I have relied on Perplexity Pro ($20 or $10 with this referral link) for awhile now, but had not given much thought or exploration time to the image generation features. That is, until today when Dr. Dru Stevenson asked me on Mastodon, “will it generate images?”

The answer, as you can see above, is “Yes.” Not only that, but you can use several different AI models to get the job done:

Generating images is pretty easy, although you can only generate 50 a day with a Pro account. I can’t believe I completely missed this and haven’t been using it.

Generate Image Tools

Here’s the Generate Image tools:

A Quick Experiment

For fun, I uploaded a photo of myself to see if it could use DALL-E to adapt the photo into a watercolor painting. It was not able, giving this rationale:

Unfortunately, as an AI language model, I do not have the capability to generate a painting from the provided portrait photo. My primary function is to process and respond to text-based inputs. I do not have the ability to create, generate, edit, manipulate or produce visual content such as paintings, images, or artwork. If you are interested in having a painting created from this portrait photo, I would suggest exploring options with human artists or specialized image generation tools designed for that purpose.

What a bummer. When I gave ChatGPT Plus the same prompt (shown below), it generated the image:

Prompt: Generate a watercolor paint from the attached portrait photo of me.

ChatGPT Plus Image Generated

That is about what I expected from ChatGPT Plus with DALL-E. I had hoped for something similar from Perplexity Pro, but not yet.

Perplexity Pro Watercolor Image

Instead, Perplexity Pro allowed me to generate a watercolor painting:

What a disappointment with the current capabilities, but a big step forward regardless in image generation. No doubt, this will be the worst it can do….

Perplexity Pro On Image Generation

I asked Perplexity Pro a simple question (how can perplexity generate an image?) about whether it could do the job, and it gave me this information:

Perplexity can generate an image through its AI technology, which is available for Perplexity Pro users on the web browser. Here’s how the process works:

  1. Accessing the Feature: Once you have a Perplexity Pro subscription, you can generate images by waiting for the search response to finish generating. After that, on the right column next to the response, you will see the option to “Generate Image”[1].

  2. Choosing a Style: Click on “Generate Image” to begin the process. You will be presented with options for some pre-written styles. If you want to customize the image further, you can click on the icon next to “Choose a style” to apply your own prompt and give additional instructions[1].

  3. Generating the Image: After selecting the style and providing any additional instructions, you will have to wait for the image to be generated. Once it’s done, the image will be displayed at the top of the column[1].

  4. Model Selection: You can select from three different models to generate your image: Playground V2.5, DALL·E 3, or Stable Diffusion XL. The model can be switched from the settings page under the Pro section[1].

  5. Usage Limit: As a Perplexity Pro user, you can create up to 50 images per day[1].

  6. Commercial Use: Images generated through Perplexity’s service are intended for personal, non-commercial use only, as per the Terms of Service[1].

  7. Error Handling: If you encounter an error while trying to generate an image, it could be due to explicit or moderated content in your query, an unstable connection, or a technical issue on Perplexity’s end[1].

This feature is currently only available on the web browser and not on the mobile app[1].

Citations: