I Made Game Background Concept Art with Stable Diffusion
I've been generating fantasy game background concept art with Stable Diffusion XL recently, and the results are surprisingly good. By using ControlNet to set the composition and applying LoRA for specific styles, I'm getting images that are almost at a commercial level. However, I do need to touch up some slightly lacking details in Photoshop. If you're interested in AI art, let's explore together!
8 answers
Wow, the results are amazing! I use SDXL too, and using ControlNet for composition is a real game-changer. Would you mind sharing which LoRA you used?
I'm working on something similar too. For areas lacking detail, try using inpainting—it's often faster than Photoshop.
A fantasy setting sounds really cool! I'm still a beginner, so I think I need to start learning from ControlNet. Do you have any recommended tutorials?
I was skeptical about AI art, but seeing it used in actual work like this makes me interested. Is it also widely used in the gaming industry?
I tried something similar with SDXL too, but Photoshop touch-ups were essential. Especially with lighting effects, it feels like AI still can't quite nail it perfectly.
Oh, commercial-grade? That sounds intriguing! Could you share a few sample results? I'd love to see them firsthand.
I'm also researching this, and I recommend a combination of tile upscale + detail correction for enhancing details. It takes a bit of time, but the quality improves significantly.
I'm active in the AI art community, and these practical tips are really helpful. I've been looking for someone to research together, so feel free to reach out!