Testing the learning pen on anima-baseI tried to make it possible to use different tags for different types of artistic vision.It turned out pretty well to separate the different directions, but in different iterations some concepts were lost.v1In this version, the Render versions of the style without the liner behave well with them, but the black-and-white sketch versions are very difficult to pull out of the model. In subsequent iterations, I will try to complement the dataset with all possible images and styles.All styles can be combined and to enhance them, you can write the tag several times. In anima, it is not recommended to put a priority through (tag:1.3), this greatly breaks the understanding of the model.Tag Guide:@JadonyBW - Render version without line.@JadonyColor - Liner versions with painting.@JadonyRender - A render of the thin-line version.@JadonySketchClean - There should be sketches, but due to improper training, the effect is weak. If you only add grayscale and sketch to the hints.@jadony - A common tag for everything

Jadony style

The first stable version. It feels like it has not yet learned most of the concepts that are in the plans, but it already produces excellent images in the Render style and with light line art

Files with SHA256: 4842e83cdc1e5f75a6ada1117bc627a537c672c32f1fc7acc821d59d87d9fcdb