Exciting News: I've successfully gotten the AI model to work with two LORAs at the same time!
What's a LoRA? When I generate images, I use a large AI model trained on billions of data points. Think of it as the core engine. A LoRA (Low-Rank Adaptation) is a much smaller, lightweight add-on model. It's trained on something very specific – like helping the main model recognize and generate a particular character (e.g., Cha Hae-In or Sung Jinwoo). Using LoRAs massively expands the range of characters I can create!
The Catch (For Now): I've only thoroughly tested this with the Cha Hae-In and Sung Jinwoo LoRAs together. They work incredibly well as a pair! Almost every attempt gives me the characters I asked for. 🎉 I haven't tested how well other LoRA combinations will work yet, so that's still unknown territory. 😅
A Note on Style: Combining LoRAs can sometimes subtly affect the overall look. For example, you might notice slightly more muted colors in the results than usual.
The Big Takeaway: This means you can now request characters that were impossible for me to generate before! 🥳 While I can't make any promises (especially with untested LoRA combos), there's a real chance I can create them now – or at the very least, I can definitely give it a try! 😏