KULIANSHU

Don't just generate backgrounds. Understand the shot first.

Comment

Inspiration

Visual effects artists spend a significant amount of time making creative planning decisions before any background can be created.

Although AI image generation has become increasingly powerful, we realized that beautiful images alone do not solve production problems. Artists still need to decide camera composition, object placement, environmental consistency, lighting, season, and continuity across multiple shots.

This inspired us to ask a simple question:

What if AI could understand the shot before generating the background?

That question became the foundation of KULIANSHU.

What it does

KULIANSHU is an AI-assisted planning system for visual effects production.

Instead of generating backgrounds immediately, it first analyzes the shot, evaluates the scene, determines object placement, builds world planning, generates a structured Prompt Contract, and finally produces an Approved Planning Package for artists.

The workflow is:

Input Shot → Shot Analysis → Tree Placement → World Planning → Prompt Contract → Background Recommendation → Human Review.

Planning comes first. Generation comes later.

How we built it

We designed KULIANSHU as a modular workflow where every stage has a single responsibility, a defined input, and a structured output.

OpenAI GPT helped us organize product architecture, workflow design, technical documentation, and planning logic.

OpenAI Codex accelerated engineering organization, documentation, and rapid iteration while building the MVP.

The current MVP demonstrates the complete planning workflow before any image generation begins.

Challenges we ran into

One of the biggest challenges was changing our own mindset.

At first, we tried to build another AI image generation workflow.

Eventually, we realized that the real problem was not image generation—it was creative planning before generation.

That insight completely changed our product direction.

Accomplishments that we're proud of

We successfully transformed an abstract production workflow into a structured AI-assisted planning pipeline.

Instead of creating another image generation tool, we built a planning layer that helps artists make better creative decisions before production begins.

What we learned

We learned that better visual effects begin with better planning.

AI should not simply generate images.

It should first understand creative intent, production context, and workflow consistency before generation.

What's next for KULIANSHU

Our next goal is to expand KULIANSHU into a complete shot-aware planning platform.

Future development includes multi-shot continuity analysis, world consistency, production pipeline integration, and support for multiple AI generation engines while keeping the planning workflow unchanged.

Built With

Share this project:

Updates