Key takeaways:
- Perplexity can make AI images, but it’s not front and center.
- Using Spaces with custom instructions speeds up the process.
- You get more natural-looking results with a better prompt.
So, you want to make AI images with Perplexity, but the process feels clunky? I’ve been there. While Perplexity is mostly known for research and writing, it actually lets you generate images too. The catch? The image feature is kind of hidden, and the default workflow is slow and awkward. Here’s how to get it done faster, with better results.
Table of Contents
- Perplexity’s AI Image Generation: What’s Good and What’s Annoying
- The Fastest Way: Using Spaces and Custom Instructions
- Why the Space Method Wins
- Comparing Default vs Space Method
- Tips for Getting the Most Out of Perplexity’s AI Image Tool
- FAQs
- How do I generate an AI image in Perplexity?
- Can I create more natural-looking images?
- Why is the image feature hard to find?
- What AI models does Perplexity use for images?
- Can I control the image style?
Perplexity’s AI Image Generation: What’s Good and What’s Annoying
Perplexity isn’t just for text. You can use it to generate pictures using models like DALL-E and Playground. The quality is solid—think on par with MidJourney or Stable Diffusion—but the image feature feels like an afterthought. There’s no big, obvious menu for images. If you want to make a picture, you have to start a new thread, type your prompt, wait for the AI to answer your question, and only then do you get a “generate image” button.
The prompt Perplexity uses for image generation is pulled from its answer to your question, which means it’s usually a summary and not a great one. If you want a picture of a woman practicing yoga, you’ll get a summary about yoga’s popularity, not a detailed scene. You can still generate an image, but the prompt is weak, so the result is hit or miss.
The Fastest Way: Using Spaces and Custom Instructions
Here’s the workaround I use to speed things up and get better images. Instead of starting a new thread every time, I make a Space dedicated to image prompts. This lets me set up custom instructions that turn my basic prompt into a strong, detailed image prompt.
Step-by-Step: Setting Up Your AI Image Space
- Open Perplexity and go to Spaces.
- Click “Create a Space” and name it something like “AI Image Prompt Generator.”
- Add a description (optional, but helps keep you organized).
- In the custom instructions, paste a prompt that tells the AI to rewrite your idea as a detailed, organic image description. You can tweak this prompt to get more natural, less “stock photo” results.
- Save the Space.
Now, whenever you want an image, start a new thread in this Space. Paste your idea (like “woman practicing yoga”), disable Pro Search and the web option, and make sure your Space’s custom instructions are active. Send your prompt.
The AI will quickly reply with a much better, more descriptive prompt. Click “Generate Image,” choose your style, and you’re set. This skips the research step, so it’s way faster.
Why the Space Method Wins
The main reason this method works better is that you’re not waiting for Perplexity to do a full research cycle. You get to the image generation step quicker, and the prompt is tailored for visuals, not just a summary of facts. Sometimes, you’ll get photos that look like they were shot on a phone—more casual, less polished. Other times, the results are still professional, but overall, you get more variety and control.
If you want to tweak your images even more, try adjusting the prompt in your custom instructions. For example, ask for specific lighting, mood, or camera angles. The more details you add, the more likely you’ll get something unique.
Comparing Default vs Space Method
Method | Steps Needed | Prompt Quality | Speed | Consistency |
---|---|---|---|---|
Default | New thread, wait for answer, click image | Weak (summary-based) | Slow | Inconsistent |
Space Method | New thread in Space, send prompt, click image | Strong (custom) | Fast | Better |
Tips for Getting the Most Out of Perplexity’s AI Image Tool
- Use clear, detailed prompts. The more you describe, the better the AI can visualize.
- Try different styles (photograph, illustration, etc.) for variety.
- If you want an organic look, mention things like “taken with a phone camera,” “natural lighting,” or “candid moment.”
- Don’t expect perfection every time—sometimes the AI nails it, sometimes not.
- Keep experimenting with your custom instructions to dial in your favorite style.
If you want to edit or resize your images after, check out this guide on how to resize pictures on PC. For making your own art or tweaking images, you might also like how to make a brush in Photoshop.
FAQs
How do I generate an AI image in Perplexity?
Start a new thread, type your prompt, wait for an answer, then click “generate image.” For faster results, use a Space with custom instructions.
Can I create more natural-looking images?
Yes, by customizing your prompt in the Space’s instructions to ask for organic, candid, or phone camera-style photos.
Why is the image feature hard to find?
Perplexity focuses on research and writing, so image generation isn’t a main feature. That’s why it’s kind of buried in the interface.
What AI models does Perplexity use for images?
It uses DALL-E, Playground, and FlagStudio for generating pictures.
Can I control the image style?
You can pick from a few preset styles or use a custom prompt for more control. The results depend on how you describe what you want.