Finally managed to released an updated and improved version of the MechArmor Lora.
This lora should help a lot in creating images of bulky, massive and powerful robots, mechs and cyborgs.
Recommended strength is 0.6 - 1.0.
Please feel free to add your images to the gallery!
Description
New merged version from different training trails and stages. Provides vastely better results compared to the previous version.
FAQ
Comments (6)
Just as a fyi: Loras created with OneTrainer currently don't work in Forge, just in ComfyUI (via this fix). I've stumbled upon this error in the console: [LORA] LoRA version mismatch for KModel
Started digging a bit and came across these github issues in Forge:
https://github.com/lllyasviel/stable-diffusion-webui-forge/issues/1747
https://github.com/lllyasviel/stable-diffusion-webui-forge/issues/1615
I currently train only in Onetrainer, but usually use Comfyui to merge and extract.
But the Onetrainer loras I have created so far work in Comfyui, I guess the issue are the nodes used for merging and extracting, resulting in Loras that work in Comfy but not necessarily in other software packages.
@Random_Maxx Yeah, the issue lies with Forge. As mentioned in the original post Comfy fixed it in a file lora.py. Found corresponding one in Forge and there is a todo comment there. Trying to fix it with the help of ChatGPT atm, but not having much luck so far. I'm not good at that. 😋
@OneViolentGentleman wish you good luck, hope you can fix the issue!
I... love you
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.