As soon as the model was trained, the first step was randomly generating a large set of images w. a simple prompt ("golem, detailed, realistic, 3D rendering")
Each golem can be extracted by removing the images' backgrounds (reco: @photoroom_app).
Some of the designs are amazing. However, the golems all look very similar to each other. Let's separate them into categories.
The following step is to prompt some more precise description (e.g. "golem with stone, lava, steel, fire")
Here's a first set of 16 golems that look consistent.
Because that initial set of lava golems was a bit simplistic, I used #img2img and an improved prompt to slightly increase the level of details (e.g. "3D rendering, trending on Artstation, etc.)
Here's the result:
And now, we can keep the same prompt and introduce many other variations while keeping the overall shape, posture, and level of detail.
Here are the ice golems (like the ones in Clash of Clans)
The forest golems! My favorites 🌲🌳
The sand golems 🏜️
And... the golden golem (directly inspired by the golden golem in Minecraft, via #img2img)
Now it's also possible to use the same finetune, and generate pictures that could be used elsewhere than in the game (ads, splash screens, etc).
Just add "Greg Rutkowski" in the prompt and start seing some cinematic composition, dramatic lighting, etc.
Some other tests I made...
As always, feel free to add any questions, feeback, remarks (and don't forget to like/follow/RT if you find this content interesting. I really appreciate it!)
Tomorrow, I'll keep sharing other explorations with #StableDiffusion 🚀🚀
(11/06 update - here's another run to create more characters, this time with "Space Marines" heavy infantry)
The model was trained on just 11 pictures (!), with only 1500 training steps, which tuned out to be quick (20 min).
As before, the first step is to "explore“ the model with a few generic prompts. The goal is to find the modifiers that will keep a consistent style going forward.
Once the "stable modifiers" are found, it's time to select some of the best output and remove the background when needed.
"A dwarf, detailed, trending on Artstation, Clash of Clans"👇
I also used #img2img to instantly generate dozens of variants, "inspired“ by a single original photograph.
This provides consistent assets (similar shape, size, or materials) with some slight variations. It's up to the artist/user to select which one looks best.
Same thing here, using #img2img - however, I prompted "steel chest" instead of "wooden chest"
While it's not 100% perfect, there's still more steel in these chests than in the previous ones.
Also, some of the assets are disjoint or show anomalies. Some fine-tuning is necessary