Trained with 4K photos of a Blood Angels Space Marine Army, however !

V1 is auto-tagged with the same caption root and incorrect/inaccurate/basic captions!

This serves as an example of what will happen to pristine data which is not correctly captioned

future versions will address this and it will be explained in my tutorial series on improving AI datasets for training image models.

Full captions will be supplied in each version once they are collected into a text only format.
Find them in my Prompt Lists (DJZ-Nodes on Github) or check the "Training data" download option here on Civit.