I’ve used other AI image makers and they all have weird shortcomings in understanding certain images. The images you get with “Pangolin” are an abomination. (Pangolins suffer enough in the world today. These images are criminal.)
Other than that this is really really cool! I like it better than Discord. I only signed up for Discord (Midjourney?) in an attempt to get images for my Forge. It just would not work. Tailoring it to the forge is genius.
Edit: What I should have added is that it’s still early days for this type of technology and I’m thrilled to be included. I am curious if/how they will continue the programs development in understanding language and requests (if that’s the correct way to word it).
There are a lot of things it does not have a clue about. Trilobites it has no idea but it knows all about Zhangjiajie. Roadrunners are a mystery but Capybaras it knows well. It thinks Lotus and Waterlilies are the same thing. And so on and etc.
I wanted the text to show above the design, but wanted the design to come through so added some transparency. It looks like the piece is a bit of tape with the design showing through even better than in the photograph.
SO I tried it today with poor results.
attached is the screen shot of the 7 different images it created with the key words “bullet thought apple” since we are a science center founded for Doc Edgerton.
one is in “super cure” which is just a girl, the rest are just apples.
Try “bullet piercing an exploding apple” and then click the button a lot. While AI art tends to not take a lot of effort to generate, it also rarely returns a useable result on the first (or second or third or Nth…) try.
