I think they are reacting to competition. MidJourney is amazing, was easier to get into, gives you commercial rights, and frankly I found more fun to use and even better output in most instances.
The only thing I don’t like about MidJourney is the Discord based interface. I think I can grok why Dave chose this route as it bakes in an active community element and allows users to pick up prompt engineering techniques osmotically… but I’d prefer a clean DALL-E style app and cli / api access.
In case you don’t know, you can at least PM the MidJourney bot so you have an uncluttered workspace.
It’s clearly personally preference, but I loathe Discord but love it for MidJourney. As you said, there’s an interactive element where I see other people doing cool things and adapting part of their prompts and vice versa. It really is fun. And when you do it in a PM, you have all your efforts saved. DALL-E is pretty clunky in that you have to manually save an image or lose it once your history rolls off.
I've completely changed my mind after spending the last few days neck deep in it around the clock. Sleep is overrated! MidJourney is awesome and the way it's implemented within Discord is a masterstroke of elegant simplicity.
Thanks. Yeah fair point; I haven’t ponied up for a subscription yet so am still stuck in public channels and often find my generations get lost in the stream. Imagine you’re right and having the PM option would change the experience drastically for the better albeit still within Discord’s visually chaotic environment.
I have access to both and they're good for different things. DALL-E seems somewhat more likely to know what you mean. Midjourney seems better for making interesting fantasy and science fiction environments.
For comparison, I tried generating images of accordions. Midjourney doesn't really understand that an accordion has a bellows [1]. DALL-E manages to get the right shape much of the time, if you don't look too closely: [2], [3]. Neither of them knows the difference between piano and button accordions.
Neither of them can draw a piano keyboard accurately, but DALL-E is closer if you don't look too hard. (The black notes aren't in alternating groups of two and three.)
Neither of them understands text; text on a sign will be garbled. Google's Parti project can do this [4], but it's not available to the public.
I expect DALL-E will have many people sign up for occasional usage, because if you don't use it for a few months, the free credits will build up. But Midjourney's pricing seems better if you use it every day?
MidJourney definitely struggles more with complex prompts from what I saw. If you like the output more, that’s subjective, but I think DALL•E is the leader in the space by a wide margin.
I think both have strengths and weaknesses, but I don’t disagree DALL-E in most instances is technically better at matching prompts. But I often enjoyed, artistically, the results of MidJourney more; it just felt fun to use and explore.
I guess it depends on what you like/enjoy? It's not good at photorealistic, but it comes up with some pretty entertaining (and pretty?) 'arty' type stuff. I go on regularly just to play around for fun.