Liberte.Redmond: A Generalist SD 1.5 Base for Versatile Image Generation
Liberte.Redmond is a generalist checkpoint fine-tuned on SD 1.5. Learn its trigger settings, how it fits the Redmond model family, and how to use it.
On this page
Liberte.Redmond: A Generalist SD 1.5 Base for Versatile Image Generation
TL;DR:
- Generalist checkpoint fine-tuned on Stable Diffusion 1.5 — no trigger words needed.
- Recommended: DPM+ SDE/2M or 2M SDE, 30 steps, test with and without negative prompts.
- Serves as the foundation for a whole family of specialized LoRAs by ArtificialGuyBR.
What Is Liberte.Redmond?
Liberte.Redmond is a generalist checkpoint model fine-tuned on Stable Diffusion 1.5. Unlike style-specific LoRAs, it aims to handle a broad range of subjects — realistic portraits, artistic illustrations, cars, game characters, and more — without requiring specialized trigger words or embeddings.
The author, ArtificialGuyBR, released it as a base model intended to anchor a dozen specialized LoRAs and fine-tunes. It sits alongside Freedom.Redmond (the SD 2.1 sibling) in a growing ecosystem of Redmond-family models.
How to Use It
Recommended settings (from the model card):
- Sampler: DPM+ SDE/2M or 2M SDE
- Steps: 30
- The author notes that some prompts produce interesting results even without negative prompts — worth testing both ways.
The model ships as a single fp16 SafeTensor file (1.99 GB), so it runs comfortably on most consumer GPUs that handle SD 1.5. It supports both standard text-to-image and inpainting workflows.
Where It Fits in the Redmond Family
Liberte.Redmond is the SD 1.5 anchor of the Redmond family. The author has since released specialized LoRAs built on top of it, including:
If you need a single generalist base for SD 1.5 that you can then specialize with LoRAs, Liberte.Redmond is designed for that role.
Commercial Use
The author explicitly authorized commercial use of the model in image-generation services, as long as you credit @artificialguyBR (linking to the Twitter account is appreciated). This makes it viable for studios and freelancers building on top of the model.
Common Errors and Fixes
Error | Cause | Fix | Source
|---|---|---|---
Blurry or low-detail outputs | Too few steps or wrong sampler | Use 30 steps with DPM+ SDE/2M or 2M SDE | Model card Over-saturated or burnt images | fp16 precision on weak GPUs | Try --no-half flag in A1111 or reduce CFG scale | Community Inpainting produces seams | Mask padding too low | Increase mask blur / padding in your UI | Community
FAQ
Do I need trigger words? No. Liberte.Redmond has no trained trigger words. It responds to natural prompts.
What base model is it trained on? Stable Diffusion 1.5.
Can I use it commercially? Yes, with credit to @artificialguyBR.
Is there an SD 2.1 version? Yes — Freedom.Redmond is the SD 2.1 sibling in the same family.
Sources
- Civitai model page — model card, description, downloads, sample images.