Turn a first encounter
into a first chapter.

Authorized manga discovery videos.
Software-assisted production. Human editorial control.

A paid, bounded pilot for one title—designed to send interested viewers to your official reading or buying page.

A recap tells enough story
to make someone curious.

Selected panels, a clear narrative, deliberate pacing, sound and captions.

The proposed publisher version introduces a protagonist and a conflict, stops before the resolution, and invites viewers to continue reading. It is an approved excerpt, with an agreed spoiler boundary.

The destination is reading.
The video is the invitation.

A reader-acquisition experiment for a title you can authorize.

Useful for a series launch, a backlist discovery campaign, or an official reading-platform promotion. The publisher supplies its channels or media budget; the offer is creative production, not an assumed influencer audience.

One title can become
a coherent creative package.

A proposed 90–180 second landscape teaser, two vertical cutdowns, and two opening-hook variants.

Captions, approved narration and music, a thumbnail, and a title-specific call to action. One language and territory; two bounded rounds of consolidated feedback.

Final scope follows feasibility. Distribution, ad spend, outside talent, further languages and engine/source-code transfer are separately agreed.

There is a real foundation.
Production readiness is still open.

The July 2026 manga archive contains an editor and processing/rendering code.

Page and panel tools, a scene-node editor, timeline controls, crop/pan operations, audio rendering and an FFmpeg video renderer are present in the inspected source.

No finished manga recap export or current runtime test was found in the inspected archive. Existing code is evidence of engineering work; it is not evidence of reliable delivery today.

Make editorial decisions
explicit before rendering.

A reviewable chain from approved source to approved film.

Page understanding suggests panels, bubbles and ordering. A person checks the interpretation, writes or approves the story, and sets pacing. A scene graph keeps image treatments and timing inspectable.

The archived source contains detection and scene processing code. Accuracy, present integrations and repeatability still require validation.

Editable reconstruction is
a research track, not a promise.

The current drawing project contains a design conversation and implementation plan.

It proposes mixed semantic, procedural and raster layers; a frozen eye-template experiment; and assisted bubble cleanup. The inspected project contains planning documents, not an implemented reconstruction engine.

Hidden pixels would be inferred or generated—not recovered original artwork. This research is optional to a conservative pilot using approved source art.

Stop where the reader
wants the next page.

“The last train has left.
Then a voice calls from the dark.”

Open on the setting, move closer to the character’s reaction, then reveal an unresolved encounter. End with the official title name and a clear invitation to read.

Illustrative invented story and motion treatment. A real pilot would use only a publisher-approved script and assets.

Your story stays
under your approval.

Agree the permissions and the creative boundaries before production.

Title-specific pages, translation, adaptation, edits, platforms, territories, term and paid/organic use. Approve model services and data handling. Keep the default artwork original to the title; no replacement drawings or cloned voices without specific agreement.

Accept the film on quality.
Evaluate the campaign on readers.

Choose the primary outcome before launch.

For a reading platform: cost per qualified new reader reaching an agreed chapter milestone. For direct sales: attributable purchases and contribution. Track the observable funnel, compare with existing creative under similar conditions, and report insufficient traffic honestly.

Production acceptance uses approved story accuracy, cleared rights, agreed files and tested links. No guaranteed views, conversion or sales.

Fund a useful proof.
Then make a go/no-go decision.

Start with a capped, paid feasibility milestone.

You provide one cleared title, a marketing owner, source assets and a destination. Arthur delivers a creative outline, a short proof sequence, a rights/data workflow and an honest production estimate.

Price follows scope and sustainable effort. Any further development is funded separately; reusable engine IP and client usage rights are explicitly agreed.

The next illustration
should be an actual cleared proof.

Approve the proposition. Select permitted material. Build one convincing sequence.

A cleared sample is the missing sales evidence. Publisher outreach remains conditional on Arthur approving this presentation; no messages or forms have been sent.