Skip to content
BENIBIX Learn

Midjourney and AI Image Tools

Level 5 · Pictures, video and voice

How Midjourney, Adobe Firefly and ComfyUI turn words into a picture. What sets each one apart, and what you may do with what it makes.

Course contentsLesson 9 of 10

After this lesson you can

  • Say how a text-to-image tool makes a picture
  • Tell Midjourney, Firefly and ComfyUI apart
  • Check the rights before you use a picture

11 mincovers 6 glossary entries

Worth reading first: ChatGPT, Claude and Gemini

How Midjourney, Firefly and ComfyUI make pictures

AI image tools turn a line of text into a picture. Midjourney, the best known, opened to the public in July 2022, and each prompt gives you four pictures to pick from. Adobe Firefly is the one built for paid work. ComfyUI is free and runs on your own computer, where you wire the steps together yourself.

All three do text-to-image. Most of the models behind them are diffusion models. They start from random noise and clean it up, step by step, until it matches the words. That is why the same prompt can give a slightly different picture every time.

Midjourney is made by a small lab of the same name. At first you could only use it by typing in a group chat app. Now it has its own website too. You type a prompt, get four images in a grid, and pick one to make bigger or change. It is known for a rich, painted look.

Adobe Firefly makes and edits pictures, video, sound and vector art. It lives in Photoshop and its maker's other apps. The maker says it trained Firefly on stock pictures it has the rights to, and on old work that is free for all. It says it did not train on its customers' files. So a firm's lawyers find it easier to say yes.

ComfyUI is a free, open source app that runs image and video models on your own computer. There is no single prompt box. You build a graph of boxes called nodes: one loads a model, one holds the prompt, one makes the image. Wires join them, and you can save the graph and run it again.

ComfyUI is not a model. It is the app that runs models, such as Stable Diffusion, and you download those yourself. It needs a strong graphics card and an afternoon to learn. In return, every step is yours to change, and the same graph gives the same result next time.

The common mistake is to think any picture is free to use. Check each tool's terms, and take care with real people, brand logos and the style of a living artist. A model trained on licensed work lowers the risk from its training, not the risk from what you ask it to draw.

Which one to pick

  1. 1

    Pick Midjourney

    When the look matters most. A book cover, a mood board, or a first idea for a poster.

  2. 2

    Pick Adobe Firefly

    When the picture goes into paid work, and you edit in Photoshop.

  3. 3

    Pick ComfyUI

    When you need full control. Or when you want to run the same steps on hundreds of images at home.

One short prompt, four pictures, then you pick one. Midjourney works this way every time.

Worked example

Jack makes a poster for a school play

Jack needs a poster for his school's play. He tries 3 prompts in Midjourney and gets 12 pictures in about two minutes. He picks a stormy castle.

Then the town paper wants the art for a paid ad. Its rules ask for art made with licensed tools. So Jack makes a new one in Firefly, and keeps the first one for the school hall.

The words for this lesson6

Check yourself

What is ComfyUI?

Finished this lesson?

Marking it complete also marks its 6 glossary entries learned, which updates your roadmap and your saved page. 0 are already marked.