fenghe12
fenghe12
same here
hello may i ask how did you use an image of a person as a input of smpl model and generate 3d representation？thanks！
hello！ i met similar problem. have you solved it？ bit accuray is nearly 100% during training.But when i use fine-tuned ldm decoder weight to generate images, i only get about...
sorry but i forgot to save training log，but i can share generation and decode code
import torch device = torch.device("cuda") from omegaconf import OmegaConf from diffusers import StableDiffusionPipeline from utils_model import load_model_from_config ldm_config = "./stablediffusion/configs/stable-diffusion/v2-inference.yaml" ldm_ckpt = "./stablediffusion/checkpoints-base/v2-1_512-nonema-pruned.ckpt" print(f'>>> Building LDM model with config {ldm_config}...
decoding code is exactly the same as decoding.ipynb
same here
bs>1估计没有显卡能跑得起来
> Download the latest version apex. DO NOT use apex 22.04 dev. Solution: `wget https://github.com/NVIDIA/apex.git` `cd apex` `pip install -v --disable-pip-version-check --no-cache-dir --no-build-isolation --config-settings "--build-option=--cpp_ext" --config-settings "--build-option=--cuda_ext" ./ ` 您好...