Getting a feel for using maquines in pre-sketch stages of character design. This workflow started out with a posemy.art image export, followed by passing the result through a figure sketch LORA to create the starting base. From there I moved through the sketching and inking process as can be seen above before experimenting wtih three different filling methods — the clip studo auto fill, which produced a nice base layer for further work, an experimental wet wash, which I think could produce wonderful results in certain contexts, such as satin clothes and waters, and finally a manga toned fill, which I very much like, although whether copying the manga tone style when you don’t have the limitations of the manga print industry and can instead use colours, which are a lot more eye-catching, is a question I’ll have to wrangle with on a case-by-case basis.

The A.I has a mind of its own! This is one of the largest barriers facing artists who want to use generative A.I in their actual workflows. It’s all very well to type prompts into Stable Diffusion or MidJourney and get amazingly beautiful images spat out at us, but if that image doesn’t represent a true realization of what we’re seeing in our mind’s eye, and, more importantly, what we want to communicate to our audiences, then it’s all so much piss in the wind. The main problem is that generative A.I has mostly been presented to us as a toy. You press a button and pretty pictures pop out. But that’s not how actual artists work. We go through a process of conceptualization from project scoping to mood-boarding, to sketch, to lineart, to coloring, and beyond. What I present here is a very quick and dirty attempt to create a workflow that uses A.I and other aritst assist tools to either massively speed up the process of character art creation, or else to help plug gaps in a skills portfolio required for that process.