YICHENAI

[YICHENAI @YICHENAI2026]

I built a UGC Director Skill for Codex: turn product details into scripts and prompts, then use Nano Banana 2 for reference images and Seedance 2.5 for video.

One handheld fan, different characters, settings, and stories. Sharing the results and the workflow.

[YICHENAI @YICHENAI2026]

Following my Amazon visuals, here's an AI-made ad for the same handheld fan.

Lighting brings out texture, airflow shows power, and everyday scenes convey cooling.

The film and process: product benefits, storyboards, AI visuals, editing, and sound design.

[YICHENAI @YICHENAI2026]

Revisiting an Amazon visual project I built: one product image expanded into a full set of visuals through an AI workflow.

I turned it into an interactive webpage and used Opus 5.5 to create a motion showcase, sharing both the results and the process.

[YICHENAI @YICHENAI2026]

I needed a canvas for my AI image and video work, so I built Yichen AI Canvas.

References, prompts, and outputs in one place. An agent writes shot prompts and connects nodes, from product photos to videos.

Built for myself, now used by clients. Still improving.

[YICHENAI @YICHENAI2026]

AI images still need revisions. I built a Codex Skill to rebuild flat images as layered PSDs, with editable text, replaceable products, and separate backgrounds.

Here's a phone case ad demo: from an AI image to a file you can keep editing in Photoshop.