pruna / p-image-ideogram

## Overview P-Image Ideogram is an image generation model by Pruna for text-to-image creation with reliable in-image text. It turns a single written description into a finished still image, with simple controls for resolution, aspect ratio, and output format. It’s a strong fit when you need posters, ads, or mockups where words must stay readable inside the design. ## What you can build - Poster and flyer concepts with readable headlines and taglines - Social ad variations with consistent layout per aspect ratio - Product promo images with labels, badges, and short on-image copy - Logo-like wordmarks and typographic quote cards - YouTube thumbnails and hero banners with bold text - Event announcements, menus, and simple packaging mockups ## Inputs - A required written description of the image you want. Include style, lighting, camera, and composition details. - Any exact words you want to appear inside the image. Put the literal words in quotes when spelling matters. - A required output resolution choice: 1K for standard delivery or 2K for higher detail. - A required aspect ratio selection, including square, landscape, and portrait options (such as 1:1, 16:9, 9:16, 4:3, 3:4, 3:2, 2:3). - A required “thinking” level that trades speed for prompt adherence and visual quality. - An optional prompt upsampling toggle that expands your prompt before generation. This can improve richness, but can also introduce extra styling. - An optional output file format selection: JPG, PNG, or WebP. - An optional output quality setting (0–100) for compressed formats. This doesn’t apply to PNG. - An optional numeric seed for reproducible generations. Leave it unset to randomize. ## Outputs - One generated image file in the format you selected (JPG, PNG, or WebP). - The image respects your chosen aspect ratio and resolution tier. - 1K outputs are typically suited for web UI and rapid iteration, while 2K outputs are better for text-heavy layouts and close-up detail. ## Recommended settings - Use **high thinking** for posters, logos, and any image with a lot of text. - Use **medium thinking** for everyday marketing visuals and balanced iterations. - Use **low or very low thinking** for quick drafts and exploring composition. - Turn **prompt upsampling off** when you need exact spelling and tight control over wording. - Turn **prompt upsampling on** when you want richer styling ideas from a short prompt. - Choose **PNG** when crisp edges matter most. - Choose **JPG or WebP** when smaller file size matters more than lossless detail. ## Limitations - This model is prompt-only in this listing. It doesn’t take reference images or masks. - In-image text can still misspell, swap characters, or change kerning, especially for long paragraphs or very small type. - Layout instructions compete with visual realism. Dense constraints can reduce image quality. - Highly specific brand marks and exact trademarked logos aren’t guaranteed to reproduce faithfully. - Low-quality prompts (missing subject, scene, or style details) can cause unpredictable composition. - Heavy compression settings can create artifacts around small text and thin lines. ## Safety & compliance - Don’t generate illegal content, sexual content involving minors, or content that promotes violence or hate. - Don’t generate real-person likenesses for deceptive uses. Get consent for commercial use. - Avoid using protected logos, characters, or brand assets unless you have rights. - Always review outputs before publishing. Add human moderation for user-generated prompts.