# Bulin 47 AI — Product overview

Canonical HTML page: [https://bulin47-ai.com/](https://bulin47-ai.com/)
Document type: public usage summary, not a verbatim page export.
Last substantive review: 2026-10-11 (Asia/Shanghai).
Purpose: Identify the preset tool, required inputs, output and example limitations.

Bulin 47 AI turns an authorized adult's photo into a character-replacement version of a preset Bulin 47 freestyle performance. The reference footage and instructions are built in. It is an independent project, not affiliated with Bulin 47 or Kevin Wai.

## Choose a task

- Create a video: read [create.md](https://bulin47-ai.com/create.md), then open [the studio](https://bulin47-ai.com/#create).
- Understand sign-in and credits: read [pricing.md](https://bulin47-ai.com/pricing.md) and check [current pricing](https://bulin47-ai.com/pricing).
- Find a result or troubleshoot a task: read [results.md](https://bulin47-ai.com/results.md).
- Decide whether you can upload, share or use media commercially: read [terms.md](https://bulin47-ai.com/terms.md) and [privacy.md](https://bulin47-ai.com/privacy.md).
- Request help or a refund: read [refund-policy.md](https://bulin47-ai.com/refund-policy.md) or [about.md](https://bulin47-ai.com/about.md).

## Inputs and outputs
Original mode uses one lead photo and aims to keep the original surrounding crowd. Custom mode adds exactly four single-person crowd photos to replace selected surrounding people. The user does not upload a seven-view sheet: the backend generates one composite character reference from the lead photo. Outputs are vertical MP4s, approximately 23.17 seconds, with the preset performance's original sound. Available exports are 480p and 720p, without an added watermark. This does not clone the user's voice or change the lyrics.

## Before submitting

Every depicted person must be 18 or older, authorize the processing and meet the [content rules](https://bulin47-ai.com/terms.md). JPG, PNG and WebP are accepted, up to 10 MB per image. Sign-in is required. A new 480p video uses 10 credits; 720p uses 22. Watching examples and selecting a photo for local preview are free, but preview does not generate a video.

## Scope and limits
This is a preset freestyle workflow, not a general prompt-to-video editor. No public developer API or MCP integration is documented; authenticated internal browser endpoints are not a public Agent API. Generation requires sign-in, enough credits and photo permission. Likeness, hands, clothes, background and lip sync can vary. Custom crowd identities may drift during occlusion or between segments. Example videos do not guarantee identical results for other inputs.

## Public examples
See [the examples](https://bulin47-ai.com/#examples). The blue-jacket example is an earlier reference-sheet experiment. The grandmother example starts from one portrait. The football Custom preview adds a manually reviewed scene reference and still has crowd identity changes; it is not evidence that this additional preparation is automatic for ordinary users.

Next: [Create a video](https://bulin47-ai.com/create.md).