Looking back at DALL·E now, its positioning is actually pretty clear—strong at understanding semantics, super low barrier to entry, perfect for that “idea to sketch” stage. You’ve got some vague concept in your head, just describe it in plain language, and it’ll give you a decent direction to work with.
That’s incredibly friendly for people who don’t know their way around parameters. But if you need fine control, consistent characters, or the ability to tweak things over and over, it’s not the best tool for that—that’s ComfyUI’s turf.
I use it as an inspiration gateway, get a direction, then move to other tools for the polish. Each tool does its own job, and it flows better that way.
“From idea to sketch” hits the nail on the head. Plain language gets you in the right direction—super friendly for people who don’t know their way around parameters.
Character consistency is really bad, the same person changes face in the next image.
Same here, I use Inspiration Portal that way too—once I’ve got the direction, I switch to MJ for fine-tuning.
For real, the semantic understanding is just on another level—it can handle even the most abstract prompts. With SD, you’d be translating that shit for ages.
“Each tool doing its own job” — that’s the right mindset. Trying to make one thing do everything is just asking for trouble.
You can’t really fine-tune it over and over again—there’s no seed or redraw strength, so every change completely alters everything.
The learning curve is basically zero, no argument there. Even my elderly relatives can crank out decent-looking images with it.