The Lorvara System

Lorvara sits between the LLM workflow you already run and the models that render it.

If you are already creating cinematic scenes with an LLM, Lorvara will help keep that process organized.

The system is based on Key Prompts, organized in Projects, each with its own Scenes & Shots and an attached Brief.

Prompts

The creation layer

Everything in Lorvara revolves around Key Prompts. Key Prompts are edited in the Studio, and generated media land in the Library

Key PromptsStructured promptsStudioEdit prompts and generateLibraryThe prompt media home

Projects

The management layer

A project holds the Key Prompts for one piece of work. Briefs, that describes that work, are attached to projects. Scenes & Shots is the shot list: each shot becomes a Key Prompt in the project.

ProjectsCollections of promptsBriefsDirection for project promptsScenes & ShotsGroupings of prompts

Models & Credits

The generation system

Models render the work: image, video and audio, each with its own limits on references, sizes and durations. Credits meter what those models charge. The cost differs by model, size and duration, and it is shown before you generate along with your credit balance.

ModelsIn-app generation modelsCreditsWhat a generation costs

Files & Delivery

What gets exported

Every generation is kept as the original file at full resolution using AWS Cloudfront servers. The grids and cards you work in show smaller previews made from it, and Download hands you the original: MP4 for video, MP3 for audio, and the image file the model returned. Those files go straight into Premiere, Resolve or Final Cut, where the cut happens.

File formatsOriginals for Premiere, Resolve and Final Cut

Storage & deliveryPreviews to work in, originals to deliver

Workspaces

Node-based workflow

The Workspace is one project spread out: its prompts as cards you place and connect, instead of a list. Connect prompts to share reference libraries. Quickly test and iterate images and videos with model generation cards.

WorkspaceNode-based testing

Agents & Skills

LLM in the workflow

Three agents do the writing: the Prompt Agent, the Brief Agent and the Skill wizard. A Skill is what the Prompt Agent reads before it writes, so it decides how one kind of shot gets made. All three propose, and nothing is generated, saved or spent until you accept it.

Agents & SkillsThe LLM interfaceInstructionsWhat the agents are toldStyles & Camera ControlsFrame, lens, light, grade and the camera

Collaborate in Teams

Collaboration built in

A team lets several people work on the same projects and briefs. It has one credit balance of its own, which every generation on a team project spends from. The usage report shows what was spent, by person and by media type.

TeamsSupporting collaborationTeam creditsOne balance, usage by person

Custom Builds

Your own private Lorvara

A custom build is the same product, forked and deployed for one company. Generation goes to your own model providers and assets stay in your own storage. Gravital Digital, who build Lorvara, deliver and maintain it.

Custom buildsCustom Lorvara

Where to start

If you are new, follow the Quick Start. If you are familiar with LLM workflows, check out What a brief is to get started. A brief is what you input to Lorvara to define all aspects of a cinematic project (the look, the people, the places) once, so that every prompt you write afterwards has context for the entire project.

If you are deciding whether this is for you, read Who is Lorvara for?.

Cards above without a link, and sections listed in the sidebar without pages behind them, are not yet written. They are there so you can see the shape of what is coming.