How to Use Grok Imagine Image 2.0: Edit, Resize, Reference

TL;DR

• Image 2.0 adds magic wand selection and segmentation, background removal, support for up to five reference images, and smart resize over the prior Grok Imagine model.
• Enter a prompt that specifies your subject, style, composition, and aspect ratio.
• Generate several options, then choose the image closest to your intended result.
• Select individual objects with the magic wand before changing details or replacing the background.
• Add reference images to guide later edits, then use smart resize to prepare the final image for another format.

You can run the same workflow through the OpenArt image generator.

What's new in Grok Imagine Image 2.0

xAI, Elon Musk's AI company, builds Grok Imagine as part of the Grok model family. Image 2.0 is xAI's update to the image side of Grok Imagine, and it adds precision editing to the earlier model's basic text-to-image workflow. You can now target specific parts of an image, change the background, combine several visual references, and adapt the finished image to another format.

The magic wand selects an object or region with a click. Segmentation identifies separate elements within the image, such as a person, product, or background, so you can edit one element while leaving the others intact.

Background removal isolates the main subject and clears everything behind it. You can keep the transparent result or place the subject into a newly generated scene.

Multi-reference input accepts up to five images for one generation or edit. Each reference can guide a particular detail, such as a face, product, pose, or visual style, which gives you more control than a single reference.

Smart resize adapts an image to a new aspect ratio or set of dimensions. Grok Imagine recomposes the content for the target frame instead of stretching the original pixels, which helps preserve proportions when you prepare one image for different placements.

Generate your first image in Grok Imagine

1. Open OpenArt's Create Image workspace, then select Grok Imagine Image 2.0 from the model menu.
2. Click the prompt box and describe the subject, setting, composition, and visual style you want. For example:

   A red sports car parked outside a 1970s desert motel, low-angle photograph, late afternoon.

3. Open the style settings and select a preset if you want one. Choose an aspect ratio such as 1:1, 16:9, or 9:16.
4. Set the number of images you want, then click Generate.
5. When generation finishes, click a thumbnail to open the full image. Save the result to your library or download it.

Use the magic wand and segmentation tools to select objects

1. Open your generated or uploaded image in the editor, then choose the Magic Wand selection tool.
2. Click the object you want to edit. For a larger or irregular subject, click and drag across the parts that belong in the selection.
3. Switch to the segmentation tool when the wand misses separate areas or includes nearby objects. Click the target subject to generate a more precise selection boundary.
4. Zoom in and inspect the selection overlay around edges such as hair, clothing, or product packaging. Use the add and subtract controls to include missed areas or remove unwanted ones.
5. Adjust the selection until the overlay covers only the intended object. Confirm the selection, then choose the edit you want to apply to that region.

Remove or replace a background

1. Open your generated image in the editor. For an existing file, choose Upload and select the image.
2. Choose Remove Background from the editing tools. Start the removal and wait while Grok Imagine separates the foreground subject.
3. Inspect the subject's edges. Use the selection controls to restore missing areas or remove leftover background pixels.
4. Confirm the edit to keep the background transparent. Download the result as a PNG to preserve transparency.
5. To add a new setting, choose Replace Background. Upload a background image or enter a prompt describing the scene you want.
6. Generate the replacement and review the composite. Save the edit or download the finished image.

OpenArt also offers background removal and replacement as standalone tools if you want a quicker edit without opening the full image editor: Background Remover for one-click transparent cutouts, and Background Changer for swapping in a new scene.

Guide edits with up to 5 reference images

1. Open Grok Imagine Image 2.0 and click Add reference beside the prompt field.
2. Set the reference-image count between one and five. The model will not accept more than five images in one generation or edit.
3. Upload each reference image from your device. Wait until every thumbnail appears before continuing.
4. Click each thumbnail and assign its role or tag. For example, mark one image as the subject and another as the visual style.
5. Mention the assigned tags in your prompt when needed. Write a direct instruction such as "Place [subject] in the composition and follow [style reference]."
6. Choose your aspect ratio and other output settings, then click Generate. For an edit, describe what should change and what the references should preserve.

Multi-reference input also helps with consistency across many generations, not just a single edit. If you're keeping a character's face steady across scenes, OpenArt's AI Character tool uses the same kind of reference-driven consistency. For product photography, AI Brand Kit saves an identity once and applies it automatically, so you don't need to re-upload references for every new product shot.

Resize and reformat with smart resize

1. Select Smart Resize in the image editor.
2. Choose a preset aspect ratio, such as 1:1, 16:9, or 9:16. For an exact output size, enter custom width and height values.
3. Click Resize to generate the reformatted image. Grok Imagine adjusts the composition to fit the new frame without stretching the original image.

Smart resize recomposes the image to fit a new frame. If you'd rather keep the original composition and add matching surroundings around it instead, OpenArt's AI Image Extender expands the canvas without cropping or distorting the original content.

Do the same workflow on OpenArt

OpenArt puts Grok Imagine beside models such as GPT Image 2, Seedream 5.0 Pro and Qwen Image 3.0. One subscription draws generation and editing costs from a shared credit pool, while your images remain available in a persistent library.

1. Open the OpenArt AI Image Generator, select Grok Imagine, and enter your prompt. Choose the aspect ratio and generate the image.
2. Open your preferred result in the editor. Use the magic wand or segmentation control to select the subject, object, or region you want to change.
3. Remove the existing background or describe a replacement. OpenArt applies the edit while preserving the selected subject.
4. Upload as many as five reference images when using Grok Imagine. Assign each reference to the relevant subject, style, or composition before generating again.
5. Use smart resize to change the format, or run the finished image through the AI Photo Editor to upscale it or make other touch-ups.

You can repeat the workflow with another available model, including Nano Banana 2 and Nano Banana 2 Lite, without moving files to a separate service or opening another account. Try Grok Imagine on OpenArt and create your first image.

Closing takeaway

Learning how to use Grok Imagine Image 2.0 starts with prompting, but precision editing deserves equal attention. TechRadar's hands-on test ran into this directly: asking Grok to add a rainbow to a landscape produced one that "looked far closer than it should and otherwise gave off an artificial feel," and a monocle added to a dog photo wasn't "clean enough to fool anyone." Small, localized edits took several rounds of rewording to land right. A good selection lets you correct one object without regenerating the full image, but expect to inspect edges and re-run an edit once or twice before the result matches your intent.

FAQs

Is Grok Imagine free to use?

Free access depends on the platform and its current usage limits. OpenArt provides 40 trial credits without requiring a credit card. Trial credits let you test image generation and editing before choosing a paid plan.

How many reference images can you upload to Grok Imagine Image 2.0?

Grok Imagine Image 2.0 accepts up to five reference images per generation or edit. OpenArt lets you upload and tag those references in the image workflow. Separate tags help the model distinguish a face reference from a style or composition reference.

Can you remove backgrounds in Grok Imagine?

Grok Imagine Image 2.0 can identify a background and remove it from an image. OpenArt provides the same background removal and replacement controls. You can isolate the subject and insert a different setting without exporting the image.

Can you edit Grok Imagine images on OpenArt?

OpenArt supports a generate-then-edit workflow with Grok Imagine and other image models. Its editing controls include object selection, background replacement, resizing, and upscaling. Model choice lets you continue the same project with GPT Image 2, Nano Banana Pro, or Seedream 5.0 Pro.