| import tempfile | |
| import numpy as np | |
| import PIL.Image | |
| import torch | |
| import trimesh | |
| from diffusers import ShapEImg2ImgPipeline, ShapEPipeline | |
| from diffusers.utils import export_to_ply, export_to_gif | |
| from typing import Dict, List, Any | |
| class EndpointHandler(): | |
| def __init__(self): | |
| self.device = torch.device("cuda" if torch.cuda.is_available() else "cpu") | |
| self.pipe = ShapEPipeline.from_pretrained("openai/shap-e", torch_dtype=torch.float16) | |
| self.pipe.to(self.device) | |
| self.pipe_img = ShapEImg2ImgPipeline.from_pretrained("openai/shap-e-img2img", torch_dtype=torch.float16) | |
| self.pipe_img.to(self.device) | |
| def to_glb(self, ply_path: str) -> str: | |
| mesh = trimesh.load(ply_path) | |
| rot = trimesh.transformations.rotation_matrix(-np.pi / 2, [1, 0, 0]) | |
| mesh = mesh.apply_transform(rot) | |
| rot = trimesh.transformations.rotation_matrix(np.pi, [0, 1, 0]) | |
| mesh = mesh.apply_transform(rot) | |
| mesh_path = tempfile.NamedTemporaryFile(suffix=".glb", delete=False) | |
| mesh.export(mesh_path.name, file_type="glb") | |
| return mesh_path.name | |
| def run_text(self, prompt: str, seed: int = 0, guidance_scale: float = 15.0, num_steps: int = 64, output_type: str = "pil") -> str: | |
| generator = torch.Generator(device=self.device).manual_seed(seed) | |
| if output_type=="pil": | |
| images = self.pipe( | |
| prompt, | |
| num_images_per_prompt=4, | |
| generator=generator, | |
| guidance_scale=guidance_scale, | |
| num_inference_steps=num_steps, | |
| frame_size=64, | |
| ).images | |
| gif_path = tempfile.NamedTemporaryFile(suffix=".gif", delete=False, mode="w+b") | |
| gif_path2 = tempfile.NamedTemporaryFile(suffix=".gif", delete=False, mode="w+b") | |
| gif_path3 = tempfile.NamedTemporaryFile(suffix=".gif", delete=False, mode="w+b") | |
| gif_path4 = tempfile.NamedTemporaryFile(suffix=".gif", delete=False, mode="w+b") | |
| export_to_gif(images[0], gif_path.name) | |
| export_to_gif(images[1], gif_path2.name) | |
| export_to_gif(images[2], gif_path3.name) | |
| export_to_gif(images[3], gif_path4.name) | |
| return gif_path.name, gif_path2.name, gif_path3.name, gif_path4.name | |
| else: | |
| images = self.pipe( | |
| prompt, | |
| num_images_per_prompt=1, | |
| generator=generator, | |
| guidance_scale=guidance_scale, | |
| num_inference_steps=num_steps, | |
| frame_size=64, | |
| output_type=output_type | |
| ).images | |
| ply_path = tempfile.NamedTemporaryFile(suffix=".ply", delete=False, mode="w+b") | |
| export_to_ply(images[0], ply_path.name) | |
| return self.to_glb(ply_path.name) | |
| def __call__(self, prompt: str, seed: int = 0, guidance_scale: float = 15.0, num_steps: int = 64) -> str: | |
| generator = torch.Generator(device=self.device).manual_seed(seed) | |
| images = self.pipe( | |
| prompt, | |
| num_images_per_prompt=1, | |
| generator=generator, | |
| guidance_scale=guidance_scale, | |
| num_inference_steps=num_steps, | |
| frame_size=64, | |
| output_type="mesh" | |
| ).images | |
| ply_path = tempfile.NamedTemporaryFile(suffix=".ply", delete=False, mode="w+b") | |
| export_to_ply(images[0], ply_path.name) | |
| return self.to_glb(ply_path.name) |