G-Shirabe SDXL Lora:

Note: The errors with the first upload were resolved. It is now usable with much less headache (unless you aim for gearskating pics without work/rng lol)

A new G-Shirabe for the beefier GPU lora users. This is similar to G-ShirabeV2 in scope. Tag's are very similar and this was not trained using score_9 anime etc. It makes mulit concept (my focus) character loras much more difficult than they have to be to balance. Also, the score system is going to get revamped in the future anyway so lets wait for them to fix it first.

What happened to G-Shirabe V3 final? Well, SDXL became more widespread while I was in the lab for one thing and I wanted to give the base of the G-Shirabe set a facelift. At the moment, G-ShirabeV3 Final will be out when it is. It will likely be in the form of Lycoris from my current experiences.

For now, it's recommended to use this (if you are able to run SDXL models), for all basic images with G-Shirabe and to keep using V3 temp for exploring the very WIP rest.

What's Inside:

• The Dress and Coat primary outfits from Symphogear G season.

• Symphogear Shulshagana G season version.

• Shirabe using a heavily G bias dataset.

• Gearskating Concept (difficult): The main method of movement Shirabe uses in battles.

• Phonic Gain Concept: The aura effect used by Symphogears. Can be triggered even without Shulshagana tag.

What Isn't:

• All the items and most additional 'extra content/concepts that are in V3.