bytedance/options

6 notes, read out of this brain and free to use. Each one was extracted from a source and is re-checked against its exam.

ByteDance image size parameter formats

The `size` parameter for ByteDance image generation accepts pixel dimensions in the format `{width}x{height}` (e.g. `2048x2048`). Resolution levels such as `1K`, `2K`, `3K`, `4K` can be passed via `providerOptions.bytedance.size`.

ByteDance image model options

Image model options available via `providerOptions.bytedance` are: watermark (boolean, whether to add watermark), outputFormat ('png' or 'jpeg', supported by seedream-5-0 and dola-seedream-5-0-pro; seedream-4-5 and seedream-4-0 always return jpeg), size (string, resolution level e.g. '1K', '2K', '3K', '4K' as alternative to pixel dimensions), sequentialImageGeneration ('auto' or 'disabled', for batch generation), maxImages (number, maximum images to generate when sequentialImageGeneration is 'auto'), and optimizePromptMode ('standard' or 'fast', seedream-4-0 supports both, other models support standard only).

ByteDance video model options - generation

Video generation options available via `providerOptions.bytedance` are: watermark (boolean, whether to add watermark), generateAudio (boolean, generate synchronized audio, supported by Seedance 1.5 Pro and 2.0), cameraFixed (boolean, fix the camera during generation), returnLastFrame (boolean, return last frame of generated video for chaining), serviceTier ('default' for online inference or 'flex' for offline inference at 50% price with higher latency), and draft (boolean, enable draft sample mode for low-cost preview, Seedance 1.5 Pro only, generates 480p preview).

ByteDance video model options - image input

Image input options for video generation via `providerOptions.bytedance` are: lastFrameImage (string, URL of last frame for first-and-last frame video generation, supported by Seedance 1.5 Pro, 1.0 Pro, and 1.0 Lite I2V) and referenceImages (string array, URLs for multi-reference image-to-video).

ByteDance video polling configuration

ByteDance video generation is task-based: the provider creates a task and polls until completion. Configure polling with the top-level `poll` option of `generateVideo()` with `intervalMs` (how often to check the task, default 5000) and `timeoutMs` (give up timeout, default 600000). Deprecated options `pollIntervalMs` and `pollTimeoutMs` in provider options emit a warning if passed.

ByteDance video aspect ratios

Supported aspect ratios for ByteDance video generation are: `16:9`, `4:3`, `1:1`, `3:4`, `9:16`, `21:9`, and `adaptive` (image-to-video only).

Give your agent this brain