Important! Seed models are intended for enhancing new merges. This isn't a series that follows a style. Every model is meant to boost a very specific action or theme. These are for checkpoint creators only. If you merge new checkpoints and want to boost a certain element or ability, then these models serve that function. They are filled with Loras related to their titles. So TrekWars_seedModel will boost space adventuring sci-fi themes significantly when merged as the tertiary model.

When I say "filled with Loras", we're talking hundreds at a time, each Lora is merged more than once. I use DARE merging only to add the higher weights. No Lora is fully merged. This allows the resulting checkpoint to "eat the weights as a free lunch" by absorbing them. I have developed a process that is extremely effective.

Lora types in each model:

• Themes

• Actions or poses

• Detailers for quality/anatomy

• Lighting and styles

• Any Lora that supports the model's overall theme or aesthetic

Base models vary. I mostly use my own models for the base. Any base model downloaded from Civtai will be in the title & credited in the description.

The modified DARE method I use allows me to merge any architecture into SDXL based models

Loras with appropriate permissions are used along side my own Loras. Questions about what was used will be answered. However, I've got a library of over 2,000 Lora across 10 different arch-types. So I don't keep a log of what was used or how many times it was merged.

• Since there is a large volume of different Lora types in each model, it can be unstable for direct generation.

• Showcase images are generated using the model as it is. But my process works best when used as a secondary model merge.

• It adds more stability & consistency to the output.

• These are handy to have for any future merge where you want to add more of a foundation to a specific concept.

• Best results overall is to use a base model that is at least familiar with the seed's main concept.