Trained on 1280x768, may work on other landscape dimensions.
Recommended RealVisXL
Usage
[your skin description] [pov, ak47skins, gloves, counter-strike, holding weapon] [other styles] [lora]
This needs more work, in futures updates/weapons Im going to try to remove the background from the dataset.
At the moment it does not work right in some checkpoints.
Best strength is 0.6-0.8, at higher starts showing more artifacts.
⚡️BUZZ FOR BEST IMAGES USING THE LORA ⚡️
Description
FAQ
Comments (3)
Would you be willing to share the training data, so I understand how to make other lora with a similar idea but different themes?
You need your dataset to contain the same background and positions, but variety on the object you want to alter, then remove all the tags but ones related with the object, so then the prompt only applies to that object
@ZyloO currently working on a car lora and this seems impossible to achieve for this exact project. Especially when the cars are multi-million dollars, and my net worth is 23 cents.