or a blank prompt. Any of these seeds an initial brief that can be shaped and iterated before storyboard is launched.
Iterate without rendering
Change the story, swap the environment, update characters, or adjust style — review each version as a brief before committing AI spend. Storyboard only runs when the brief is locked.
Brief editing
Story and style as separate controls
Ideation replaces the blank storyboard prompt with a structured creative workspace. Story and style are decoupled: swap the narrative without touching the visual language, or change the style reference while keeping the same plot. Each combination is reviewable as a brief before any render starts.
Claymation demo — analysis to final output
Analyzed a German shoe company ad built in a 1970s Will Vinton claymation style. Analysis Mode extracted the visual language. Ideation seeded it into a brief, then iterated through three story versions before locking the final: "An elderly Latina mother watches her 20-year-old son receive his college diploma; he walks over and hands it to her."
Style seed — seeded from analysis
Medium / look: 1970s Will Vinton-style Claymation — handcrafted clay figures with visible fingerprints, smear-frame morphing transitions, and the warm imperfect texture of hand-sculpted polymer clay; characters morph and squash expressively in the tradition of Vinton's Closed Mondays and Rip Van Winkle; tactile clay environments with rough-thumbed surfaces, visible armature weight, and analog stop-motion charm.
Lighting: Warm tungsten practical lighting casting soft directional shadows across clay surfaces, highlighting texture and depth; golden side-lighting that emphasizes the handmade dimensionality of sculpted faces and fabric folds; no digital glows — all warmth comes from physical light sources on physical sets.
Lens: 16mm film grain aesthetic — slightly soft focus with natural vignetting; medium shots for character expression; slow zooms for emotional payoff; close-ups that reveal the beautiful imperfection of clay skin and fingerprint texture.
Tone: nostalgic, handcrafted, warmly humanistic, emotionally resonant, analog.
Palette
Storyboard hand-off
Final brief launches into storyboard
After three story iterations with the same claymation style, the locked brief launches directly into storyboard. Six scenes — The Waiting Chair, The Walk Across Stage, Eyes Full, He Comes to Her, The Diploma Passed, Her Embrace — each with matched character and environment prompts drawn from the original analysis.
Output · Ideation → Storyboard → Render
Graduation Reunion · claymation output
The finished render from the Ideation-seeded brief: 1970s Will Vinton-style claymation, six scenes, locked after three story iterations. The style came from analysis; the story came from ideation.
Output · Ideation → Storyboard → Render
Omo · 1:30 ideation output
A 1:30 video produced entirely through the Ideation flow — 9 scenes with full narration and captions. Brief seeded in Ideation, iterated before any render spend, then handed directly to storyboard for final output.