Skip to content

Writer-to-Screen Convergence: block-native writing, visual production, and Afterglow acceptance harness #2165

Description

@BryanHarrisScripts

Purpose

Capture the product architecture emerging from the September Human UAT and design discussion before further implementation. The goal is not another disconnected workflow. PlotPickle should let a writer create the story in the same deterministic structure that later drives Outline, Storyboard, Previs, Scene Workspace, and Production.

Core product principle

The written Block is the creative source that progressively attracts the other production components around it.

The same 4 Acts → 12 Sequences → 24 Blocks → 96 Mini-Blocks structure should organize learning, outlining, writing, visual references, storyboard decisions, previs, synchronized scene work, and production evidence. As the work moves downstream, more components are attached to the same story addresses rather than copied into new authorities.

The Human described this as a gravitational/convergent model: the original story idea and written Block remain the center while structure, characters, references, shots, movement, audio, timing and provider instructions are progressively pulled into one final production object.

Writer-first chain

WRITE — What is the story on the page?

  • The writer works in named Blocks, using a Block title/subtitle as a human-readable story unit.
  • A feature-length screenplay can be shaped in roughly Block-sized writing increments instead of forcing the writer to think in implementation objects.
  • Each Block maps to four Mini-Blocks and exposes the relevant real screenplay text.
  • Learning/education should use the same Block/Mini-Block language, so the user learns the same structure they later writes in.

OUTLINE — What happens in this Block and why?

  • Act / Sequence / Block / Mini-Block structure.
  • Dramatic function, purpose, progression and written source material.
  • Writer can see the Block as a story unit before any visual production detail is required.

STORYBOARD — What does it look like?

  • approved/candidate character references;
  • representative poses/expressions;
  • locations, props, palette, lighting, tone and composition;
  • key visual moments associated with the Block/Mini-Block and real Scene/Beat material.

PREVIS — How does it move and play?

  • visual coverage summary first;
  • camera intent, blocking, motion, pacing, transitions and rough timing;
  • low-cost/rough motion is valid; the purpose is decision-making rather than final rendering.

SCENE WORKSPACE — How do the pieces synchronize over time?

A scene-level workbench should combine screenplay context, playback, inspection and a block/timeline surface for synchronized evidence such as:

  • Dialogue;
  • Action;
  • Shot;
  • Audio;
  • later lighting / FX / transition / camera motion where useful.

This is the place where Scene → Beat → Shot → Frame becomes visually understandable rather than an architecture phrase.

PRODUCTION — Generate/render the approved result

  • compile approved upstream story and visual intent into provider-neutral production instructions;
  • use already-selected provider/runtime routing;
  • generated output remains candidate material until Human acceptance;
  • provider prose is disposable and does not become canon by itself.

Structural relationship

  • Block / Mini-Block = stable story address.
  • Scene / Beat = variable-density authored material related to that address.
  • Shot / Frame = variable-density visual material related to that authored material.
  • technical RenderClip addresses remain production plumbing and are not creative shot quotas.
  • relationships are not forced 1:1.

Agentic mode

A writer who only wants to write should be able to ask PlotPickle to project the story downstream automatically:

Written Block → Outline interpretation → visual references → Storyboard candidates → Previs candidates → synchronized Scene Workspace → Production candidate

Agents may interpret and propose each downstream layer, but they do not silently rewrite the source story or promote candidates to accepted creative truth. The Human can accept, reject, revise, regenerate or return upstream to rewrite.

Afterglow as the reference acceptance story

Use Afterglow: Reflections of Sentience as a canonical worked example and deterministic acceptance fixture. Its written screenplay should let PlotPickle verify story-to-screen fidelity at a known Block/Mini-Block address.

A CI/UAT fixture should be able to select one known Afterglow Block and verify that the same story address survives through:

  1. correct Block title/subtitle and source text;
  2. correct Mini-Block projection;
  3. Outline purpose/structure;
  4. Storyboard visual references/candidates;
  5. Previs coverage/movement/timing intent;
  6. Scene Workspace Dialogue/Action/Shot/Audio synchronization;
  7. provider-neutral production instructions;
  8. candidate output provenance back to the original story address.

The test must verify continuity and provenance, not subjective artistic quality. Missing downstream material must remain truthfully missing rather than fabricated just to make CI pass.

Relation to existing PlotPickle architecture

This work must reuse existing authorities and compiler boundaries rather than create parallel systems:

Mathematical implication

The deterministic structure becomes more useful when every stage shares the same addresses. PlotPickle Score and future measurable signals can evaluate continuity, completeness, progression and drift across the same story graph without pretending to judge creative quality. The stronger the shared address/provenance model, the stronger the measurable structural evidence becomes.

Implementation phases

  1. Writing Block contract — formalize Block title/subtitle, source-text span/page range, four Mini-Block projections and shared Learn/Write/Outline vocabulary.
  2. Afterglow fixture — choose and freeze one or more reference Blocks with expected story addresses and source evidence.
  3. Storyboard convergence — attach visual reference/character/location/lighting/tone evidence to the same addresses.
  4. Previs convergence — attach motion/camera/timing evidence without new canon authority.
  5. Scene Workspace — synchronize Dialogue/Action/Shot/Audio against real Scene/Beat/Shot identities and source story addresses.
  6. Production compiler path — compile approved creative state into provider-neutral production instructions and candidate renders.
  7. Story-to-screen CI/UAT — deterministic continuity/provenance checks across all stages.
  8. Agentic one-click projection — allow a writer to generate candidate downstream layers from the written story, with Human review at every promotion boundary.

Non-goals

  • no forced five-page exact Block size;
  • no forced one-Scene-per-Mini-Block or fixed creative Shot count;
  • no AI authority over the writer's story;
  • no automatic acceptance of generated visual or production material;
  • no second project/store/compiler authority;
  • no subjective CI score for whether a scene is creatively "good";
  • no requirement that every writer use visual/production stages.

Acceptance direction

PlotPickle should eventually demonstrate this simple Human story:

I can write my screenplay in understandable named Blocks. PlotPickle can show me what each Block means structurally, visualize it, rough out how it moves, synchronize the scene, and generate a production candidate — while every downstream decision remains traceable to what I actually wrote and nothing becomes my accepted story without my approval.

September 17 convergence clarifications

The Story Card / Afterglow / curriculum review added three governing principles that apply across the child issues.

1. The screenplay is evidence; 24/96 is the shared structural coordinate system

  • 24 Blocks / 96 Mini-Blocks remain the canonical PlotPickle planning, learning and diagnostic addresses.
  • A source screenplay may be organized differently.
  • Source section count, page count and PlotPickle Block count must not be conflated.
  • Afterglow v9 is the most complete baseline and contains 20 explicit titled source sections; v8 is a historical complete comparison source; v10 is a later partial rewrite whose recovered source covers the opening frontier only.
  • Afterglow Golden Story Fixture: map all 24 Blocks / 96 Mini-Blocks to source evidence #2168 owns the version-aware evidence crosswalk and the distinction between Covered, Condensed/Shared, Gap/Underdeveloped and Unresolved structural responsibilities.
  • Source-density metrics support diagnosis but do not prove dramatic completeness.

2. Character Truth is not automatically audience knowledge

Character profiles/backstory can explain motivation, fear, strategy, relationships and intended emotional start/end states even when none of that material is spoken or shown directly.

The system must distinguish:

  • Character Truth / writer knowledge;
  • Character Arc State at a story address;
  • audience-visible Screenplay Evidence.

#2178 owns source-profile preservation, provenance and character-arc mapping across the same 24 Blocks. Backstory may inform diagnosis without becoming automatic exposition or canon.

3. Learn and Story use the same conceptual coordinates

PlotPickle's curriculum should explain the same concepts the writer applies in PLAN/Story Cards, WRITE and OUTLINE.

#2167 owns the contextual bridge:

This is an almost one-to-one concept-to-use relationship, not one lesson per Block. A writer should be able to move from a lesson into the relevant story position and from a Block/Mini-Block/arc checkpoint back to the teaching that explains it.

Curriculum explains and asks questions; it never dictates the creative answer or manufactures missing material.

Current child-issue ownership

No child issue may create a parallel story store, curriculum store, character authority or address system.

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions