ShotWise

Inspiration

As a filmmaker and trainer, I regularly see students and emerging directors choose shots without being able to explain why.

A dolly move because it looks cool. A close-up because that's what the last film they watched did. The shot list gets built, but the intent behind each choice is often missing — and without a full crew of department heads to pressure-test those decisions, that gap can go unnoticed until it's expensive to fix.

ShotWise came from wanting a tool that makes a director articulate the why behind a shot while also confronting the can I actually shoot this? reality of budget, equipment, crew, and time.

What It Does

ShotWise is an AI directing companion that turns a script into a production-ready shot list.

A director starts with a scene and defines their real-world constraints — budget, available equipment, crew experience, and production requirements. ShotWise uses Google Gemini at runtime to analyze the scene and generate shot recommendations, reasoning through:

• Framing
• Lens
• Camera angle
• Camera movement
• Lighting
• Creative intent
• Equipment and crew requirements
• Practical setup considerations

Each shot recommendation is generated live by Gemini based on the specific scene and constraints supplied — rather than selected from a fixed template.

Recommendations can include a previsualization frame so the director can evaluate the proposed composition visually, along with relevant reference material for further study.

Once refined, the shot list can be locked and exported for the production team. ShotWise also supports a Master Slate, allowing multiple scenes to be organized within a single production rather than treated in isolation.

How We Built It

ShotWise was built as a working web application using Replit, with Google Gemini providing the intelligence behind the directing workflow.

Scene and production information is sent to Gemini at runtime using the google-genai SDK. Gemini analyzes the scene and generates shot recommendations and reasoning dynamically. That model reasoning is then integrated into a structured production workflow covering scene management, shot organization, visualization, references, production constraints, and export.

Challenges

The core challenge was getting an AI model to reason about creative intent and production reality simultaneously.

A visually compelling shot isn't necessarily a useful recommendation if it requires equipment, crew, time, or lighting resources that aren't available. The challenge was designing the workflow and prompts so that Gemini doesn't simply suggest the most cinematic option, but weighs that option against the constraints of the actual production.

Building the application also meant iterating on how AI-generated reasoning should be presented to a filmmaker: enough detail to understand and challenge a recommendation, without turning the directing process into a wall of AI-generated text.

Accomplishments We're Proud Of

We're particularly proud of getting Gemini to generate scene-specific, constraint-aware shot reasoning rather than generic coverage.

The system can take the same creative intent and produce different recommendations depending on the production realities supplied by the director — making the AI useful as a decision-making partner rather than simply a shot generator.

What We Learned

Building ShotWise reinforced the value of giving an AI agent explicit constraints and a structured workflow.

The most useful output isn't necessarily the most cinematic option. It's the strongest creative option that can actually be made.

What's Next

The next step is moving ShotWise from a directing companion into a deeper production intelligence layer — connecting creative decisions to scheduling, equipment, crew availability, and continuity across an entire production.

Built With

• google-gemini
• google-genai
• node.js
• postgresql
• react
• replit
• replit-agent