Holotard - Hololive VTuber LoRA Collection
| Property | Value |
|---|---|
| License | CreativeML OpenRAIL-M |
| Language | English |
| Task | Text-to-Image Generation |
What is holotard?
Holotard is a comprehensive collection of LoRA (Low-Rank Adaptation) models and textual inversions specifically designed for generating high-quality images of Hololive VTubers using Stable Diffusion 1.5. This collection includes carefully crafted models for VTubers across HoloEN, HoloJP, and HoloID branches, with many characters featuring multiple outfit variations.
Implementation Details
The repository contains merged models with hll4-p3-ep8 (latest vtuber-finetuned model) and includes the blessed2 VAE integration. The LoRA models are designed to work with stable-diffusion-webui and can be implemented using the syntax <lora:filename:1> in prompts.
- Multiple base models including HeavenOrangeVtubers, AOM3, AOM2hard, and Grapefruit4.1
- Extensive collection of character-specific LoRAs with multiple outfit variations
- Built-in VAE support for improved color reproduction
Core Capabilities
- Generate high-quality images of Hololive VTubers with accurate character details
- Support for multiple outfit variations per character
- Compatible with popular Stable Diffusion workflows
- Extensive coverage of VTubers from all Hololive branches
Frequently Asked Questions
Q: What makes this model unique?
This collection stands out for its comprehensive coverage of Hololive talents and the inclusion of multiple outfit variations per character, allowing for diverse and accurate VTuber image generation. The models are optimized for use with specific base models and include VAE integration for superior color reproduction.
Q: What are the recommended use cases?
The models are ideal for creating fan art, thumbnails, or reference images of Hololive VTubers. They work best with the stable-diffusion-webui interface and can be enhanced with recommended additional resources like EasyNegative for improved results.
