metadata
license: mit
tags:
- pytorch
- diffusers
- unconditional-image-generation
- diffusion-models-class
library_name: diffusers
This model is a diffusion model for unconditional image generation of uwall velocity field in a turbulent channel flow with Retau = 1000. This model is trained using just almost 1000 fields which are very few since in the most of case DDPM are trained with thousands of images.
Usage
from diffusers import DDPMPipeline
import torch
import torchvision
from torchvision import transforms
import numpy as np
from matplotlib import pyplot as plt
from huggingface_hub import hf_hub_download
import joblib
pipeline = DDPMPipeline.from_pretrained('RobbyDalmonts/DDPM_uwall_morefields')
image = pipeline().images[0]
to_tensor = transforms.ToTensor()
tensor_output_normalized = to_tensor(image)
tensor_output_normalized = tensor_output_normalized *2 -1
fig,ax = plt.subplots()
ax.imshow(tensor_output_normalized[0], cmap ='coolwarm')
ax.set_title('normalized sample')
plt.show()
plt.close()
scaler_path = hf_hub_download(repo_id='RobbyDalmonts/DDPM_uwall_morefields', filename="scaler.pkl", repo_type='model')
scaler = joblib.load(scaler_path)
tensor_output = scaler.inverse_transform(tensor_output_normalized.squeeze(0).numpy().reshape(-1,1)).reshape(128,64)
fig,ax = plt.subplots()
ax.imshow(tensor_output, cmap='coolwarm')
ax.set_title('sample')
plt.show()
plt.close()