Lora trained on color images by 1930-1970s soviet children's book illustrator Genrih Valk. There were too few of them in dataset, but it is impossible to find more in good quality (most of his illustrations were black and white linearts). Try to combine lora with tokens like "boy, girl, 1960s, soviet, sun, happy, forest, fox, hare, hedgehog" and etc. with weights from 0.6 to 1.1