
Keywords: armwrestling
My first published Lora. I've done a few trained locally I didn't like enough to release, but the Civitai built in one is just getting much better results.
This is my first attempt at doing a pose one - as we all know, Stable Diffusion does really bad at anything involving close contact. I've done some quite acceptable Arm Wrestling images using Controlnet but having seen Bing generate great Armwrestling images reliably, I thought a lora could be worth a shot.
This is the results, and I felt it good enough to release. Due to this first dataset, it's biased towards man vs woman matches, but can generate WvW and MvM, so I decided to just call it Armwrestling rather than Mixed Armwrestling. In reality, I'd use this to generate the pose (and most importantly, the grip) and then inpaint what I wanted for the figures.
However, all the demo images are just generated with no editing or inpainting. There's some image to image ones, but they're all based on images generated by this lora - though there's a couple with some extra Loras in.
Reviews and tips appreciated. Especially tips - it costs me 500 buzz to do a training, so to do more Loras, I'm reliant on tips.
描述:
训练词语: armwrestling
名称: 255420_training_data.zip
大小 (KB): 12417
类型: Training Data
Pickle 扫描结果: Success
Pickle 扫描信息: No Pickle imports
病毒扫描结果: Success
名称: armwrestling.safetensors
大小 (KB): 36980
类型: Model
Pickle 扫描结果: Success
Pickle 扫描信息: No Pickle imports
病毒扫描结果: Success
