CAVN AI vs PATINA

CAVN AI is a music creation platform for generating songs, cloning voices, producing music videos, and refining tracks. Song generation accepts text, image, or video input. Production tools include stem separation, mastering, MIDI export, audio enhancement, and 12-track editing with local adjustments. Its video generator supports lip-sync and videos up to 10 minutes long at up to 4K resolution. Users can clone or replace voices and create an AI singer or virtual artist. The platform integrates Smodel, Mureka, Minimax, Ace Step, and its own CAVN model. Its API documents prompt-based song generation, result queries, stem splitting, and audio-to-MIDI tools, with agent integrations aimed at MCP, OpenClaw, WorkBuddy, Hermes-style memory agents, and enterprise automation. The Free plan costs nothing and includes 10 daily renewing credits, about four songs, and two concurrent generations. Free output is for lawful personal, non-commercial use with attribution, and generated songs are public by default. Basic costs $10 a month, billed annually at $120/year; Standard costs $30 a month, billed annually at $360/year. Paid subscribers receive commercial-use rights subject to the terms.

PATINA is a paid API and web service for generating material textures from text or images. Text prompts produce complete, seamlessly tiling physically based materials, with upscaling available to 8192 pixels. The endpoint can output basecolor, normal, roughness, metalness, and height maps. An image URL can be used to make texture variations, while an input image and mask support regeneration of white mask regions while preserving black areas. Tiling may run in both directions, horizontally, or vertically, with tile size and stride controls. Requests support one to four texture images and one to eight denoising steps. Exports are listed as PNG, JPEG, and WebP, and the model listing permits commercial use. Prompt expansion is enabled by default and adds approximately $0.0025. PATINA also has a material-extraction workflow that isolates and flattens material from a photo, makes it seamless, and predicts PBR maps. API examples are provided for JavaScript, Python, and cURL, using a FAL_KEY environment variable.