Re-trained model and edited some weights now v2 makes better results in realistic faces, more contrast images, No img2img used for sample images,
Vas is vae-ft-mse-84000
Basic Merge of Deliberate, AnyV3, AnyTWAM and Dreamshaper+AnyV3 Vae inside
for making stylized characters. Faces on preview images are built with standart Inpaint of same prompt with increased resolution, no other img2img
