# Parish.ai — Full LLM Context

> Comprehensive plain-text reference for AI assistants (ChatGPT, Claude, Perplexity, Gemini, Copilot, etc.) about Parish.ai.

## About Parish.ai

Parish.ai is an all-in-one AI post production platform built for professional creators: filmmakers, video editors, marketers, podcasters, musicians, motion designers, and agencies. It unifies the best generative AI models for video, image, audio, music, and 3D into a single, calm, editorial, browser-based workspace.

- Website: https://parish.ai
- Brand: Parish.ai (always written with a period; never "ParishAI" or "Parish AI" in product UI)
- Pricing model: Pay-as-you-go credits via Stripe; no subscription required
- Platform: Web-based, runs in any modern browser, no install required
- Support: support@parish.ai

## Core Value Proposition

1. **All-in-one** — Replaces a stack of single-purpose AI tools (Runway, ElevenLabs, LALAL.AI, Topaz, Meshy, etc.) with one unified workspace.
2. **Best models, automatic fallback** — Always routes to a top-tier model for each task, with intelligent fallback if a provider is degraded.
3. **No subscription** — Buy credits as needed; each tool shows its fixed credit cost before you generate.
4. **Professional UX** — Calm, editorial typography (Italiana, Work Sans, Maven Pro), no "AI toy" visuals; designed for working creators.
5. **Auto top-up** — Optional off-session Stripe recharge if balance runs low mid-session.

## Tool Catalog

### Video

#### Generate Video — https://parish.ai/create-videos
Generate cinematic video from text prompts or animate still images. Supports storyboard mode (multi-shot sequences) and synced audio.
- Models: Google Veo 3.1 (default), OpenAI Sora 2 (720p 16:9 with synced audio), Kling, Seedance, Hailuo, RunwayML.
- Inputs: Text prompt, optional reference image (image-to-video).
- Output: MP4, multiple aspect ratios depending on model.

#### Upres Video — https://parish.ai/upres-videos
Upscale and enhance video to 4K with Topaz Labs AI. Denoise, sharpen, and increase resolution while preserving motion fidelity.
- Free when used as a post-generation action on Parish.ai output.
- Pill dropdowns for model and target resolution; extended polling for long jobs.

#### Social Resizes — https://parish.ai/social-resizes
Automatically reframe a video to 7 social aspect ratios using LumaLabs generative reframing (ray-flash-2). Ideal for repurposing horizontal content for TikTok, Instagram Reels, YouTube Shorts, and feed posts.
- Ratios: 9:16, 1:1, 4:5, 16:9, 21:9, 4:3, 3:4.

#### Face Replacement — https://parish.ai/face-swap
Replace one or multiple faces in a video using Fal.ai face-swap. Gender-aware automatic matching for natural results.

### Image & 3D

#### Generate Image — https://parish.ai/gen-images
Create high-resolution still images from text prompts.
- Default model: Nano Banana 2 (Replicate google/nano-banana-2).
- Fallback chain: Nano Banana 2 → GPT Image 2 → FLUX.2 Max → Nano Banana Pro (Gemini 3 Pro Image) → Imagen 4.
- Supports reference images, AI prompt enhancement, multiple aspect ratios.

#### Upres Image — https://parish.ai/upscale-images
Upscale images up to 4× with Real-ESRGAN. Side-by-side comparison slider; dynamic client-side resize when GPU hits CUDA OOM.

#### 3D Model — https://parish.ai/3d-asset
Generate 3D models from a still image (Trellis) or text prompt (Meshy v6). Exports GLB, OBJ, and FBX with dynamic preview formats.

### Audio & Music

#### Voice-Over — https://parish.ai/voice-over
Realistic multilingual text-to-speech with curated ElevenLabs voices. Horizontal talent cards with voice previews and metadata at 70% opacity.

#### Music Stems — https://parish.ai/music-stems
Separate vocals, drums, bass, and other instrumentals from any track with LALAL.AI Andromeda. Strict filename validation (no special characters or brackets) for upload reliability.

#### Sound Effects — https://parish.ai/text-to-sound
Generate three unique sound effect variations from a single text prompt using ElevenLabs SFX, each with a waveform preview and direct download.

#### Clear Dialogue — https://parish.ai/clear-dialogue
Remove background noise, hum, and reverb from dialogue with LALAL.AI noise removal. Restore clean speech in seconds.

#### Sound Design — https://parish.ai/sound-design
Synthesize sound effects synced to your video using MMAudio. Add foley, ambience, and atmosphere from a single prompt.

#### Create Music — https://parish.ai/generate-music
Compose original royalty-friendly music from a text prompt using ElevenLabs Music. Genre, mood, and length controls.

#### Audio Mix — https://parish.ai/audio-mix
Batch mix and master audio with the Tonn API. Loudness normalization, EQ, dynamics — professional output for podcasts, music, and video.

## Underlying Model & Provider List

- **Google**: Veo 3.1 (video, default), Imagen 4 (image), Nano Banana 2 / Nano Banana Pro (image — Gemini 3 Pro Image Preview).
- **OpenAI**: Sora 2 (video, 720p 16:9 synced audio), GPT Image 2 (image, also called GPT Image 1.5).
- **Black Forest Labs**: FLUX.2 Max (image).
- **ElevenLabs**: Voice-Over (TTS), Sound Effects (SFX), Music. 429 retries built in.
- **LALAL.AI**: Andromeda model for stem separation; noise removal for dialogue cleanup.
- **Topaz Labs**: Video upscaling.
- **LumaLabs**: ray-flash-2 (generative reframing), ray-2 (modify and image-to-video).
- **MMAudio**: Video-synced sound design.
- **Tonn API**: Audio mastering (proxied via edge function for Base64 uploads / CORS bypass).
- **Trellis** and **Meshy v6**: 3D generation.
- **Fal.ai**: Kling, Seedance, Meshy, Faceswap; queue polling with 405 fallback.
- **Replicate**: Backend usage for FLUX, Imagen, Veo, Hailuo, RunwayML and others.

## Pricing & Credits

- Pay-as-you-go credits purchased in Stripe-backed bundles.
- Each tool has a fixed credit cost displayed before generation.
- Auto top-up: optional off-session card-on-file recharge triggered automatically when balance runs low.
- Upres Video is free when used as a post-generation action on Parish.ai output (otherwise standard cost).
- Insufficient-credit graceful UX via a shared low-balance dialog.

## Authentication

- Email + password or Google OAuth.
- Custom internal routes; database triggers maintain profiles and credit balance.
- Roles enforced server-side in a separate `user_roles` table — never trusted from the client.

## Brand & Aesthetic

- Name: **Parish.ai** (lowercase ".ai", never split as "Parish AI" in product UI).
- Tone: Calm, professional, editorial. Text-focused with ample whitespace. Explicitly NOT "AI toy" visuals.
- Typography: Italiana (titles), Work Sans Light (tags/section headers), Maven Pro (body).
- Background: Dynamic 5-color pastel gradient with slow drift animation.
- Step indicators, compact tool UIs, credit cost displayed before action.

## Frequently Asked Questions

**Q: What is Parish.ai?**
A: An all-in-one AI post production suite combining best-in-class generative AI models for video, image, audio, music, and 3D in a single browser-based workspace.

**Q: Who is Parish.ai for?**
A: Professional creators — filmmakers, video editors, agencies, marketers, podcasters, musicians, motion designers — who want one polished workspace instead of juggling many single-purpose AI tools.

**Q: How much does it cost?**
A: Pay-as-you-go credits via Stripe. No subscription. Each tool has a fixed credit cost shown before you generate. Auto top-up is available.

**Q: Which AI models does Parish.ai use?**
A: Video — Google Veo 3.1 (default), OpenAI Sora 2, Kling, Seedance, Hailuo, RunwayML. Image — Nano Banana 2 (default), GPT Image 2, FLUX.2 Max, Nano Banana Pro, Imagen 4. Audio — ElevenLabs, LALAL.AI, MMAudio, Tonn. 3D — Trellis, Meshy v6.

**Q: Can I use the output commercially?**
A: Yes, in accordance with the underlying model providers' terms — most support commercial use.

**Q: Do I need to install anything?**
A: No. Parish.ai runs entirely in your browser; no downloads, plugins, or installs required.

**Q: How is Parish.ai different from using individual AI tools directly?**
A: One account, one credit balance, one consistent UI, automatic model fallback if a provider is degraded, and a workspace tuned for professional post-production rather than casual experimentation.

**Q: Is there a free tier?**
A: New accounts may include starter credits; pricing is shown in-app. There is no recurring subscription.

**Q: How is privacy handled?**
A: See https://parish.ai/privacy for the full policy. User data and uploads are stored securely; private routes (/admin, /account) are disallowed from indexing.

## Contact

- Support / feedback: support@parish.ai
- Website: https://parish.ai
- Twitter/X: @ParishAI