Player not loading? Watch on YouTube
This webinar presents Wan 2.7 image and video capabilities, then explains three creator workflows built around its APIs. The examples use cloud services through Alibaba Cloud Model Studio; the session does not demonstrate how to run models locally.
Ryan from the Wan team describes image generation, reference-based editing and color control. He says the image model supports 3,000-token prompts, 12 languages, sequences of up to 12 images and up to nine reference images. His video presentation covers localized edits, changes to dialogue and motion, and continuation guided by a target final frame. These are the team's capability claims rather than independent benchmark results.
Benjamin Law explains how he used OpenClaw to build an API client for batch e-commerce image editing and video generation. His workflow supplies input paths, output locations and reusable project instructions to an AI agent.
David demonstrates an n8n construction-video workflow built around initial, midway and completed images. Users can approve results or edit prompts before the workflow generates and merges video segments. He advises keeping human review because generated prompts can miss the intended result.
Zachary Huang uses his open source PocketFlow framework to turn markdown into educational cartoons. The pipeline plans scenes, uses character references for consistency, generates voices and animation, then joins clips with FFmpeg. He stresses that teaching examples and narrative flow still need human supervision.
