Cool Pictures from the Mind of a Machine: AI Generated Pictures

Spoiler the Iron Tower looming over a desolate landscape, digital fantasy art in the style of Laurel Burch :


4ba297aee5624bd4afe98502fec00c10.jpg

5d1ad1091413497cb86e5f128c24f091.jpg

c670ebe302c94cb9a5154d12c771d431.jpg

07cbe14cbcce4c8e97ed82a2e7b994c2.jpg


 
Spoiler the Iron Tower looming over a desolate landscape, digital fantasy art in the style of Kilian Eng :

77ecf5a8a83d41258c24bcaa7c356e6e.jpg

17bbf2bbd2ca432a87279d34cbcd5744.jpg

e20989b6b99645e0aefad72c4287033b.jpg



This is really not Kilian Eng's style at all...
 
Midjourney? Not very good at style imitation. It has its own midjourneylian style. Stable diffusion plus the proper LORA is better in that aspect.
 
Xl or 1.5?
 
1.5 is better at styles I think. For training XL they allowed artists to request being excluded so no images by them would be used to train XL base model. You can add them using available LORAs made by the community of course.
 
A big reason why the AI isn't replicating the artist's style perfectly is because of the 'digital fantasy art' prompt I'm putting in; I'm mostly experimenting to see how the AI tries to reconcile two different style requests in the same prompts, which is why the Marc Chagall images seem off style-wise.

Besides, I'm using mage.space, and LORAs are a premium feature.
 
Weirdly shaped guitars were kind of Prince's thing, though... :D
 
20240331202450-[5.0][dynavisionXLAllInOneStylized_releaseV0610Bakedvae][3936]-HiResFix(latent...jpeg


Spoiler another one... :

20240331202611-[5.0][dynavisionXLAllInOneStylized_releaseV0610Bakedvae][3940]-HiResFix(latent...jpeg

 
They're Will Smiths from alternate universes.
 
Top Bottom