reve / create v2

## Overview Create V2 by Reve is a text-to-image model that turns one written prompt into a finished image. It’s built to follow instructions closely, including layout-like requests and readable on-image text. You describe the subject, style, and any exact wording, then the model renders a single image at a chosen aspect ratio. This is useful when you need quick, polished visuals without manual design edits. ## What you can build - Poster and banner concepts with clear, legible headlines - Social creatives sized for square, landscape, or vertical placements - Product mockups and marketing scenes from a single description - Storybook-style pages and multi-panel layout concepts (generated as one image) - Visual drafts for art direction, mood boards, and concept exploration ## Inputs - A detailed image description written in plain English. Include subject, setting, lighting, and style. - Any exact text you want inside the image. Put the text in quotes and describe its placement. - An aspect ratio choice for the final canvas. Choose auto, or a fixed ratio like 16:9, 1:1, or 9:16. ## Outputs The model returns 1 generated image. The result is typically delivered as a downloadable image file and/or a hosted image URL, depending on the platform. Reve’s standard output size stays near 1.5 megapixels, with these common resolutions: - 16:9: 1536 × 864 px - 3:2: 1536 × 1024 px - 4:3: 1536 × 1152 px - 1:1: 1024 × 1024 px - 3:4: 1152 × 1536 px - 2:3: 1024 × 1536 px - 9:16: 864 × 1536 px ## Limitations - Output resolution is fixed per aspect ratio at about 1.5 MP. Large print work may need upscaling. - Long or conflicting prompts can reduce instruction accuracy. - Small, dense typography can still fail, even when the model is text-strong. - The model may simplify complex layouts into a cleaner design than you expect. - Low-quality inputs can hurt results. Short prompts, vague style notes, or unclear text placement often produce weaker images. ## Safety & compliance - Safety filters may block or replace outputs that violate content rules, including explicit sexual content. - Only generate content you have rights to create and use, including logos, brand marks, and likenesses. - Don’t use the model to produce harmful, illegal, or privacy-invasive content.