how can i run demo on my own dataset without training?

[Sk-bishnoi]

Namespace(N_importance=0, N_rand=4096, N_samples=64, basedir='data/nerf_llff_data/', chunk=32768, config=None, datadir='./data/llff/fern', dataset_type='llff', expname='exp', factor=8, ft_path=None, half_res=False, i_embed=0, i_img=500, i_print=100, i_testset=50000, i_video=50000, i_weights=10000, lindisp=False, llffhold=8, lrate=0.0005, lrate_decay=250, multires=10, multires_views=4, netchunk=65536, netdepth=8, netdepth_fine=8, netwidth=256, netwidth_fine=256, no_batching=False, no_ndc=False, no_reload=False, perturb=1.0, precrop_frac=0.5, precrop_iters=0, random_seed=None, raw_noise_std=0.0, render_factor=0, render_only=False, render_test=False, shape='greek', spherify=False, testskip=8, use_viewdirs=False, white_bkgd=False)
2023-07-10 12:24:27.867780: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library 'libcuda.so.1'; dlerror: libcuda.so.1: cannot open shared object file: No such file or directory
2023-07-10 12:24:27.867809: E tensorflow/stream_executor/cuda/cuda_driver.cc:318] failed call to cuInit: UNKNOWN ERROR (303)
2023-07-10 12:24:27.867827: I tensorflow/stream_executor/cuda/cuda_diagnostics.cc:156] kernel driver does not appear to be running on this host (sureshkumar-HP-ProBook-640-G8-Notebook-PC): /proc/driver/nvidia/version does not exist
2023-07-10 12:24:27.868425: I tensorflow/core/platform/cpu_feature_guard.cc:142] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 AVX512F FMA
2023-07-10 12:24:27.898674: I tensorflow/core/platform/profile_utils/cpu_utils.cc:94] CPU Frequency: 2611200000 Hz
2023-07-10 12:24:27.899453: I tensorflow/compiler/xla/service/service.cc:168] XLA service 0x55a7dfc7fce0 initialized for platform Host (this does not guarantee that XLA will be used). Devices:
2023-07-10 12:24:27.899497: I tensorflow/compiler/xla/service/service.cc:176] StreamExecutor device (0): Host, Default Version
MODEL 63 0 <class 'int'> <class 'int'> False
(?, 63) (?, 63) (?, 0)
Found ckpts []
get rays
Killed

[NikitaPovelkin]

Hello,

You can't. NeRF needs to be trained on your data before you can generate new views from the model because NeRF " memorises " the given scene.

Best regards,
Nikita