Text-to-Image
Diffusers
lora
ernie
image-generation
anime
illustration
flat-style
vector-art
cel-shading
stylization
character-design
Instructions to use HexDuck/FMIX with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Diffusers
How to use HexDuck/FMIX with Diffusers:
pip install -U diffusers transformers accelerate
import torch from diffusers import DiffusionPipeline # switch to "mps" for apple devices pipe = DiffusionPipeline.from_pretrained("baidu/ERNIE-Image", dtype=torch.bfloat16, device_map="cuda") pipe.load_lora_weights("HexDuck/FMIX") prompt = "Fimix:, a highly energetic magical girl with twirling pink pigtails, shouting happily while swinging a glowing star-topped wand, wearing a frilly cosmic-themed dress with floating ribbons, casting a massive bright neon spell that materializes the word \"FMix\" in glowing solid magical energy, a vibrant abstract dimension of shooting stars and geometric auroras, dynamic low angle shot, intensely bright magical rim lighting casting sharp cell-shaded shadows, explosive and magical atmosphere, premium flat style aesthetic, flat colors background, vector art style background, cell shading" image = pipe(prompt).images[0] - Notebooks
- Google Colab
- Kaggle
- Local Apps
- Draw Things
- DiffusionBee
File size: 4,438 Bytes
d0903c2 6bc8d2e d0903c2 6bc8d2e d0903c2 6bc8d2e d0903c2 6bc8d2e a0102cf 6bc8d2e d0903c2 2cf825a d0903c2 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 | ---
license: other
base_model:
- baidu/ERNIE-Image
pipeline_tag: text-to-image
library_name: diffusers
tags:
- lora
- ernie
- text-to-image
- image-generation
- anime
- illustration
- flat-style
- vector-art
- cel-shading
- stylization
- character-design
model_type: lora
trigger_words:
- Fimix:
widget:
- text: >-
Fimix:, a highly energetic magical girl with twirling pink pigtails, shouting happily while swinging a glowing star-topped wand, wearing a frilly cosmic-themed dress with floating ribbons, casting a massive bright neon spell that materializes the word "FMix" in glowing solid magical energy, a vibrant abstract dimension of shooting stars and geometric auroras, dynamic low angle shot, intensely bright magical rim lighting casting sharp cell-shaded shadows, explosive and magical atmosphere, premium flat style aesthetic, flat colors background, vector art style background, cell shading
output:
url: Examples/01.png
- text: >-
Fimix:, a woman with voluminous auburn curls and copper-colored eyes, wearing a brown leather corset with brass gears and buckles over a white ruffled blouse, mechanical brass goggles resting on her forehead, holding a steaming copper pipe, background of a Victorian clockwork city with airships, sepia-toned anime style, high detail on metal reflections and leather grain, flat colors background, vector art style background, cell shading
output:
url: Examples/02.png
- text: >-
Fimix:, space explorer with short silver hair, a smudge of dirt on her cheek, gazing upwards in absolute awe, wearing a sleek white and orange spacesuit, scratched helmet visor reflecting a massive galaxy, standing on a barren alien landscape covered in jagged obsidian rocks, low angle wide establishing shot, flat colors background, vector art style background, cell shading
output:
url: Examples/03.png
inference:
parameters:
guidance_scale: 4.0
num_inference_steps: 50
width: 1024
height: 1024
---
# 🌌 FMix (Flat Style Mix)
**FMix** is a premium stylization LoRA designed to achieve a high-quality, professional **Flat Style aesthetic** with heavy, clean cell shading.
This model excels at creating vibrant, vector-inspired illustrations that maintain sharp outlines and bold colors. Whether you're aiming for modern graphic design, clean character sheets, or stylized anime art, FMix delivers a crisp 2D look that feels both modern and sophisticated.
---
## 🧩 Recommended Settings
* **Trigger word:** `Fimix:`
* **Strength:** `0.8 ~ 1.2`
* **Resolution:** `1024x1024`
* **Samplers:**
* DPM++ 2M
* Euler Ancestral
* res_2m
* LCM
---
## 🚀 Advanced Setup (Recommended)
For optimal results and faster generation, it is recommended to combine this LoRA with a Turbo LoRA.
**Suggested stack:**
```
FMix LoRA (1.1) + Turbo LoRA R64 (0.8)
```
**Turbo LoRA link:**
[https://civitai.red/models/2551262/ernie-turbo-lora-extracted?modelVersionId=2867130](https://civitai.red/models/2551262/ernie-turbo-lora-extracted?modelVersionId=2867130)
**CFG Scale:**
* `1.0 ~ 2.0`
*(Keep it low for Turbo compatibility)*
---
## 💡 Pro Tip
FMix is highly responsive to prompt modifiers.
To push the flat aesthetic to its limit, use:
```
Fimix:, [your prompt here], flat colors background, vector art style background, cell shading
```
---
## 📝 Creator Notes
### Iterative Perfection
This LoRA is the result of extensive experimentation. Over **10 different versions and training configurations** were tested to achieve the ideal balance between style and flexibility.
### Artistic Control
FMix is designed to be versatile—it handles complex character details (gears, goggles, intricate clothing) while forcing them into a clean, flat-shaded 2D space.
---
## ⚠️ Known Limitations & Recommendations
* **Rock-Solid Stability:** No known issues.
* **Mutation Resistant:** Strong resistance to common AI deformities.
* **Anatomical Consistency:** Maintains integrity even in complex poses.
* **Reliable Output:** Designed as a plug-and-play solution for consistent, clean results.
---
## 👤 Contact & Support
* 💬 Discord: `hex_duck`
* 💖 Brazilian Discord Server:
[https://discord.gg/Q29RtePts5](https://discord.gg/Q29RtePts5)
If you enjoy this work and want to support future development:
☕ [https://ko-fi.com/retrowavex32](https://ko-fi.com/retrowavex32) |