Spaces:
Running
on
Zero
Running
on
Zero
try: | |
import spaces | |
except ImportError: | |
# Define a dummy decorator if spaces is not available | |
def GPU(func): | |
return func | |
spaces = type('spaces', (), {'GPU': GPU}) | |
import os | |
import torch | |
import argparse | |
from typing import * | |
from diffusers import StableDiffusionPipeline | |
from collections import deque | |
from triplaneturbo_executable.utils.mesh_exporter import export_obj | |
from triplaneturbo_executable import TriplaneTurboTextTo3DPipeline, TriplaneTurboTextTo3DPipelineConfig | |
# Initialize configuration and parameters | |
prompt = "a beautiful girl" | |
output_dir = "output" | |
adapter_name_or_path = "pretrained/triplane_turbo_sd_v1.pth" | |
num_results_per_prompt = 1 | |
seed = 42 | |
device = "cuda" | |
max_obj_files = 100 | |
# download pretrained models if not exist | |
if not os.path.exists(adapter_name_or_path): | |
print(f"Downloading pretrained models from huggingface") | |
os.system( | |
f"huggingface-cli download --resume-download ZhiyuanthePony/TriplaneTurbo \ | |
--include \"triplane_turbo_sd_v1.pth\" \ | |
--local-dir ./pretrained \ | |
--local-dir-use-symlinks False" | |
) | |
# Initialize the TriplaneTurbo pipeline | |
triplane_turbo_pipeline = TriplaneTurboTextTo3DPipeline.from_pretrained(adapter_name_or_path) | |
triplane_turbo_pipeline.to(device) | |
def generate_3d_model(prompt, num_results_per_prompt=1, seed=42, device="cuda"): | |
""" | |
Generate 3D models using TriplaneTurbo pipeline. | |
Args: | |
prompt (str): Text prompt for the 3D model | |
num_results_per_prompt (int): Number of results to generate | |
seed (int): Random seed for generation | |
device (str): Device to use for computation | |
Returns: | |
dict: Output from the pipeline | |
""" | |
output = triplane_turbo_pipeline( | |
prompt=prompt, | |
num_results_per_prompt=num_results_per_prompt, | |
generator=torch.Generator(device=device).manual_seed(seed), | |
device=device, | |
) | |
# Initialize a deque with maximum length of 100 to store obj file paths | |
obj_file_queue = deque(maxlen=max_obj_files) | |
# Save mesh | |
os.makedirs(output_dir, exist_ok=True) | |
for i, mesh in enumerate(output["mesh"]): | |
vertices = mesh.v_pos | |
# 1. First rotate -90 degrees around X-axis to make the model face up | |
vertices = torch.stack([ | |
vertices[:, 0], # x remains unchanged | |
vertices[:, 2], # y = z | |
-vertices[:, 1] # z = -y | |
], dim=1) | |
# 2. Then rotate 90 degrees around Y-axis to make the model face the observer | |
vertices = torch.stack([ | |
-vertices[:, 2], # x = -z | |
vertices[:, 1], # y remains unchanged | |
vertices[:, 0] # z = x | |
], dim=1) | |
mesh.v_pos = vertices | |
# If mesh has normals, they need to be rotated in the same way | |
if mesh.v_nrm is not None: | |
normals = mesh.v_nrm | |
# 1. Rotate -90 degrees around X-axis | |
normals = torch.stack([ | |
normals[:, 0], | |
normals[:, 2], | |
-normals[:, 1] | |
], dim=1) | |
# 2. Rotate 90 degrees around Y-axis | |
normals = torch.stack([ | |
-normals[:, 2], | |
normals[:, 1], | |
normals[:, 0] | |
], dim=1) | |
mesh._v_nrm = normals | |
# Save obj file and add its path to the queue | |
name = f"{prompt.replace(' ', '_')}_{seed}_{i}" | |
save_paths = export_obj(mesh, f"{output_dir}/{name}.obj") | |
obj_file_queue.append(save_paths[0]) | |
# If an old file needs to be removed (queue is at max length) | |
# and the file exists, delete it | |
if len(obj_file_queue) == max_obj_files and os.path.exists(obj_file_queue[0]): | |
old_file = obj_file_queue[0] | |
try: | |
os.remove(old_file) | |
except OSError as e: | |
print(f"Error deleting file {old_file}: {e}") | |
# Run the pipeline | |
output = generate_3d_model( | |
prompt=prompt, | |
num_results_per_prompt=num_results_per_prompt, | |
seed=seed, | |
device=device | |
) | |