File size: 2,427 Bytes
552170d 7f0ef9a 552170d 5d2195a 0c9527e 552170d 5d2195a f58a47d 5d2195a 552170d 4bf134b b261654 552170d 4bf134b 552170d 7dc746a b261654 552170d 7dc746a 7b0d2c4 552170d b261654 7dc746a 552170d 7dc746a 552170d 7dc746a | 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 | ---
license: creativeml-openrail-m
language:
- en
pipeline_tag: text-to-image
library_name: diffusers
---
# LiteVision-v1
LiteVision-v1 is a lightweight (only about 2GB), fast Stable Diffusion 1.5-based model optimized for low-step inference using LCM scheduler.
- Built for low-end GPU.
- Minimal steps.
- Clean results.
# Benchmark test on P100 GPU
## 🚀 Key Features
- 🔹 Based on Stable Diffusion 1.5 architecture
- 🔹 Optimized for LCM (Latent Consistency Models)
- 🔹 High-quality output in **4–8 inference steps**
- 🔹 Works with standard `StableDiffusionPipeline`
- 🔹 No custom pipeline required
- 🔹 Fully compatible with Diffusers 0.36.0
---
## ⚡ Recommended Settings
For best results:
num_inference_steps = 6 <br>
guidance_scale = 1.5 <br>
scheduler = LCMScheduler <br>
Lower guidance gives cleaner results.
Higher guidance may introduce instability due to LCM behavior.
# 🧠 Why LiteVision?
Traditional SD 1.5 models require 20–30 steps for good quality. LiteVision ONLY requires about 6 steps.
# 🛠 Usage
```python
import torch
from diffusers import DiffusionPipeline
pipe = DiffusionPipeline.from_pretrained(
"HyHorX/LiteVision-v1",
torch_dtype=torch.float16
).to("cuda")
image = pipe(
"cinematic cyberpunk city, ultra detailed",
num_inference_steps=6,
guidance_scale=1.5,
height=512,
width=512
).images[0]
image.save("litevision_output.png")
```
# ⚠ Notes
Designed specifically for LCM scheduler.
Not tuned for traditional DDIM/PNDM high-step sampling.
Use FP16 for optimal performance.
# 👤 Author
HyHorX
# License
This model is open access and available to all, with a CreativeML OpenRAIL-M license further specifying rights and usage. The CreativeML OpenRAIL License specifies:
- You can't use the model to deliberately produce nor share illegal or harmful outputs or content
- The authors claims no rights on the outputs you generate, you are free to use them and are accountable for their use which must not go against the provisions set in the license
- You may re-distribute the weights and use the model commercially and/or as a service. If you do, please be aware you have to include the same use restrictions as the ones in the license and share a copy of the CreativeML OpenRAIL-M to all your users (please read the license entirely and carefully) Please read the full license here |