Back to AI game news
AI game news

Hy-Nomas adds reference images to start interactive world-generation sessions

By GameSlash

The October 31 demo update starts a new world from an uploaded picture, then continues with text and camera instructions; existing sessions cannot swap in a new reference.

Tencent Hunyuan added reference-image initialization to the Hy-Nomas interactive demo on October 31, 2026. Users can start a new generated world from an uploaded picture, then continue with text prompts and camera control, according to the project's update log.

GameSlash explanatory diagram: upload a reference, initialize a new world, continue with text and camera control
Graphic: GameSlash — original explanatory diagram based on Tencent Hunyuan's demo documentation; not a screenshot or a generated-world result.

A picture becomes the starting point

The demo accepts JPEG, PNG or WebP files up to 10 MB. It captions the image, prepares the initial reference segment and then processes the user's continuation instruction. The reference initializes a new world; it cannot replace an image in a session already underway. The deployment guide describes that distinction.

GameSlash diagram of Hy-Nomas image-reference constraints:JPEG PNG WebP up to10MB, new-world initialization only, no mid-session replacement
Graphic: GameSlash, based on Tencent Hunyuan demo documentation. Not a product screenshot or generated-world result. Source: Tencent Hunyuan documentation

A world-model update, rather than a finished game

For game creators, the change offers a way to anchor a visual exploration in an existing concept image before directing what happens next. It does not establish that the output has a game's rules, persistent simulation or editable engine assets. Hy-Nomas had already released its initial code on October 21; the new event is the image-reference demo update.

View source