Dola turns product screenshots into launch campaigns with Seedream 5.0 Pro

• One product paragraph and one screenshot produced three campaign routes in MengTo's Dola demo.
• Posters combine Seedream 5.0 Pro scenes with real product UI, according to MengTo's walkthrough.
• Seedance 2.0 turned a finished poster into a five-second launch clip in the motion demo.

One ThreeUI campaign route came back as “774 components. Zero boilerplate.” ByteDance's Seedream 5.0 Pro announcement also describes layer separation for editing generated images.

MengTo gives Dola two inputs in his desktop run:

• A paragraph describing what the product does and who it serves.
• One screenshot of the actual website.

His route-generation prompt, broken into its requested fields:

[Product] does [what] for [who]. Give me 3 launch routes, each with:

MengTo describes the poster pass in three parts:

1. Seedream 5.0 Pro builds a scene from the route's hero idea.
2. Dola places real screenshots from the site onto that scene.
3. Dola sets the headline in clean type.

Keeping the actual interface is the workflow's best detail.

MengTo animates the “3D for designers.” poster into a five-second clip with Seedance 2.0 inside Dola.

The desktop app syncs with the phone app, letting MengTo review routes away from his computer. At the time of his post, he said the agent and image generation were free, while video generation used credits.

Dola's App Store listing describes daily limits on free image and video generation, alongside paid subscriptions for more frequent use. “Free” comes with usage caps.

In a separate demo, MengTo made seven browser-based 3D worlds with Viggle's PINOC, running on its JST-2 model:

• Inputs: One image per world and one per character.
• Generation: PINOC turned the images into Gaussian-splat worlds and rigged characters.
• Direction: Claude Code, connected through PINOC's MCP server, walked a character named Fern through the worlds and controlled the camera.