Guide · Image Gen · 3 min read
Four starting points for different workflows. Here’s how to choose where to begin.
AI image tools offer different ways to work: a conversation, a visual creation app, an Adobe editing workflow, or a model you run yourself. These four starting points are a researched guide to workflow fit, not a ranking from hands-on testing.
Midjourney is an option for concept art and moodboards. Its prompting guide recommends starting with short, clear descriptions of the subject, medium, lighting and mood. That makes it a useful starting point for exploring a visual direction; it does not establish that its output is universally better than another tool’s.
If you already use ChatGPT, ChatGPT Images lets you create and edit images in the same conversation. Describe what you want, then ask for changes. Availability and usage limits depend on your account. DALL·E is a legacy reference here, not the name of ChatGPT’s current image-generation experience.
Stable Diffusion models have downloadable weights and can be run locally or through hosted interfaces. They offer options for fine-tuning and technical control, depending on the model and software you choose. Local use also means handling hardware and setup. Check the licence for the specific model: downloadable weights do not mean unrestricted commercial use, and the controls available depend on your chosen interface.
Firefly connects with Adobe’s creative apps, including Photoshop and Express. Adobe says its own Firefly models are trained on licensed and public-domain content. The Firefly app also offers partner models, so check which model you select and the applicable terms. “Designed for commercial use” is not a guarantee that every output is cleared for every client, brand or campaign.
Start with one real image you need. Check whether the result follows the brief and how much editing it needs before committing to a workflow. You do not need several subscriptions just to begin.
Related