Open shudct opened 3 days ago
First of all, thx for release the code!
when I test ,I notice the total time for the process is quite long, especially in the following function.
scripts/mutiview_inference.py: geo_reconstruct
img_list = [front_pil] + run_sr_fast(refined_rgbs[1:]) -------> 40s
rm_normals = predict_normals([img.resize((512, 512), resample=Image.LANCZOS) for img in img_list], guidance_scale=1.5) -------> 90s
vertices, faces = run_mesh_refine(vertices, faces, rm_normals, steps=100, start_edge_len=0.02, end_edge_len=0.005, decay=0.99, update_normal_interval=20, update_warmup=5, return_mesh=False, process_inputs=False, process_outputs=False) ------> 90s
I wonder if this is normal and is there any way to speed it up? thx a lot.
First of all, thx for release the code!
when I test ,I notice the total time for the process is quite long, especially in the following function.
scripts/mutiview_inference.py: geo_reconstruct
img_list = [front_pil] + run_sr_fast(refined_rgbs[1:]) -------> 40srm_normals = predict_normals([img.resize((512, 512), resample=Image.LANCZOS) for img in img_list], guidance_scale=1.5) -------> 90svertices, faces = run_mesh_refine(vertices, faces, rm_normals, steps=100, start_edge_len=0.02, end_edge_len=0.005, decay=0.99, update_normal_interval=20, update_warmup=5, return_mesh=False, process_inputs=False, process_outputs=False) ------> 90sI wonder if this is normal and is there any way to speed it up? thx a lot.