Next Gen Shots

An AI image and video studio you run on your own computer.

Create images from a description, edit them in plain language, generate short videos, and keep every project organized in one quiet, uncluttered workspace.

---

What it is

Next Gen Shots is a small, self-contained studio for making visual work with AI. You type what you want and it produces an image. You point at an image and describe a change, and it makes the change while keeping the lighting and feel of the original. When you need motion, it can turn a prompt into a short video. Each piece of work lives in a thread — a named project you can return to — so your output accumulates into something organized rather than a pile of loose files.

It runs entirely on your machine. Your images stay on your disk, there's no account to create, no subscription, and nothing is sent anywhere except the AI provider you choose. You bring your own API key and pay those providers directly for what you generate — often a few cents an image — with no markup in between.

What you can do

Create images from words. Describe a scene in a sentence or a paragraph and get an image back, in the aspect ratio you need for a post, a thumbnail, or a print.

Edit by describing the change. Remove a background, relight a photo, swap an element, or restyle a picture — by saying what you want, with no manual masking. The original's composition and mood are preserved.

Combine and refine. Blend elements from several images, transfer a style, or keep adjusting one image step by step until it's right.

Make short videos. Turn a prompt into a clip, with control over length, resolution, framing, and camera movement, then preview it in place. (Optional — needs a Replicate key.)

Stay organized. Group everything into threads by project, see how many pieces each one holds at a glance, and keep a complete local history of what you've made.

Start with a direction. Built-in starting points for common jobs — social media, e-commerce, content creation, UI design, marketing — tune the framing and context so you spend less time fiddling and more time making.

Who it's for

People who make visual work regularly — marketers, sellers, designers, creators — and would rather have a focused tool they control than a browser tab tied to a subscription. If you can describe what you want to see, you can use it; and if you care that your work and keys stay on your own computer, it was built with that in mind.

Scope. This is a single-user tool meant to run locally on your own machine. It has no login and is not hardened for public hosting — keep it on your computer or local network. See TECHNICAL_DEBT.md for current limitations.

---

Getting started

You'll need Python 3.13+ with uv, Node.js 18+, and a Gemini API key (get one here). For video, also add a Replicate key (here) and install FFmpeg. The Gemini key is the only thing you strictly need to begin.

# 1. Backend — install dependencies and add your key
cd backend
uv sync
cp .env.example .env        # then open .env and paste in your GEMINI_API_KEY

# 2. Frontend — install dependencies
cd ../frontend
npm install

Then, from the project root:

./start.sh

This starts everything and opens the studio at http://localhost:3000. There's no login — you land directly in the workspace. (start.sh also clears stale ports and waits for the backend to be ready before launching the interface.)

Prefer to run the two halves yourself:

# Terminal 1 — backend
cd backend && uv run uvicorn app.main:app --reload --port 8000

# Terminal 2 — frontend
cd frontend && npm run dev

• Studio: http://localhost:3000
• API: http://localhost:8000 · interactive docs at http://localhost:8000/docs

---

Under the hood

A FastAPI backend talks to the AI providers and keeps a local SQLite history; a Vue 3 + TypeScript frontend (Vite, Pinia, Tailwind, shadcn/vue) provides the studio. Images come from Google's Gemini; video and some image models run through Replicate.

API overview

Images

• POST /api/text-to-image — generate an image from a prompt
• POST /api/image-edit — edit an existing image
• POST /api/multi-image-compose — compose multiple images
• POST /api/conversational-refine — refine an image conversationally
• POST /api/prompt-enhance — expand or improve a prompt
• POST /api/upload-image · POST /api/import-uploaded-image
• GET /api/generations · DELETE /api/generations/{id}

Video

• POST /api/text-to-video — generate a video (via Replicate)
• POST /api/stitch-videos — combine clips
• GET /api/video-generations · DELETE /api/video-generations/{id}
• POST /api/upload-video · POST /api/import-uploaded-video

Threads

• GET|POST /api/threads · GET|PUT|DELETE /api/threads/{id}
• GET /api/threads/{id}/generations

Project structure

NextGenShots/
├── backend/
│   ├── app/              # FastAPI app (main, models, schemas, crud, database)
│   ├── scripts/          # one-off historical migration scripts (not part of app)
│   └── .env.example
├── frontend/
│   └── src/              # Vue 3 app (components, views, stores, composables)
├── prompt-library/       # library of hand-tuned, outcome-specific prompts
├── docs/archive/         # historical working notes
├── ARCHITECTURE.md · CONVENTIONS.md · TECHNICAL_DEBT.md
└── start.sh

---

License

Released under the MIT License — see LICENSE. You're free to use, modify, and build on it.