AI Image Generator for Study

Study visuals: labeled diagrams (heart, cell, atom), concept illustrations (how photosynthesis works), and flashcard images. You describe what you want in a text prompt — or upload a reference image in the Image Studio — and Scholarly generates 2–3 candidate images to choose from.

It depends where you generate. In a page or flashcard, the image you pick is downloaded, converted to JPEG, and attached to that component in place. In the Image Studio, images are saved to a persistent grid with their prompt, model, and aspect ratio, and you can download them. In chat, the image is embedded directly in the message.

Yes — in the Image Studio. You can upload 1–3 reference images (or pick from your content library) and Scholarly uses them to edit or remix a new image. In-page and flashcard generation are text-prompt-only and don't support reference images.

When generating inside a page or flashcard, you pick from three styles: Cartoon, Realistic, or Sketch. The Image Studio goes further — you can select one or more models, choose presets (like labeled-diagram or anatomical), and set an aspect ratio of 1:1, 16:9, 9:16, 4:3, or 3:4.

Generated images aren't editable in place — to change one, regenerate with a new prompt, or use it as a reference image in the Image Studio to remix it. There's no timeline-style editor.

Yes. Image Studio and chat share a separate beta allowance: 2 lifetime images on Free, 40 per week on Premium, and 80 per week on Laureate. Free images do not reset. Paid allowances reset weekly with no rollover, and limits may change during beta. Images inside other agent-created artifacts remain included in those features.

Each prompt returns a grid of 2–3 candidate images so you can pick the clearest one. There's no batch mode — each prompt generates separately rather than queuing several at once. In the Image Studio, selecting multiple models runs a concurrent generation for each.

Quality and style adherence depend on the underlying model, so very complex or highly specific requests may not match perfectly on the first try. If it misses, regenerate with a clearer prompt, switch styles or models in the Image Studio, or add a reference image to steer the result.

Yes. You can generate an image during a chat conversation and it's embedded right in the message with its metadata, so you can keep revising your study material without leaving the thread.