Create polished ad visuals from text or precisely transform up to three existing images.
Create a new visual from text, or edit up to three uploaded images with precise instructions.
Describe the new visual or the exact changes to apply. Prizmad accepts up to 8000 characters.
Low is the lighter draft tier. Medium is the default for final ad creatives.
Choose 1–4 variants. The complete run uses 1 token.
Sign in to spend tokens
Prompt is required
Your generated images will appear here
Grok Imagine Image 2.0 creates ad-ready images from text and edits up to three existing images. It supports crisp typography, complex layouts, realistic detail, precise editing, 1K or 2K output, and 1–4 variants per run.
Grok Imagine Image 2.0 is a create-and-edit image tool for campaign visuals. Create Image turns a prompt into a new visual. Edit Images transforms up to three uploaded images while following your instructions.
The tool supports 13 fixed aspect ratios for new images and adds Auto when editing. Choose 1K or 2K resolution, Low or Medium quality, and JPEG, PNG, or WebP output. Processing is asynchronous; an official completion-time guarantee is not published.
On Prizmad, a complete request uses 1 token whether you choose one or four outputs. The same total is shown before generation and used by the standard debit and automatic failure-refund flow.
Polished launch and product ads with designed typography and realistic materials.
Feed, Story, and vertical campaign creatives sized for their placement.
Replace backgrounds, adjust lighting, and restyle product imagery from uploaded assets.
Generate multiple compositions for campaign testing without changing the core brief.
Build poster and social-ad compositions with sharp text and structured layouts.
Create detailed product, material, lighting, and lifestyle visuals.
Transform up to three uploaded images while preserving the details named in your instruction.
Create square, landscape, portrait, panoramic, and phone-native creative sizes.
Request up to four variants in one generation for creative comparison.
Choose Create Image or Edit Images.
Write a specific prompt covering subject, composition, typography, lighting, and the intended placement.
For Edit Images, upload one to three JPG, PNG, or WebP files and describe exactly what should change or remain unchanged.
Choose aspect ratio, resolution, quality, output format, and the number of outputs.
Create the image. Completed results are saved to your Prizmad generation history.
A complete request uses 1 token, including runs that request two, three, or four output variants.
Both modes use the same visible token rule. Editing supports one to three uploaded images.
If a generation definitively fails, Prizmad's guarded async lifecycle returns the debited tokens once.
Yes. Choose Edit Images, upload one to three JPG, PNG, or WebP images, and describe the exact transformation. Auto aspect ratio follows the first input image.
Create mode supports 2:1, 20:9, 19.5:9, 16:9, 4:3, 3:2, 1:1, 2:3, 3:4, 9:16, 9:19.5, 9:20, and 1:2. Edit mode also supports Auto.
Choose 1K or 2K resolution and JPEG, PNG, or WebP output. Low and Medium quality tiers are available.
Each request can return one to four output images, and the complete request uses 1 token.
The request runs asynchronously. An official completion-time guarantee is not published, so timing can vary with the selected settings and service load.
Published 2026-08-15 · Last updated 2026-08-15