Uwarp

OiiOii — a seven-agent AI animation pipeline that turns one prompt into a scripted, storyboarded and scored short film while keeping characters on-model across every shot

Describe an idea and seven agents script, storyboard, animate and score it into a short film — characters stay on-model across every shot.

  • AI Animation
  • Multi-Agent Pipeline
  • Character Consistency
  • Short Film
Publisher
HOGI.AI
Type
AI Animation Agent
Pricing
Freemium
Reviewed
24 September 2026
Official site

Quick verdict

Use when

  • The thing you want to end up with is a sequence rather than a clip, and the same character has to stay recognisable through every shot of it
  • You have an idea, a script or a video you want restyled, rather than finished artwork or a shot list you have already blocked out
  • You would rather brief the job than operate it: the script, the cast, the scenes, the storyboard, the shots and the score all come out of one pass
  • The look matters more than any single engine, so you want the pipeline to pick a rendering backend per shot instead of committing the whole piece to one model
  • You are producing animation at volume for social channels and want many shot generations running at once rather than waiting on them in turn
  • You expect to revise one shot without reshooting the film, and you want the timeline to reflow around the change
  • You would rather hand setup to a coding assistant than click through a studio, since the platform ships a CLI and agent skills for exactly that

Skip when

  • You need photoreal footage of real people, because this is built around animation styles rather than cameras
  • You want to author camera moves, blocking or a continuous take yourself — the pipeline generates shots and sequences them, it does not hand you a rig
  • You need frame-level command of an edit, a grade or a composite
  • You are making a single short clip and consistency between shots is irrelevant, which is a problem one model already solves
  • Your output is not motion at all — a document, a page or an interface gains nothing from an animation pipeline
  • You need everything to run on your own machine or offline, since generation is a hosted service

Try instead

If you would rather drive the shots yourself than hand the whole pipeline to agents, or you only need one clip rather than a sequence, those live here too.

OiiOii vs LTX Studio vs Runway vs Pika

OiiOii is the odd one out in this shortlist, because it does not sell a model or a clip — it sells a team of agents that writes, casts, storyboards and scores a whole short across 25+ rented engines, and meters you by the credits those engines burn. LTX Studio is the closest rival on that premise: a storyboard-first studio where saved Elements carry a character between shots, but a person still drives it. Runway and Pika sit at the other end of the axis, single-clip generators with the most control per shot and the shortest institutional memory of a character. The split that decides it is whether you want a sequence you supervise or a film you brief.

Tap a dimension to focus

Pricing

Roughly even
  • OiiOiiThis page
    • Credit is the only currency, and it is spent by what a shot actually consumes — number of storyboard panels, chosen engine, resolution and seconds rendered
    • A daily free allowance plus a sign-up bonus covers evaluation; the video engines are reserved for paid plans and images cost roughly half as much on one
    • Subscription credits are valid for about a month while top-up packs last a year, and the balance is drawn down oldest-first
    • Concurrency is sold as concurrent storyboards and scales with the tier, from ten on the entry plan upward
    • Business tiers add seats, shared credit pools and project permissions; high-volume contracts are quoted rather than listed
  • LTX Studio
    • The free tier issues a single one-time credit grant rather than a recurring allowance, so free is a trial rather than a small monthly budget
    • Paid plans are built around a monthly credit balance that is discounted on annual billing and rises several-fold between entry and Pro
    • The lower tiers carry a personal-use licence only, with commercial rights starting at the mid tier — a real cost step for anyone working for a client
    • Top-ups are sold on the upper tiers rather than every plan, and a refund window exists while usage stays under a published credit ceiling
    • Enterprise is quoted rather than listed
  • Runway
    • The free plan is a one-time credit deposit that does not expire — enough to test the models, not enough to finish anything
    • Paid tiers are monthly credit allotments discounted annually, with per-model burn published as credits per second, per image and per song
    • Monthly credits reset on the billing date and do not roll over except on the top tier, where one month carries forward; separately purchased credits never expire
    • Extra credits can be bought from the dashboard with a minimum purchase, so running out is a top-up rather than a plan change
  • Pika
    • The free plan carries no recurring credits at all — it can generate, but only from credit packs you buy, which makes it a demo rather than a standing free tier
    • Paid tiers step up monthly credits, parallel generations and priority queue access together, so throughput is part of the price
    • The top tier is a price range rather than a single number, which reads as sized-to-workload rather than fixed
    • Commercial licensing starts at the mid tier, and payments are non-refundable once a period has begun
  1. OiiOii — official site
  2. OiiOii — how the seven agents divide the work
  3. OiiOii — plans, credit rules and per-model rates
  4. OiiOii — FAQ
  5. OiiOii — script-to-film pipeline
  6. OiiOii — animation canvas and models
  7. OiiOii — video models and frame reference
  8. OiiOii — character design
  9. OiiOii — terms of service
  10. OiiOii — privacy policy
  11. LTX Studio — plans and credits
  12. Runway — plans and credits
  13. Pika — plans and credits

Preview of OiiOii - not the live app. Confirm details on the official site.

Does this overview help you decide?

—

(—)

Learn more

Details below the decision summary—features, workflow, and scope notes.

What is OiiOii?

Describe an idea and seven agents script, storyboard, animate and score it into a short film — characters stay on-model across every shot.

What it costs

Free daily credits cover evaluation, then plans are metered in credits that are spent by the shot — engine, resolution and seconds all move the price.
Free tier
Yes
Pricing summary
Pricing here follows the shape of the work rather than a seat count. Credit is the only currency, and it is not sold as a flat allowance you can predict from the tier name: the platform deducts it by task complexity, which in practice means the number of storyboard panels in a project, the resolution and length of each shot, and above all which rendering engine that shot was assigned. That last variable is where budgets are actually won and lost, because the model lineup spans a very wide band — a fast cheap engine and a flagship cinematic one can differ by an order of magnitude for the same number of seconds, and the agents pick per shot. The published reference table lists a per-second credit rate for each video model at each resolution, an image rate per model at each resolution, and a music rate per track, which is enough to estimate a project before spending anything, provided you know which engines the pipeline will choose. Membership changes the arithmetic as well as the allowance: image generation costs roughly half as much on a paid plan, and the video engines are reserved for members outright, so the free path is genuinely for evaluating rather than for shipping. Access starts with a daily credit allowance plus a sign-up bonus rather than a locked demo, which is enough to see how the agents behave on a short piece. There is also a short promotional trial plan aimed at new users. Above that, personal plans step up both the monthly credit balance and the per-credit rate until the rate reaches its floor, and the tier ladder buys concurrency rather than features: concurrent storyboards climb with each plan, which is the number that decides how fast a multi-shot project actually renders. Business plans add multi-seat management, shared credit pools, project-level permissions and a shared asset library, with high-volume contracts quoted rather than listed. Three terms deserve attention before you commit. Subscription credits are valid for about a month, while top-up packs last a year, so unused subscription credits do not carry forward indefinitely. The system draws down the balance oldest-first, which is protective rather than arbitrary. And refunds are narrow: promotional purchases are explicitly non-refundable, and a regular purchase qualifies only while nothing has been consumed. Auto-renewal also depends on the payment method rather than the plan, so it is worth checking rather than assuming. Confirm current rates, allowances and concurrency limits on the official plans page, because the per-second model rates are the part of this that moves most.

Reviewed on 24 September 2026 · OiiOii — plans, credit rules and per-model rates

What OiiOii includes

The capabilities as the product documents them, read against what each one actually removes from an animation workflow.

  • Seven agents with named jobs, not one model with a prompt

    An art director sets style, palette, lighting and aspect ratio and then chooses the rendering engine for each shot. A scriptwriter expands an idea into scenes, beats, dialogue and voice-over. A character designer locks faces, wardrobe and proportions. A scene creator builds reusable environments. An animator turns the storyboard into moving shots. An editor sequences and paces them. A sound engineer scores and mixes. The reason the split matters is that consistency in animation is an information problem before it is a generation problem, and each of these is the piece of information the others need.

  • A shared scene graph that survives the whole film

    The agents work from a common internal model rather than handing clips down a line, which is what lets characters, environments and visual style persist across dozens of shots instead of being re-guessed at each one. Editing the script updates that graph and everything downstream reflows, so a change to a scene is a change to the film rather than a patch on one clip.

  • Characters that do not drift

    The character designer produces a multi-pose sheet — front, three-quarter, side, back, plus expressions — from a text description or a reference image, and those identifiers are passed as constraints to every shot the character appears in. Face drift is the failure people actually notice in AI animation, and this is the specific mechanism aimed at it. Scenes get the same treatment, carrying time of day, weather, mood and key props.

  • 25+ engines behind one pipeline, chosen per shot

    The roster spans flagship cinematic models, faster cheaper ones, image models and a music model, and the art director picks per shot rather than locking the project to one engine. That is a real capability and also the thing that makes cost hard to predict, which is worth understanding before you brief something long.

  • 150+ built-in styles

    A large preset library spanning anime and manga, 3D, claymation, stop-motion, Ghibli-adjacent painterly looks, pixel art, ink, watercolour and a long tail of niche aesthetics, including styles modelled on recognisable published art directions. It removes the style prompt from the problem, at the cost of inheriting whatever each preset was tuned on.

  • A canvas with per-shot regeneration and linked editing

    Storyboard panels stay in sync, so changing one element updates the related scenes rather than leaving the sequence inconsistent. Individual shots can be regenerated in place and the timeline reflows automatically, which is the mechanism that turns a take-it-or-leave-it generation into something you can actually direct.

  • Batch rendering and a concurrency ceiling you buy

    Shot generations run in parallel rather than one at a time, which is the difference between a storyboard being an afternoon and being a week — but the number of concurrent storyboards is set by your tier, so throughput is a commercial decision as much as a technical one.

  • Rendering, sound and score in the same pass

    Music is generated against the story’s emotional arc and diegetic effects are matched to on-screen action, with levels ducked under dialogue and cuts matched to beats. You can bring your own score instead and the sound engineer will mix it, which is the more useful half of this for anyone with an existing composer or library.

  • A canvas model plus a CLI and agent skills

    Alongside the browser workspace, the platform ships a command-line path and skills that let a coding assistant drive generation — picking a workflow, running it and troubleshooting symptoms. For a product whose value is orchestration, exposing the orchestration to an agent rather than only to a UI is the coherent choice.

How to brief an agent pipeline instead of operating a studio

The setup that works, and the five places a multi-agent animation run goes wrong.

  1. Decide whether you are briefing or directing before you start

    The honest question is whether you want to supervise shots or specify a film. If you want to place a camera, cut on a beat and grade a shot, this is the wrong shape of tool and no amount of prompt engineering will change that. If you want to describe a story and receive an edit, then leaning into the pipeline — giving it the whole thing rather than a shot at a time — is what produces the result it was built for.

  2. Lock the cast and the look before you render anything expensive

    The art director inherits style to everything downstream and the character designer generates the multi-pose sheets everything else references, so both are cheap to change early and very expensive to change late. Settling the palette, the aspect ratio and what the protagonist actually looks like while the project is still a script is the single highest-leverage thing you can do.

  3. Bring a script if you have one

    A one-line idea works, and the scriptwriter will build a structured screenplay from it. But if you already know the story, supplying it parses into the same internal scene graph the agents share, which means the narrative beats are yours rather than reconstructed. The same applies to character reference images, scene references, voice-overs and music — everything you bring is one less thing guessed.

  4. Estimate the credit cost from the model table before you commit

    The published per-second and per-image rates are the only way to predict a budget, because cost is driven by which engines the art director selects and you do not choose those directly. Do a short piece first at the resolution you actually intend to ship, then multiply. The gap between the fastest and the flagship engines is the largest single variable in the whole exercise.

  5. Watch the concurrency ceiling, not just the credit balance

    A multi-shot project is limited by how many storyboards can render at once as much as by credits, and that number is set by your plan. A project that feels instant on a high tier can feel stalled on a low one with plenty of credits left, which is the constraint people discover after upgrading for the wrong reason.

  6. Treat the output as a cut you curate, not a render you approve

    The workflow that gets results is regenerating individual shots that miss and letting the editor reflow, rather than judging the whole film and starting again. Non-destructive shot regeneration only pays off if you are willing to use it shot by shot — and if you are not, a single-clip tool with more control per shot would have been the cheaper choice.

Who OiiOii is for

The builds this product actually maps onto.

  • Solo creators making serialised animation

    Short-form channels that publish episodes rather than one-offs are the clearest fit, because the value compounds: a locked cast and reusable scenes mean episode four costs less than episode one, which is exactly the economics a single-clip workflow never delivers. Batch rendering keeps a publishing cadence feasible for one person.

  • Marketers with a story rather than a product shot

    Explainer animation, brand films and campaign pieces where the message needs characters and a narrative arc rather than a montage of footage. The pipeline removes the storyboard and score from your critical path, which is usually what makes this kind of work expensive rather than the animation itself.

  • Game studios producing cinematics and cutscenes

    Anything that needs a consistent cast, a consistent world and many shots against a deadline that does not allow a full animation pass. The style presets cover the game-adjacent looks, and the CLI plus agent skills mean the pipeline can be driven from a build or a script rather than only from a browser.

  • Educators and explainer producers

    Teaching material, training simulations and lesson content where the same narrator or presenter has to recur across a series. Consistency is the requirement that breaks naive AI video workflows for this kind of work, and it is the one this product is engineered around.

  • Teams that would rather brief an agent than hire a pipeline

    If the constraint is time and headcount rather than control, paying a platform to run script, storyboard, render and score in one pass is cheaper than assembling the equivalent chain of tools — provided you accept that you are curating the result rather than authoring it.

When OiiOii is the right pick

The useful way to place OiiOii is to notice what it does not sell. It is not a model and it is not a clip generator; the thing you are buying is coordination. Every other tool in this category asks you to hold the film in your head — which engine for this shot, what the character looked like two shots ago, whether the light in the third scene matches the first. OiiOii’s bet is that those are jobs for software, and its architecture reflects that: characters become locked sheets, scenes become reusable environments, style decisions are set once by an art director and inherited by everything downstream, and the rendering engine is treated as an interchangeable part chosen per shot. That is genuinely different from a prompt box, and it is what makes a one-minute narrative with a recognisable protagonist feasible for one person. It follows that the product is at its best when the sequence is the point. A short film, a manga-style episode, a game cinematic, a serialised social series — anything where shot four has to look like shot one. It is also the right call when you would rather brief than operate: if describing the story to something that returns a finished edit is the workflow you want, none of the single-clip tools will give you that, however good their individual frames are. The trade is control, and it is a real one. There is no rig, no keyframe-level camera work, no compositing, no frame-accurate grade. You can regenerate a shot and reflow the timeline, but you are curating outputs rather than authoring them. Two further caveats are worth stating plainly rather than burying. Cost is hard to predict before you know which engines the pipeline will pick, and the difference between a cheap and a flagship engine is the single largest variable in a budget. And the commercial and data terms are thinner than the tooling: no enterprise privacy framework or no-training commitment is published, the policy pages sit behind a locale-scoped consent gate, and the operating entity is not named on the site — all of which matters if this is going into a client deliverable or a procurement review. Where it earns its place is any workflow that has been bottlenecked not by the ability to generate a good frame but by the ability to keep a hundred of them coherent, and where the person doing the work is a writer, a marketer or a solo creator rather than an animation team.

Platform and product notes

What it is, what it will not do, and the facts worth verifying at the source.
It orchestrates other companies’ models, and charges you for them
The platform’s own layer is the agents and the shared scene graph; the frames come from a roster of third-party engines that includes several flagship models from other labs, alongside image and music models. That is why the credit table is long and why cost tracks model choice so closely — you are paying pass-through rates with a margin, and the published per-second figures make that visible if you read them.
It is an animation platform, not a video editor
The output is assembled and scored for you, but there is no frame-accurate timeline, no compositing, no colour grading and no camera rig. Shot-level control goes as far as regenerating a shot and reflowing the sequence. If your work needs a grade or a composite, it needs a different tool at the end regardless of where the footage came from.
Commercial rights arrive with a paid plan
The product documents paid plans as carrying commercial usage rights, with the exact scope per plan set by the terms rather than by the pricing page. That is normal for the category, but it means the free tier is for evaluation and personal experiments, and a client deliverable should be planned on a paid tier from the start.
Generation is hosted, and the data terms are thin
Everything runs on the vendor’s servers, so scripts, references and prompts are uploaded rather than processed locally. No enterprise privacy framework, data-residency option or no-training commitment is published, and the policy and terms pages are locale-scoped and consent-gated. For personal and small commercial projects that is usually acceptable; for anything going through a security review, read the terms first rather than after.
The operating entity is not named on the site
Support runs through the operator’s own domain and the press material attributes the product to the same company, but no registered legal entity, address or jurisdiction is published on the site or in the policy pages. If your procurement process needs a counterparty with a name on it, that is a question to resolve before signing off, not one this page can answer.
Credits expire, and refunds are narrow
Subscription credits are valid for roughly a month and top-up pack credits for a year, with the balance drawn down oldest-first. Promotional purchases are explicitly non-refundable, and a regular purchase only qualifies for a refund while nothing has been used. Credits are non-transferable and cannot be redeemed for cash, and auto-renewal depends on the payment method rather than the plan — all of which is worth reading rather than assuming.
Published rates and limits are vendor-stated
The per-second model rates, credit allowances, concurrency ceilings and generation times on this page come from the vendor’s own pricing and FAQ pages and are not independently audited here. They are the right things to re-check at the source before budgeting, because model rosters and credit rates are the part of this product that changes most often.

Frequently Asked Questions

Quick answers about this tool—open a question to read more.