Text to Image (v3-text-to-image): Generate images from detailed text prompts, with options for TURBO/BALANCED/QUALITY rendering speed, style presets (AUTO, GENERAL, REALISTIC, DESIGN), MagicPrompt expansion, multiple aspect ratios (square, portrait, landscape), and seed/negative prompt control up to 5000 characters.
Image Editing (v3-edit): Upload an image and a mask of the same size, then use a prompt to inpaint or modify only the masked region; supports rendering speed, MagicPrompt, and seed for reproducibility.
Image Remixing (v3-remix): Provide an input image and a remix prompt, optionally setting style, size, speed, expand_prompt, number of images, strength, seed, and negative prompt to explore new variations on the same base.
Reframing (v3-reframe): Reframe existing images into different target aspect ratios and resolutions (square, square_hd, portrait_4_3, portrait_16_9, landscape_4_3, landscape_16_9), with optional style, speed, num_images, and seed.
