ArtificialGuyBR

Home / Blog / Training LoRAs

Toy.Redmond LoRA for SDXL: Collectible Toy Figures

SDXL LoRA that generates toy-style figures and collectible figurines with a distinctive 3D rendered aesthetic.

2 sources cited Training LoRAs

On this page

  1. What Toy.Redmond Does
  2. How to Use
  3. Visual Examples
  4. Training and Credits
  5. Common Errors and Fixes
  6. FAQ
  7. Sources

Toy.Redmond: A Toy Figure LoRA for SDXL 1.0

If you want your subjects rendered as detailed collectible toys, Toy.Redmond is a purpose-built LoRA for Stable Diffusion XL that nails the look — plastic surfaces, vivid paint, miniature scale, and a consistent 3D aesthetic across a wide range of themes.

TL;DR:

What Toy.Redmond Does

Toy.Redmond fine-tunes SDXL 1.0 to render any subject as a collectible toy figure. The style evokes high-quality miniature figurines with glossy or matte plastic textures, bold colors, and a sense of handcrafted charm. Unlike style LoRAs that apply a painterly filter, this one transforms the entire rendering pipeline into a toy manufacturing aesthetic.

The model handles diverse subjects well — from human figurines to animals to fantasy creatures — while keeping the toy look consistent. This versatility makes it useful for creating character concept art, toy product mockups, or simply stylizing any scene into a miniature world.

How to Use

Toy figure generated with Toy.Redmond LoRA

Trigger Word

The trigger word is FnkRedmAF. Include it in your positive prompt:

FnkRedmAF, a detailed dragon figurine, glossy plastic finish, collectible toy, vibrant colors

Setting | Value

|---------|-------

Base model | SDXL 1.0 Resolution | 1024×1024 LoRA weight | 0.7–1.0 (start at 1.0, reduce if artifacts appear) Sampler | DPM++ 2M Karras or Euler a Steps | 25–30

In ComfyUI

Load the LoRA with a LoraLoader node, wire it between your checkpoint and sampler. Set the strength sliders to your preferred values (start at 1.0). Ensure the FnkRedmAF token appears in your CLIP text encoding.

In Automatic1111

  1. Place the LoRA file in models/Lora/
  2. Add <lora:ToyRedmond-xx:1.0> to your prompt
  3. Include FnkRedmAF in the positive prompt

Visual Examples

Toy figurine with detailed finish

The Toy.Redmond LoRA produces consistently toy-like results across different subjects. The second example shows how the model applies the plastic figurine aesthetic to different subjects while maintaining that collectible feel.

Another toy-style generation

Training and Credits

Toy.Redmond was fine-tuned with GPU compute provided by Redmond.AI. The model is free to use under the CreativeML OpenRAIL-M license, which permits commercial and non-commercial use with some restrictions on harmful content.

Common Errors and Fixes

Error | Cause | Fix | Source

|-------|-------|-----|--------

No toy effect | Missing trigger word | Add FnkRedmAF to prompt | HF Model Card Artifacts at high weight | LoRA weight too high | Reduce to 0.6–0.8 | Community reports Blurry results | Wrong resolution | Use 1024×1024 as recommended | HF Model Card

FAQ

Does this work with SD 1.5? No. Toy.Redmond is trained specifically for SDXL 1.0. Using it with SD 1.5 checkpoints will produce garbage output.

What does FnkRedmAF mean? It's the instance token used during training. Including it in prompts tells the model when to activate the toy style.

Can I combine this with other LoRAs? Yes, but test at reduced weights (0.5–0.7 each) to avoid conflicts. Combine with detail-enhancing LoRAs for best results.

Sources