Overview
Create a PNG image from a written description. Name the subject, setting and composition, then choose a style and size. Review the result against your prompt before using or downloading it. Some controls may depend on the active workspace. This tool runs online and requires sign-in and an internet connection.
The AI generates an image from the text description and configured form inputs you provide. The configured output format is PNG.
When to use
- Create a new image from a written description when you don’t want to start with a photo. Choose a square, landscape or portrait Size to suit the space you need to fill.
- Explore a visual concept or style. If the image needs an exact date, price, label or brand spelling, inspect every character and add the final wording separately in a design editor.
How to use
Open AI Image Generator and sign in to its online workspace.
Enter a description in Prompt, up to 1,000 characters. Name the subject, setting and composition. For example, “A blue ceramic mug on a pale wooden table, soft window light, empty space on the right.” Choose a Style and Size. Styles are Photorealistic, Digital Art, Anime, Oil Painting, Watercolor and 3D Render. Sizes are Small (512×512), Medium (768×768), Large (1024×1024), Landscape (1024×768) and Portrait (768×1024). Quality currently offers Better. Steps ranges from 15 to 40, with a default of 25; Prompt Strength ranges from 1 to 15, with a default of 7.5; Seed ranges from 0 to 2147483647. The Seed label says to leave 0 for a new image each time. If What to Avoid is available, enter unwanted details there, up to 500 characters. Control availability may depend on the active workspace, so review the controls and restrictions shown there.
Click Generate Image and compare the result with your description. Check the number of subjects, object shapes and any lettering. If an important detail is missing, revise that part of Prompt and generate again. When you have an image to keep, click Download if the workspace offers that option, then open the saved PNG and check it before use.
Settings reference
| Setting | Options | Default |
|---|---|---|
| Prompt | Text | Empty |
| Style | photorealistic: Photorealistic digital-art: Digital Art anime: Anime oil-painting: Oil Painting watercolor: Watercolor 3d-render: 3D Render | photorealistic |
| Size | 512x512: Small (512×512) 768x768: Medium (768×768) 1024x1024: Large (1024×1024) 1024x768: Landscape (1024×768) 768x1024: Portrait (768×1024) | 1024x1024 |
| Quality | better: Better | better |
| Steps | 15 – 40 | 25 |
| Prompt Strength | 1 – 15 | 7.5 |
| What to Avoid | Text | Empty |
| Seed. Leave 0 for a new image each time | 0 – 2147483647 | 0 |
Formats & limits
| Detail | Requirement |
|---|---|
| Input formats | Text and configured form inputs |
| Output formats | PNG |
Access & cost
| Detail | Requirement |
|---|---|
| On-device | Not available for this tool |
| Online processing | Sign-in required; Internet connection required |
Examples & use cases
A social media creator
Make a wide social post concept
You need a wide illustrated reading nook with a chair, lamp and plant, leaving open space on one side for your layout. The image needs a scene, not exact wording.
Text and configured form inputs to PNG
- Prompt:
- A cozy reading nook with one armchair, a floor lamp and a leafy plant, viewed straight on, warm evening light, open space on the right
- Style:
- digital-art
- Size:
- 1024x768
- Quality:
- better
- Choose Digital Art for the illustrated direction and Landscape (1024×768) for a wide composition.
- Use the available Better quality setting and review whether the generated scene follows your prompt.
- Check that the chair, lamp and plant appear and that the right side has room for your layout.
Expected outcome
A landscape PNG concept to review for the requested objects and open space before placing it in your layout.
A teacher preparing a story activity
Create a portrait story illustration
You want a tall anime-style illustration of a child explorer beside one friendly robot. The activity calls for two characters, without extra figures.
Text and configured form inputs to PNG
- Prompt:
- A child explorer standing beside one friendly small robot in a forest clearing, full figures visible, bright adventurous anime illustration
- Style:
- anime
- Size:
- 768x1024
- What to Avoid:
- extra people, extra robots
- Choose Anime for the requested visual direction and Portrait (768×1024) for a taller frame.
- If What to Avoid is available, enter “extra people, extra robots” to identify details you want left out.
- Check whether the child and one robot are visible and whether the unwanted extra figures were excluded.
Expected outcome
A portrait PNG draft to review for the two-character composition and any unwanted extra figures.
A small-business owner
Draft a square product concept
You need a square concept of a plain ceramic soap dish on a bathroom shelf, without labels or extra products. It is for an early design discussion, not a finished branded graphic.
Text and configured form inputs to PNG
- Prompt:
- A single plain white ceramic soap dish on a light bathroom shelf, simple composition, soft natural daylight, no label
- Style:
- photorealistic
- Size:
- 1024x1024
- What to Avoid:
- bottles, packaging, labels
- Choose Photorealistic for the requested photographic direction and Large (1024×1024) for a square composition.
- If What to Avoid is available, name the bottles, packaging and labels you want left out.
- Check that the result shows one dish and inspect it for extra objects or visible lettering before sharing the concept.
Expected outcome
A square PNG concept to review with your team for the requested subject and unwanted details.
More ways to use it
- A designer planning a website banner: You need a wide visual starting point for a banner and want to explore a described scene in a particular style before adding layout and precise wording in a design editor. A landscape PNG concept to assess for the scene and composition. Add exact banner wording separately in your design editor. Prompt: A quiet coastal path with dune grass, distant sea and open sky, wide composition, Style: watercolor, Size: 1024x768.
- An illustrator exploring a character idea: You want a square visual draft of one clockwork bird for a project mood board, without extra birds in the scene. A square PNG draft to review for the single-bird composition and requested visual direction. Prompt: One small clockwork bird perched on a branch, clear silhouette, centered composition, Style: 3d-render, Size: 1024x1024, What to Avoid: other birds.
Troubleshooting
The image is missing a requested subject or composition detail.
Try this: Revise the relevant details in Prompt and generate again.
What to expect: The revised description gives the generator clearer details to follow. Check the new result; this doesn’t guarantee every requested detail will appear.
Unwanted objects or details appear in the image.
Try this: If What to Avoid is available, enter those details there and review the next result.
What to expect: The field communicates which details you want excluded. Inspect the result to see whether they were left out.
Privacy
Online processing uploads files. Completed results may remain available for later download. See https://mioffice.ai/privacy for current file-handling and retention details. Check the processing mode before starting.
Frequently asked questions
Do I need to upload a photo?
No. This generator creates images from a description entered in Prompt and accepts text and configured form inputs.
Can I describe things to leave out?
Yes. If the active workspace offers What to Avoid, enter unwanted details there; the field accepts up to 500 characters. Inspect the result to see whether those details were excluded.
Why does text inside an AI image have spelling errors?
Generated lettering can look convincing while containing incorrect words or distorted characters. Check every visible word. For an exact date, price, heading or brand name, add the final text separately in a design editor.
How do I write a prompt for the image I want?
Name the subject, setting and composition, then check the result against those details. For example: “A blue ceramic mug on a pale wooden table, soft window light, empty space on the right.” Prompt accepts up to 1,000 characters.
What styles can I choose?
Style options are Photorealistic, Digital Art, Anime, Oil Painting, Watercolor and 3D Render. Choose one that fits your intended visual direction, then inspect the generated image.
What image sizes are available?
Choose Small (512×512), Medium (768×768), Large (1024×1024), Landscape (1024×768) or Portrait (768×1024). Pick a square, wide or tall frame to suit the space you’re preparing the image for.
What resolution can I export?
The listed Size options go up to Large (1024×1024), Landscape (1024×768) and Portrait (768×1024). The configured output is PNG; check the downloaded file before using it.
Can I use a reference photo or sketch to guide generation?
The configured inputs are text and form inputs, and generation starts from the description in Prompt. A reference-image input isn’t listed, so describe the subject and composition in your prompt instead.
Can the generator create an SVG?
The registered output is PNG, not SVG. Plan for a PNG file when using this generator.
Can I use the generator without signing in?
This tool runs online and requires sign-in and an internet connection. Review the access and processing charge shown in the workspace before starting.
What can I make with this text-to-image generator?
It generates images from text descriptions. Describe the subject and scene, choose an available style and size, and review the resulting PNG.
What does the Quality setting do?
Quality currently offers Better. Start with the default Steps and Prompt Strength, then compare the result with your prompt. Processing time varies with demand and the request.
How are uploaded files and generated results handled?
Online processing uploads files. Completed results may remain available for later download. See https://mioffice.ai/privacy for current file-handling and retention details. Check the processing mode before starting.