AIVI is an AI video studio that turns ideas and raw audio into cinematic videos with consistent characters and brand styles. Itamar Katan designed and built it end to end as its sole creator.
AIVI is an advanced, AI-powered video creation studio designed to transform static ideas and raw audio into professional-quality cinematic videos.
The motivation behind AIVI stems from two major pain points: the "Cold Start" problem and the Control vs. Automation trade-off. Current tools are "black boxes"; I wanted a solution that balances AI heavy-lifting with human creative intent.
As the sole creator and lead developer, I owned the entire lifecycle—from product definition and market research to high-fidelity UX/UI design and full-stack orchestration using React 19, Firebase, and AI Pipelines.

The "Last Frame" Bridge: To solve the common AI problem of "visual jumping" between scenes, AIVI implements a Last Frame Continuity logic. By injecting the final frame of a previous scene as a reference for the next, we ensure a fluid narrative flow.
Character Consistency: Using dedicated Character Reference (Cref) slots, AIVI allows users to "lock" the appearance of a subject, maintaining their features, clothing, and persona across different environments.
Script-First Orchestration: The multi-step wizard breaks down complex storytelling into manageable scenes, allowing the user to act as a proper Director at every stage.



Reference-First Generation: Every project utilizes a "Style Reference" image acting as the north star, enforcing consistent lighting, textures, and artistic medium across all generated frames.
Iterative Refinement (Editing Mode): When a user wants to tweak a specific scene, the system uses the current frame as a Base Image. This allows for precise modifications while strictly preserving the original composition.
Multi-Model Orchestration: The platform intelligently routes requests between Google Gemini (for semantic enrichment) and Runware (for high-fidelity image synthesis).

Brand Style Manager: One of AIVI's core innovations is the ability to bake brand identity into the AI's "DNA." Users can define a brand kit including Primary/Secondary colors and logos.
Strict Color Enforcement: Unlike generic AI tools, AIVI forces the generation models to adhere to specific HEX codes by injecting Color Constraints into the prompt-engineering layer.
Zero-Training Personalization: Instead of time-consuming fine-tuning, AIVI uses real-time style injection, allowing any brand to achieve a "custom-model" look instantly.

An AI-powered video creation studio that turns static ideas and raw audio into professional-quality cinematic videos.
Itamar Katan, as sole creator and lead developer: product definition, market research, UX/UI design and full-stack build with React 19, Firebase and AI pipelines.
It passes the last frame of each scene as a reference for the next one, and uses character reference slots to lock a character's look across scenes.
A Brand Style Manager stores brand colors and logos, and the prompt layer forces generation to follow exact HEX codes, without training a custom model.
Google Gemini for semantic enrichment and Runware for high-fidelity image generation.