Meowscles Shadow

Look out, behind you! Its Meowscles shadow creeping up on you!

This lovable Fortnite character lurks, preying on all our hearts. Its lovable personality and deliciously toned physique makes most of us want him to stay.

This model was requested by @zeniph_852 (Miru), who provided the training data, I did the image captioning and LoRA training.

These LoRAs were trained on the Noob AI XL model.

The vPred version was trained on vPred 0.9R and shows to be the better-trained version.

The epsilon Pred version was trained on epsilon-pred 1.0 and shows some deficiencies, but understands the character well.

On all versions:

⯄ This LoRA was trained on Meowscles Shadow form and not the original Meowscles (the one which looks like a calico cat with dot eyes).

⯄ This LoRA is capable of generating Meowscles (Shadow) with a simple "meowscles" prompt. No need to add anything else. However, when other artists are used: one may need to add the other prompts in order for Meowscles (Shadow) to gen properly.

⯄ The "flexing" may need to be added to the neg in some gens if the hands are up even though no prompt for "hands up" is present. This is due to most training images source hacving Meowscles (Shadow) with its hands up.

⯄ A strength of 0.3 can influence character generation. The Meowscles (Shadow)'s features will influence the gen, but not necessarily generate a Meowscles (Shadow). Use this if you want to influence/boost a particular character/gen that you are creating.

⯄ Both LoRAs were tested extensively at a strength of 0.7 . I recommend that the vPred version of the LoRA be used at this strength. Most artists styles will work on the vPred model with this strength.

⯄ The Epsilon Pred model may need more strength in order to generate Meowscles with accuracy. Scars and earring placements are irregular with 0.7 strength. If this is not an issue: 0.7 strength in the epsilon pred model may be good for you. But if accuracy is a concern: increase the strength.

⯄ On both LoRAs: A strength of 1 may gen Meowscles in 3d instead of it being generated in an artist style. If this is desired: use a strength of 1. The LoRA may become slightly disobedient with a strength of 1 (it does not follow prompts). If you notice this: lower the strength.

Happy Genning!