---
tags:
- text-to-image
- stable-diffusion
- lora
- diffusers
- template:sd-lora
widget:
- text: >-
    Beautiful woman, bubblegum pink, lemon yellow, minty blue, futuristic,
    high-detail, epic composition, watercolor.
  output:
    url: images/sd21_base.png
base_model: stabilityai/stable-diffusion-2-1-base
instance_prompt: null
license: mit
---
# TCD-SD21-LoRA

<Gallery />

## Model description 

Official SD21(base) Model of the paper [Trajectory Consistency Distillation](https:&#x2F;&#x2F;arxiv.org&#x2F;abs&#x2F;2402.19159). 

For more usage please found at [Project Page](https:&#x2F;&#x2F;huggingface.co&#x2F;h1t&#x2F;TCD-SDXL-LoRA&#x2F;)

Here is a simple example:

&#x60;&#x60;&#x60;python
import torch
from diffusers import  StableDiffusionPipeline, TCDScheduler

device &#x3D; &quot;cuda&quot;
base_model_id &#x3D; &quot;stabilityai&#x2F;stable-diffusion-2-1-base&quot;
tcd_lora_id &#x3D; &quot;h1t&#x2F;TCD-SD21-base-LoRA&quot;

pipe &#x3D; StableDiffusionPipeline.from_pretrained(base_model_id, torch_dtype&#x3D;torch.float16, variant&#x3D;&quot;fp16&quot;).to(device)
pipe.scheduler &#x3D; TCDScheduler.from_config(pipe.scheduler.config)

pipe.load_lora_weights(tcd_lora_id)
pipe.fuse_lora()

prompt &#x3D; &quot;Beautiful woman, bubblegum pink, lemon yellow, minty blue, futuristic, high-detail, epic composition, watercolor.&quot;

image &#x3D; pipe(
    prompt&#x3D;prompt,
    num_inference_steps&#x3D;4,
    guidance_scale&#x3D;0,
    # Eta (referred to as &#x60;gamma&#x60; in the paper) is used to control the stochasticity in every step.
    # A value of 0.3 often yields good results.
    # We recommend using a higher eta when increasing the number of inference steps.
    eta&#x3D;0.3, 
    generator&#x3D;torch.Generator(device&#x3D;device).manual_seed(0),
).images[0]

&#x60;&#x60;&#x60;
![sd21_base.png](https:&#x2F;&#x2F;cdn-uploads.huggingface.co&#x2F;production&#x2F;uploads&#x2F;630b77f68b327c7b8b98c409&#x2F;ifzBOlPA7E4IKkysMpelC.png)


## Download model

Weights for this model are available in Safetensors format.

[Download](/h1t/TCD-SD21-base-LoRA/tree/main) them in the Files & versions tab.