Stylization by Zdzislaw Beksinski.

The training was conducted on the CIVITAI website.

The training dataset is 1024 x 1024

360 photos were used.

Sometimes, when using LoRa models, there are still problems with limbs, whereas on a pure DEV model, there are much fewer of them, keep this in mind when generating.

we have to temporarily make early access to the model in order, if possible, to slightly compensate for the BUZZ spent on training.

it turned out to be a fairly flexible model that follows the promt quite well