Bringing fantasy characters to life c:
Hi there!This page is meant to explain my work in a more thorough manner.I employ a mixture of AI and regular digital artwork techniques to ensure that I can keep things at a high quality level whilst not losing any control due to excessive AI reliance. If I have a creative vision, I can bring it to fruition with no compromises.Unlike the many purely AI-based pages that you might've stumbled upon, I do not see AI as a substitutive for artistic vision and skill, but as something that can be used in conjunction with it; it is a tool that, if used properly, can help with bringing a preexisting vision to life (or to a higher quality level), rather than being utilized to pump out meaningless images by the hundreds.If you have dabbled in AI, you likely know that things get much trickier - if not outright impossible to pull off with AI alone - when you have an exact vision you want to bring to reality. Since I don't rely entirely on AI, this is not an issue for me.You can see for yourself what I mean down below!
References
Result
(Clients also provide text-based descriptions, or plenty of other references to convey pose, or perhaps peculiarities where the preexisting character material falls short -- all the differences/peculiatiries were requested.)
The difference is in the AI, too!
Even though most people are familiar with web-based AI generators, the technology itself can get much more complex than that.I use Stable Diffusion, which is ran locally and enables a much higher degree of customization. Extensive knowledge and usage of inpainting, ControlNet, LoRAs and all other sorts of plugins allow me to create a superior artwork even if I were to rely on AI alone, when compared to simply getting a Midjourney subscription or something that is a bit more user-friendly.Below, I will compare the pros and cons of my method with the typical, average AI art.
Purely AI Art
1 - Composition/pose left to chance;
2 - Forget specific tattoos or scars;
3 - Struggles with non-human races;
4 - Anatomy issues (hands, feet, eyes);
5 - Low resolution, or limited to generic upscaling;
6 - The background that comes is what you get, typically "non-Euclidean" (one floor becomes two, etc.).
7 - More often than not, doesn't really portray anything "meaningful", or a preexisting idea, i.e. generic characters meant to look "passable" at a quick glance. It's much easier to make something that looks "okay" when there's no clear goal and anything goes.
- Quick and low effort, can pump out dozens of "passable" generic artworks a day.
My Art
1 - Everything in my control, no "RNG";
2 - Specific tattoos or scars are possible, or any other specific trait;
3 - Any fantasy race possible and perfectly accurate;
4 - Perfect hands, feet, eyes, etc.;
5 - Attention, time and extra polish given to minor details, making everything look far better than a mere upscaling;
6 - You can pick any background, separate from the character itself, and it makes geometrical sense.
7 - The end goal/vision comes first, and the available tools are used to bring it to fruition. All characters are actual characters - not generic eye candy - as corroborated by the fact that I only have a couple of active OCs at best.
- Each artwork takes me as much time as a regular non-AI artwork, i.e. up to dozens of hours, depending on complexity level.
What to Expect/Commission me for:
- Fantasy characters in general, from portraits with no background all the way to full body, dynamic portrayals.
- Any fantasy non-human races are doable!
- SFW and NSFW of all degrees, including SFW/NSFW alt versions.
- I work strictly with fantasy characters. I do not create artwork based on real people.
- I do not work with underage characters.
Timelapse
Here, you can see a simplified timelapse from the very first iteration all the way to the finished piece. As I didn't record it in real time (I'm a chronic alt-tabber and it'd require way too many cuts), it's more of a glorified slideshow between iterations, but it's a good way to see how I get to the final piece nonetheless.Not displayed: Controlnet OpenPose (shown above), real-time Photoshop steps (you'll probably notice them when you see something magically pop up or change, though some are a bit more subtle).