Brand promo video · AE/PR · Voice Mode feature launch · Lovart AIGC · product demo film · voice-controlled canvas
Lovart recently brought Voice Mode to the canvas — a voice interaction layer with visual perception. While putting this demo together, my focus was clear: not to prove "voice input works," but to show that the chain of "speak, edit on the fly, and trigger image or video generation from vague instructions" actually holds up.
Traditional voice features often stop at "turn speech into text." Voice Mode is different because it can see the canvas — say "make that product on the left a bit larger," and the system knows which region you mean. That visual perception upgrades voice from an input method to a control method: direct canvas manipulation, composition tweaks, selection changes — speak and operate.
A lot of creative requests are inherently fuzzy — "make it more cyberpunk," "darken the mood," "give me three directions." Voice Mode lets users issue these kinds of creative instructions in natural language; the system reads the context and triggers image or video generation right on the canvas. You don't need to specify every parameter upfront — the conversation feels closer to how creative work actually happens.
One successful command doesn't prove much. What matters is multi-round continuous dialogue — speak, see, adjust, speak again. The key to presenting Voice Mode isn't "supports voice" — it's "continuously understand and execute." When that process is made trackable in the demo, viewers treat it as an everyday creative tool, not just a feature showcase.