Hello Richard!
Somebody still had to include the elements of Vermeer in the
catalog of options. That's where the limiting comes in.
Well yes and no.
It gets trained with a large number of pictures and their corresponding descriptions. Usually from huge catalogs.
So there is no person selecting anything, except lets throw at it as much data as we can find.
But there is still the human element of programming it right.
There is nothing self-determining really.
There are all kinds of filters that can be applied to photos to
make them look like "in the style of" already.
Yes, the only difference is that this is absolutely NOT what DALL-E does. It creates a new picture based on a huge library of existing input pictures. That is a extremely big difference!
True enough.
--
../|ug
--- OpenXP 5.0.57
* Origin: Mobile? Join CHAT here:
https://tinyurl.com/y5k7tsla (1:153/757.21)