Since most of us are using at least one AI image tool at this point, lets have an actual comparison based on real design work not just “look at this cool image i made”
ive used all three extensively for design concepting, mood boards, and texture generation. heres my honest breakdown:
MIDJOURNEY: best overall image quality. the v6 photorealistic output is stunning. great for mood boards, concept visualization, editorial imagery. downsides: discord-based workflow is clunky, no direct editing/inpainting, commercial licensing exists but has limits, and the “midjourney look” is becoming recognizable
ADOBE FIREFLY: best for production workflows because it integrates with photoshop. generative fill is genuinely revolutionary for extending backgrounds and removing objects. commercial licensing is clearest since adobe trained on licensed content. downsides: standalone image generation quality is noticeably behind midjourney, and the “creative” output feels safe/corporate
CHATGPT (DALL-E): best for quick concepting and iteration because you can have a conversation about changes. “make it warmer” “add more space on the left” “now make it sunset instead of noon” - this back-and-forth workflow is uniquely powerful. downsides: image quality is the weakest of the three, inconsistent with details, and the style range feels limited
for my actual design work: midjourney for mood boards and hero concepts, firefly (via photoshop) for production editing, dall-e for quick sketching conversations with clients
which AI image tool are you using for design work and for what specifically?
For motion graphics work specifically: midjourney for style frames and mood boards, firefly for background extensions and compositing in after effects, dall-e for quick storyboard concepts when im on calls with clients
the conversational aspect of dall-e is underrated for client presentations honestly. being able to say ‘lets try it more minimal’ in real-time during a meeting is a superpower
One tool youre missing: stable diffusion (local install). yeah the setup is technical but if you have a decent GPU the control you get is unmatched. controlnet for pose/composition matching, inpainting thats better than firefly, and you own everything locally. no subscription, no usage limits
for web design mockups specifically i use comfyui with a custom workflow that generates hero images, textures, and icon concepts. saves me hours
For packaging design: firefly wins because of the commercial licensing clarity. when youre designing for products that go on store shelves the last thing you want is ambiguous IP rights. adobe’s ‘commercially safe’ training data claim gives my clients legal team something to work with
midjourney outputs better images but try explaining their licensing terms to a fortune 500 legal department. its not fun
Unpopular opinion: none of these replace actual illustration for my work. i do childrens books and the AI tools cant maintain character consistency across pages, cant follow specific art direction reliably, and produce a homogeneous style that defeats the purpose of illustrated books
i use midjourney exclusively for initial concept exploration with clients. ‘something like this feel?’ then i draw everything from scratch. the AI is a communication tool not a production tool