Wan 2.7 Video

Overview

wan2.7-video/text-to-video is a text-to-video model in Alibaba's WAN video series, called through the unified HiAPI.ai interface. Simply input a text prompt to generate a video clip with visuals and an optional audio track. It supports five aspect ratios — 16:9, 9:16, 1:1, 4:3, and 3:4 — with an adjustable duration of 2–15 seconds, a choice of 720P or 1080P resolution, and support for negative prompts, random seeds, and prompt rewriting. It is suited to developers who need to drive short video assets from plain text, for batch content production and creative storyboard previews.

Key Highlights

• Programmable text-to-video parameters — duration, aspect ratio, resolution, and seed are all passed in as API parameters, making scripted batch generation and result reproduction easy.
• Optional audio track and prompt rewriting — a wav/mp3 audio URL can be passed in as the audio track, and prompt rewriting is enabled by default to fill in visual details.
• Negative prompts constrain the visuals — supports negative prompts of up to 500 characters to exclude unwanted elements and reduce the rate of unusable outputs.

Capabilities

Generate video clips by parameter: Generate video from a prompt together with duration (2–15 seconds), aspect ratio, and resolution tier. Output specifications are determined at request time, reducing later cropping and transcoding.

Prompt rewriting and negative constraints: prompt_extend is enabled by default and can expand short descriptions; negative_prompt is passed in separately to exclude specific visual content. Together they improve the controllability of outputs.

Seed reproduction and stable results: The seed parameter supports reproducing or fine-tuning generation results under the same prompt, making it easy to compare the effects of different parameter combinations.

Attached audio and optional watermark removal: An optional audio URL serves as the input for the generated audio track, and the watermark switch controls whether an AI-generated watermark is added, letting you choose the finished form as needed.

Use Cases

• Content teams producing short video assets in bulk — Use scripted prompts to batch-generate 2–15 second clips, producing vertical or horizontal short video assets that can be published directly or edited further.
• Creative storyboarding and previews — Use the same prompt with different seeds to rapidly generate multiple visual versions, producing dynamic preview assets for pitch comparisons and shot design confirmation.
• Social media shorts with music — Generate finished clips with an audio track by combining a prompt with an audio URL, producing short content ready for publishing on social platforms.

Limitations and Selection Advice

A single generation is capped at 15 seconds; longer videos must be generated in segments and stitched together, with narrative continuity depending on post-processing. Currently only text-to-video is supported; if you need to constrain visual consistency with images or reference material, use the image-to-video route in the same series. Resolution tops out at 1080P, so scenarios requiring high-resolution finished videos need to evaluate post-production upscaling options. Prompts are capped at 5000 characters and negative prompts at 500 characters; overly long scripts must be split across requests.