Frequently Asked Questions

No subscriptions, no monthly fees. After registering, top up credits and consume on a pay-as-you-go basis. US$1 = 543–628 credits (volume-tiered, transparent pricing; in CNY: ¥1 = 100 credits); you can top up any amount (US$29–7,000, or ¥200–¥50,000). Top-ups never expire.

Credits are the universal unit (US$1 = 543–628 credits; ¥1 = 100 credits). Image generation costs 31 credits/image, voiceover (TTS) 49 credits per 1,000 characters, video is billed by resolution and duration (95 credits/sec at 480p, 212 credits/sec at 720p), and copy is counted by tokens. Compositing and stitching do not consume credits. A ~2-minute episode (around 26 shots) costs about ¥150 at 480p or ¥320 at 720p.

Top-up credits never expire and will not be cleared — they remain in your account until used.

Yes. You receive trial credits when you sign up so you can run a full workflow before deciding to top up. All features are fully available and there is no watermark.

WeChat Pay, Alipay, and bank card; overseas users can top up with USD bank cards — the options are shown automatically by region.

An order receipt is generated automatically after top up. For VAT special invoices in Mainland China, email [email protected].

Platform Key: We provide an enterprise-grade API account — you consume against your topped-up credits, simple and convenient.
Self-provided Key (BYO): You enter your own Volcengine/OpenAI/MiniMax key; AI fees go directly to the vendor and do not consume your credits. The platform does not take a margin.

Text: OpenAI, Grok, OpenRouter, and custom OpenAI-compatible endpoints.
Image: Gemini 3, Volcengine Seedream, OpenAI gpt-image-2.
Video: Doubao Seedance, Vidu, Alibaba Tongyi Wanxiang.
Speech: MiniMax TTS.

Claude models can be accessed via OpenRouter or a custom OpenAI-compatible endpoint.

Planned. We'll first support ComfyUI remote nodes, then add built-in local inference (requires a discrete GPU).

Yes. We use a "character sheet anchor + reference image injection" strategy: generate a unified character sheet, then automatically inject references into subsequent images and videos to lock facial features and wardrobe, maintaining cross-shot and cross-episode consistency (cross-shot character consistency).

Single-shot duration is limited by the selected vendor model (about 15 seconds max). Our planning will automatically split shots at vendor limits; longer narratives are achieved with continuous long takes and automatic stitching of multiple segments.

One device at a time. Logging in on a new device will automatically log out the old device and send an email notification. When switching Macs or reinstalling the OS, just log in again once.

Project data is stored locally. Copy ~/Library/Application Support/Tianying/ to migrate when changing machines; projects with cloud sync enabled will be restored automatically. Account top-up credits follow your account and are unaffected by device changes.

Click "Forgot password" on the login page, enter your email and you'll receive a reset link valid for 1 hour.

Yes. Go to "Account → Delete" — the account can be recovered within 30 days, and will be permanently deleted after 30 days. Any remaining topped-up credits will be forfeited upon deletion.

When using the cloud, project data is stored in your account space and used only to generate and composite for you; when using the macOS client offline, scripts, characters, and storyboards are stored locally. When AI is used for generation, prompts are sent to the selected AI vendor (this is required for AI calls and is the same for all AI tools).

Project metadata (script text, storyboard parameters) and generated assets (images/videos, stored in object storage). All transfers are encrypted over HTTPS.

Primary service nodes are in Tokyo; video compositing nodes are distributed, and static assets are CDN accelerated. Services are not located in Mainland China and there is no ICP filing.

We support enterprise custom deployments. Email [email protected] for pricing.

The web app works across platforms and is available upon registration; a Windows desktop client is planned.

We support only Apple Silicon (M1+). Intel Macs lack VideoToolbox hardware encoding and will perform poorly.

The first launch automatically detects and installs it. If you installed it via Homebrew, the client will use that installation.

The installer is about 400 MB; runtime storage grows per project. A typical episode project (including source video and intermediate compositing artifacts) is about 200–500 MB.

All data is in ~/Library/Application Support/Tianying/data/. Copy that directory to back up.

The client core is planned to be open-sourced once the product matures; cloud services will remain closed source. Skills (Agent prompts) are open from day one.

Copyright belongs to you and commercial use is permitted. However, AI model service terms are set by each vendor and should be checked by you (OpenAI / Gemini explicitly permit commercial use).

Technically yes, but creating defamatory content, deepfakes, or using portraits without authorization is prohibited. Violations may result in account suspension.

Yes. The agency / affiliate commission system is live — email [email protected] to enable.