How to Generate AI Images on magicdoor.ai

magicdoor.ai gives you access to 9 image models from multiple providers in one interface. You do not need separate provider accounts or separate image subscriptions; each generation is charged per image through your magicdoor.ai balance.

This guide walks you through the entire process: picking the right model, writing effective prompts, editing and upscaling images, and keeping costs low.

Quick Start: Generate Your First Image

1. Open a chat on magicdoor.ai
2. Select an image model from the model dropdown (or stay on a chat model — see below)
3. Describe the image you want in plain language
4. magicdoor.ai generates the image and displays it in the chat
5. Download, regenerate, or refine from there

You can generate images two ways:

• Image models directly — Select an image model like Flux 2 Pro or Seedream 4.5 from the dropdown and type your prompt
• From a chat model — While chatting with Claude, GPT, or another text model, use the image generation tool. The chat model can help refine your prompt before generating.

Choosing the Right Image Model

The model you pick determines quality, style, speed, and cost. Here is how they compare:

Rule of thumb: Start with Seedream 4.5 ($0.03). If the result does not meet your needs, move up to Flux 2 Pro ($0.05) or ChatGPT Image 2 ($0.15). For a deeper comparison, see our image model comparison, advanced image workflows, and model cost guide.

Writing Effective Image Prompts

The quality of your prompt determines the quality of your image. Here is how to write prompts that work:

The anatomy of a good prompt

A strong prompt includes four elements:

1. Subject — What is in the image (a golden retriever, a mountain landscape, a logo)
2. Style — How it should look (photorealistic, watercolor, minimalist, cinematic)
3. Composition — How the scene is arranged (close-up, wide shot, bird's eye view, centered)
4. Mood/lighting — The atmosphere (warm golden hour, dramatic shadows, soft diffused light)

Examples: weak vs. strong prompts

Let magicdoor.ai enhance your prompts

You do not need to master prompt engineering. magicdoor.ai's automatic prompt enhancement takes your basic description and adds the technical details that image models respond to: lighting terms, composition guidance, and style references. Write what you want in plain language, and the system handles the rest.

Step-by-Step Workflows

Workflow 1: Create a Social Media Image

1. Pick your model — Seedream 4.5 for general content, ChatGPT Image 2 if you need text overlays
2. Write your prompt — Describe the scene and specify the aspect ratio (e.g., "square format for Instagram" or "16:9 for Twitter header")
3. Generate — Review the result. If close but not right, adjust your prompt and regenerate
4. Upscale if needed — Use Recraft Upscaler ($0.006) to increase resolution for print or high-DPI displays
5. Download — Click the download button on the generated image

Cost: $0.03–0.15 per first image, plus any regenerations.

Workflow 2: Design a Logo or Brand Asset

1. Start with Recraft V4 ($0.04) — It is tuned for clean, design-oriented output
2. Be specific about style — Include terms like "minimalist," "vector," "flat design," or "geometric" in your prompt
3. Specify colors — Name exact colors or palettes: "navy blue and gold," "monochrome," "pastel earth tones"
4. Iterate — Logos usually take 3–5 generations to get right. At $0.04 each, that is $0.12–0.20 total
5. Upscale — Run the best result through Recraft Upscaler for a high-resolution final version

Cost: $0.15–0.25 for a polished logo concept.

Workflow 3: Generate Photorealistic Images

1. Use Flux 2 Pro ($0.05) — Most consistently photorealistic output
2. Use photography language — Terms like "35mm lens," "shallow depth of field," "studio lighting," "golden hour" guide the model toward realism
3. Avoid contradictions — Do not mix photorealistic terms with illustration styles
4. Refine — If a specific area needs fixing, switch to Flux.1 Kontext Pro to edit just that part of the image without regenerating from scratch

Cost: $0.05–0.15 per final image.

Workflow 4: Edit an Existing Image

1. Upload your image to the magicdoor.ai chat
2. Select Flux.1 Kontext Pro ($0.04) — Purpose-built for image editing
3. Describe the change — "Change the background to a sunset beach," "Remove the person on the left," "Make the shirt blue instead of red"
4. Review and iterate — Targeted edits are more predictable than full generation, so you usually need fewer attempts

Cost: $0.04 per edit.

Workflow 5: Batch Content Creation

For creating multiple related images (e.g., a series of blog illustrations):

1. Start with your cheapest option — Seedream 4.5 at $0.03/image
2. Establish a style template — Write a base prompt with consistent style terms, then swap out only the subject for each image
3. Generate in bulk — At $0.03 per image, 20 images costs $0.60
4. Upgrade selectively — If a few images need higher quality, regenerate only those with Flux 2 Pro or ChatGPT Image 2

Cost: $0.60–2.50 for 20 images, depending on how many you upgrade.

Cost-Saving Tips

Try the cheapest model first. Seedream 4.5 at $0.03 per image is genuinely good. Many users find it sufficient for 70–80% of their needs. Only move to more expensive models when the result visibly falls short.

Use the Recraft Upscaler. Generating a cheap image and upscaling it ($0.03 + $0.006 = $0.036) is cheaper than jumping straight to a premium model when resolution is the main issue.

Refine your prompt before regenerating. Each generation costs money. Spend a few seconds improving your prompt rather than generating five versions and hoping one works.

Use prompt enhancement. magicdoor.ai's automatic enhancement often produces better results on the first try, reducing the need for regenerations.

Match the model to the task. Do not use ChatGPT Image 2 ($0.15) for a quick social media graphic when Seedream ($0.03) would work. Reserve premium models for work where the quality difference genuinely matters.

For more cost strategies across all of magicdoor.ai, see our guide to saving money on AI.

Image Generation vs. Chat Models: When to Use Which

You might wonder whether to generate images directly or ask a chat model to help:

Use an image model directly when you already know what you want. Write the prompt, generate, done.

Use a chat model first when you need help figuring out what to generate. Describe your goal to Claude or GPT-5.4 Mini, and ask it to write an optimized image prompt for you. Then switch to an image model to generate it. This is especially useful for complex scenes or when you are not sure how to describe what you envision.

FAQs

Which image model should I start with?

Seedream 4.5. At $0.03 per image it is the best value, and the quality is strong enough for most use cases. Move to Flux 2 Pro ($0.05) for photorealism or ChatGPT Image 2 ($0.15) for complex scenes with text. See our complete image model comparison for detailed guidance.

How much does AI image generation cost on magicdoor.ai?

Between $0.006 and $0.15 per image depending on the model. Each generation is priced per image and shown against your magicdoor.ai balance. Check the model cost page for current pricing.

Can I edit images after generating them?

Yes. Upload the generated image back into the chat and use Flux.1 Kontext Pro ($0.04) to make targeted edits — change colors, remove objects, swap backgrounds. You can also upscale any image with Recraft Upscaler ($0.006).

Do I need to learn prompt engineering?

Not really. magicdoor.ai's automatic prompt enhancement adds technical details to your plain-language descriptions. That said, a few basics help: be specific about what you want, mention a style, and describe the lighting or mood. The prompts section above covers everything you need.

Can I use generated images commercially?

Commercial usage depends on the provider's current terms. Check the model provider's latest terms for your specific use case before relying on the output commercially.

What should I read next?

For model-by-model tradeoffs, read the image model comparison. For more advanced edits, references, and upscaling patterns, read advanced image workflows.

---

Ready to start generating? Try magicdoor.ai — access 9 image models starting at $0.03 per image, with a $6/month base subscription and pay-per-image usage.