HeyGen's HyperFrames becomes pro editor

HeyGen’s HyperFrames is no longer just a way to have AI agents write video compositions in code. The company’s April 2026 product update described a new HyperFrames Studio as a “free-form canvas” inside HeyGen Studio for visually arranging, editing and iterating compositions, extending a tool that had previously been framed as an open-source, code-first rendering system. That change matters because HyperFrames had been introduced by HeyGen as a framework that lets AI agents create videos using HTML, CSS and JavaScript rather than a conventional timeline editor. In HeyGen’s help documentation, the company says the product is aimed at users comfortable with code or AI coding tools such as Claude Code, Cursor, Gemini CLI and Codex. ### So what exactly changed in HyperFrames? (heygen.com) HeyGen’s May 6, 2026 product update said HyperFrames now includes a visual workspace inside HeyGen Studio, giving users a “designer-friendly workspace” for building video more like a presentation or Figma board. That language marks a shift from pure prompt-and-code generation toward direct manipulation of layouts and assets. HyperFrames’ own documentation also references “Studio Manual DOM Editing,” indicating that manual editing controls are now part of the product’s documented workflow. (help.heygen.com) The surfaced documentation page did not expose full text in the search result, but its title and description point to an inspector for manual editing capabilities, user experience and constraints. ### Why does manual editing matter for video teams? (heygen.com) HeyGen’s help center says HyperFrames originally replaced manual timeline work with programmatic composition, letting users create product videos, social clips, animated explainers, captions, overlays and data visualizations through code and prompts. That approach is fast when the first draft is good, but it can slow down when users need precise visual adjustments. (hyperframes.heygen.com) A canvas with manual positioning closes part of that gap. In practice, it means a user can keep the speed advantages of agent-generated layouts while still nudging elements, rearranging blocks and refining composition without going back through a full prompt-edit-render loop. That reading is an inference from HeyGen’s published description of a visual editing canvas layered onto a code-based rendering system. (help.heygen.com) ### Where does this leave UniFab? UniFab is addressing a different part of the workflow. The company’s product pages pitch AI enhancement rather than composition, offering upscaling to 4K, 8K and 16K, denoising, face enhancement and audio upmixing for playback improvement. UniFab says its tools are designed to restore clarity, remove noise and sharpen footage, including low-light material and compressed video. That makes it more of a finishing and restoration product than a scene-building or layout-editing environment. (heygen.com) ### Is HeyGen now competing for professional editing work? HeyGen’s own materials now place HyperFrames between two categories: AI generation and hands-on editing. The company still describes HyperFrames as an open-source rendering framework built for AI agents, but it is also marketing a studio surface for visual arrangement and iteration. (unifab.ai) For newsroom and brand video teams, that means the competitive set is widening. (unifab.ai) HyperFrames is moving closer to tools used for layout control and rapid revision, while UniFab remains focused on playback quality and enhancement after footage already exists. That distinction is based on the companies’ published product descriptions as of June 2, 2026. ### What should readers watch next? HeyGen’s HyperFrames site, help center and GitHub repository are the clearest places to track how far the manual editing layer expands from here, including any broader rollout of Studio editing controls. (hyperframes.heygen.com) UniFab’s next visible changes are likely to appear on its video enhancer product pages, where the company is already advertising model updates and cloud-versus-local processing options. (github.com) (heygen.com)