The learning process continues. In the past, when creating Loras for SD 1.5, SDXL, and Pony, I could use only headshots, and the checkpoints would intuitively fill in the bodies. However, with Flux, this is not always the case. I’ve found that similar to Blair's situation, I need to include a few body shots as a reference for Flux to work effectively.
Admittedly, I only ran Blair’s Lora through a basic boot camp test, checking solely for likeness and pose-ability. I should have conducted a more thorough test; my apologies for that.
With V2, I have included both full-body images and headshots.
I'm starting to feel burnt out again. While I have all the templates needed to test each Lora, I often get bored with the same poses and locations. This desire to try new things ends up bogging me down and causing frustration.
I have a few more Lora's to release, and then I'll take another break.
I appreciate your support. /bow
