Genjutsu API and batch rendering: what this site actually offers

Is there a Genjutsu API on this site?

No. This independent site does not expose a public API key, API documentation, batch submission endpoint or customer webhook. The routes used by the web app serve the signed-in generator; they are not a supported integration surface for external scripts or businesses.

You can still produce several videos manually, one submission at a time. The sections below explain what that workflow looks like, what a real production API would need and where this site’s boundary sits.

• Review the current AI tool boundaries
• Build a repeatable manual workflow

What do people mean by a Genjutsu API?

The phrase can refer to several different things: the hosted model provider’s API, an official Higgsfield integration, this site’s internal web routes, or an unofficial wrapper sold by a third party. They are not interchangeable, and a key from one does not grant access to another.

Treat unsolicited API keys, reseller credits and "unlimited" integration offers as untrusted. Use the official product or the documented provider if you need a supported integration, and verify the current terms there.

• Model-provider API: infrastructure access, not a Higgsfield customer plan.
• Official product API: verify availability and terms with Higgsfield.
• This site’s web routes: internal app endpoints, not a public API.
• Unofficial wrappers: unsupported, unstable and potentially fraudulent.

How do you batch renders without an API?

A manual batch is a shot list, not a queue. Prepare each source clip and reference image in advance, submit one job at a time, review the result before spending the next set of credits, and keep the approved outputs together for the edit.

This is slower than an API, but it catches systematic problems early. If the first three clips drift in the same way, stop and fix the reference set or prompt instead of rendering the rest of the list.

• Write a shot list with source file, image, prompt and quality per row.
• Test the first shot at 720p before running the batch.
• Submit one render, review the whole clip, then continue.
• Track credits and the displayed estimate before each run.
• Download approved outputs and assemble them in an editor.

What would a production integration need?

A real automation surface needs authentication, documented endpoints, idempotent job creation, status polling or webhooks, rate limits, storage rules, error codes and a support path. None of those are offered as a customer API on this site.

That is a deliberate boundary, not a hidden feature. Do not build a business process around browser automation or undocumented routes; they can change without notice and may violate the site’s terms.

• Authenticated API keys with scoped permissions.
• Documented request and response schemas.
• Job status, retry and failure semantics.
• Webhook signatures and delivery guarantees.
• Published rate limits, SLAs and support contacts.

Can you use ComfyUI or another automation tool?

No. There is no official Genjutsu node, plugin or public endpoint for ComfyUI, and this site does not provide one. Generic motion-transfer pipelines in ComfyUI are a different technology, not a shortcut to this hosted model.

If your goal is a local, fully controlled pipeline, read the ComfyUI guide for the honest trade-offs. If your goal is this model, use the browser generator and plan the work manually.

• Understand the ComfyUI and open-source boundary

How do credits and failed jobs work in a manual batch?

Each run shows a credit estimate based on source duration and selected quality before submission. The server recalculates from the uploaded file, so a longer clip can cost more than the first preview suggested.

Failed generations are designed to refund the credits charged for that job. Check the final job message and updated balance before retrying, especially when a browser disconnect or timeout makes the job look unfinished.

• Understand credit estimates and balances
• Diagnose failed or stalled renders

When is a manual workflow enough?

Manual batching is enough for small campaigns, creative tests and teams that need a human review between shots. It is a poor fit for high-volume publishing, deterministic rendering or a product that needs machine-to-machine handoffs.

If the work is creative and review-heavy, the lack of an API is not the bottleneck; the source footage and reference quality are. If the work is high-volume and repetitive, use a tool that documents an integration instead of forcing this one.

Key takeaways

• This site has no public API, API key, batch queue or customer webhook.
• Several videos can be produced manually, one submission at a time.
• A production integration needs auth, docs, status, retries and support.
• ComfyUI and unofficial wrappers are not supported access paths.
• Failed generations are designed to refund the charged credits.