Image Generation

Generate stunning images from text prompts with AI

Upload product photos to guide the generation

Choose the image model. GPT Image 2.0, Grok Imagine, and HiDream-O1-Image support reference images. Z Image Base supports text-to-image only.

1

Sign in to spend tokens

Prompt is required

Your generated images will appear here

Overview

AI Image Generator — Nano Banana, GPT Image 2, Seedream & Z Image

Image Generation is Prizmad's multi-model AI image tool: one prompt field, seven models to choose from — including Nano Banana 2, GPT Image 2.0, and Grok Imagine — from 3 tokens per image (model- and size-dependent). Write a prompt, pick a model and aspect ratio, generate up to 4 images at once.

Background

What is Image Generation on Prizmad?

Image Generation is a single workspace that puts seven image models behind one prompt field: Nano Banana 2, GPT Image 2.0, Nano Banana Pro, Z Image Base, GLM Image, Grok Imagine, and HiDream-O1-Image. Instead of maintaining accounts with multiple providers, you pick the model from a dropdown and generate — same prompt box, same asset library, same token balance.

The tool supports reference images where the model allows it: GPT Image 2.0, Grok Imagine, and HiDream-O1-Image accept up to 4 uploaded photos to guide the generation, which is how you keep your actual product in the frame instead of a lookalike. Z Image Base is text-to-image only. Aspect ratio options cover the formats that matter for ads — 9:16, 16:9, 1:1, and 4:5 — and GPT Image 2.0 adds its own canvas-size presets, a quality setting, and a choice of output format. A slider lets you generate 1 to 4 images per run.

On Prizmad, Image Generation uses the same token balance as AI avatars, video, voiceover, captions, and music. The images you generate flow straight into your video projects and ad creatives, and everything is yours to use commercially — no per-model billing, no API keys.

Use cases

What you can create with Image Generation

Product photography without a studio

Studio-style hero shots, lifestyle scenes, and dramatic lighting setups for products you've never physically photographed.

Static ad creatives

Feed and Stories visuals in the exact aspect ratio your ad platform wants — generate four variants and split-test them.

Product-page and listing imagery

Clean e-commerce visuals guided by reference photos of your actual product, for Shopify pages and marketplace listings.

Social content at volume

Daily post visuals, story backgrounds, and campaign imagery generated in batches instead of sourced from stock sites.

Capabilities

Key capabilities

Seven models, one prompt field

Switch between Nano Banana 2, GPT Image 2.0, Nano Banana Pro, Z Image Base, GLM Image, Grok Imagine, and HiDream-O1-Image without leaving the page.

Reference-image guidance

Upload up to 4 product photos to guide the generation on GPT Image 2.0, Grok Imagine, and HiDream-O1-Image — keep your real product in the shot.

Ad-ready aspect ratios

9:16 for Stories and Reels, 16:9 for YouTube, 1:1 for feed, 4:5 for Meta — set per generation, no cropping afterward.

GPT Image 2.0 fine controls

When GPT Image 2.0 is selected, you also get canvas-size presets, a quality setting (higher fixed quality can use more tokens), and output format choice.

Batch generation

Generate 1 to 4 images per run to compare variations of the same prompt and pick the strongest creative.

From 3 tokens per image

Generations start at 3 tokens, making it cheap to iterate on prompts before committing to a full campaign's worth of creatives.

Workflow

How to use Image Generation on Prizmad

1

Write a specific prompt: subject, setting, lighting, and style. For product shots, name the surface, background, and mood — "organic skincare bottle on marble, fresh green leaves, soft natural light" beats "nice product photo".

2

Optionally upload up to 4 reference images of your product. Use GPT Image 2.0, Grok Imagine, or HiDream-O1-Image if you want the references to guide the result.

3

Pick your model. Z Image Base is text-to-image only; the others each bring their own look — generate the same prompt across two or three models to compare.

4

Set the aspect ratio — 9:16, 16:9, 1:1, or 4:5. On GPT Image 2.0, choose a canvas-size preset, quality level, and output format instead.

5

Set the number of images (1–4) and click Generate. Results arrive in roughly 1–5 minutes and are saved to your asset library.

Pricing

How Image Generation fits Prizmad pricing

From 3 tokens per image

Generations start at 3 tokens (Z Image Base) and vary by model — Nano Banana 2 is 16 tokens. On GPT Image 2.0, higher fixed quality settings can use more tokens — the Auto setting lets the provider choose. Tokens come from your token balance.

Seven models, one token balance

All seven image models are unlocked on the same token balance you use for AI avatars, video, voiceover, and music — no separate provider accounts, no API keys, no per-model billing.

Run out? Top up, don't upgrade

If your tokens run out mid-campaign, buy a one-off top-up directly from the top-up page. Every image you generate stays yours with full commercial rights.

FAQ

Frequently asked questions

Is Image Generation free to use?

It runs on your Prizmad token balance. There is no separate charge per model — generations start at 3 tokens per image.

How much does an image cost to generate?

From 3 tokens per image. On GPT Image 2.0, choosing a higher fixed quality can use more tokens, while Auto lets the provider decide. You see the cost in the tool before generating.

Which models are available?

Seven: Nano Banana 2, GPT Image 2.0, Nano Banana Pro, Z Image Base, GLM Image, Grok Imagine, and HiDream-O1-Image. You switch between them with a dropdown — same prompt field, same library, same tokens.

Can I upload my own product photos as references?

Yes — up to 4 reference images (PNG, JPG, or WebP). GPT Image 2.0, Grok Imagine, and HiDream-O1-Image support reference images; Z Image Base is text-to-image only.

Can I use the images commercially in ads?

Yes. You keep full commercial rights to every image generated on Prizmad — use them in paid ads, on product pages, in email, and in print.

Which aspect ratios are supported?

9:16, 16:9, 1:1, and 4:5 on most models. GPT Image 2.0 uses its own canvas-size presets instead, with an Auto option that infers the size from your prompt or reference images.

How many images can I generate at once?

1 to 4 per run, set with a slider. Generating multiple variants of the same prompt is the fastest way to find a creative worth testing in ads.

How does this compare to using Midjourney or a single-model tool?

The practical difference is choice under one roof: instead of committing to one provider's look and account, you run the same prompt through several models — including reference-image-capable ones — and keep whichever result wins, all billed from one token balance that also covers Prizmad's avatar, video, and audio tools.

Related tools

More tools like this

Luma UNI-1Generate polished images and transform existing ad creatives with reference-guided editing.KREA 2Generate high-quality, art-directed images with strong style, mood, and creative control.Krea 2 Turbo Style ReferenceStyle-guided image generationMAI-Image-2.5Generate photorealistic images or refine existing visuals with precise prompt-based edits.Qwen Image 3.0 ProGenerate premium campaign images or precisely edit up to three references.Seedream 5.0 ProGenerate and precisely edit ad visuals, product mockups, posters, and layouts with strong text rendering.

Published 2026-07-16 · Last updated 2026-07-16