Dev Log: Reforging

This is what “reforging” looks like when I create a AI art.

I usually start with pose reference. Real photos. I bring those into Photoshop, push the pose, proportions, and silhouette until the foundation feels right.

From there, I run it through image models to clean it up and push toward photoreal. Sometimes I’ll try different background passes with tools like Nanobanana or Seadream, to explore mood and context. Props get rendered separately and composited back in Photoshop. Weapons, gear, accessories, most often handled as their own pieces.

And that’s only the beginning.

After that, I move into custom models I’ve trained, then do heavy manual editing in Photoshop. Paintovers. Value control. Edge cleanup. Lighting tweaks. Color balance. Detail passes.

People think AI art is text-to-prompt.

It never works  that way for me, not to get the results I am looking  for.

This is work. Iteration. Judgment calls. Throwing away versions. Rebuilding parts that don’t hold up. It is often removing detail even if I like it to make the piece read better. The AI gives options. I decide what survives.

I have a long history taking drawing classes, art history, design, production. That training is still doing the important work here. AI doesn’t replace that. It just lets me move faser through exploration.

Could I draw these characters by hand?

Yes.

But I like the results I’m getting. And at the end of the day, that is what matters.