That’s hilarious. Here’s what Midjourney makes of it:
2 Likes
I tried everywhere for the restaurant at the end of the universe, but I guess I just don’t know how to work it.
That one is GREAT and actually could make an awesome Infocom cover. I don’t have access to DALL-E so I occasionally grab ideas on NightCafe. Even if this type of art is not used, it’s great for mood-board type of inspiration.
Also, the people at that center table are paying a premium for that view.
Here’s another one for experimentation from hotpot(?).
They want people to pay $8.50 to license the AI images (and I suppose get it larger and upscaled.)
2 Likes
That one’s Midjourney v4, which blows DALL-E away.
Here’s a “mood” image I made for my WIP. Guess the theme.
1 Like
<uncannyvalleyofficeparty>
1 Like
Okay is it Disney’s abandoned “SCP Job Interview Experience Starring Brooke Shields and Mr. Burns Jr.” (as animatronics)?
4 Likes
Got it in two.
1 Like
Serious or sarcasm? 
1 Like
Neither. It was an outright lie. 
kind of an x-files parody, but that’s not quite accurate.
1 Like
LOVE it! What did you use? …Dall E, right?
Midjourney.
1 Like
Might have already been mentioned, but DALL-E is superseded by stable diffusion. There are several online demos. here one.
4 Likes
Yes, there’s a complete web generator/editor available now at DreamStudio.ai.
Plus a Python SDK for advanced use.
I was able to generate this character via the web site:
3 Likes
Thanks for mentioning that. I actually already tried out DreamStudio. I was interested in feeding it a “seed” picture and a description to see if I could control the content a bit more. It claims you can do this.
Well, my results were not especially impressive. it mostly ignored my picture completely. Maybe i was doing something wrong. But i like the idea of giving a picture as a guide.
2 Likes
There are coding examples where you can present an image as a starting point for the next. But that image is going to get morphed quite radically.
I’m going to continue to experiment. I think the key is to home in on a particular set of text constraints which repeat a particular style.
Quoting names of artists, photographers, styles, etc is the way forward from what I can see.
It would be great to be able to generate a cast of half-a-dozen characters like this guy above, all in the same art style, for an IF Agatha Christie scenario…
1 Like
Midjourney gives you the ability to use an image as “inspiration” and it works reasonably well with people. For instance:
I’ve never been able to get different shots of the same person before Midjourney v4.
3 Likes
It seems to know how to apply light to a portrait shot. Can you ask for different styles, eg: noir (monochrome, harsh single light source)?
1 Like
