About the depth-guided ray sampling

[EchoTHChen]

dense_depth_priors_nerf/run_nerf.py

Line 579 in d541ebe

return forward_with_additonal_samples(z_vals, raw, z_vals_2, rays_o, rays_d, viewdirs, embedded_cam, network_fn, network_query_fn, raw_noise_std, pytest)

Why is network_fine not used here (network_fn is used) after the depth-guided sampling for final rendering? The network_fine isn't used totally in the rendering process.

[barbararoessle]

Hi, we do not use hierarchical sampling, so it is correct that there is just one network used. This is explained in the paper in section 3.2 in the paragraph on depth-guided sampling.

[EchoTHChen]

thanks!

…

---Original--- From: "Barbara ***@***.***> Date: Sun, Mar 12, 2023 18:11 PM To: ***@***.***>; Cc: ***@***.******@***.***>; Subject: Re: [barbararoessle/dense_depth_priors_nerf] About the depth-guidedray sampling (Issue #27) Hi, we do not use hierarchical sampling, so it is correct that there is just one network used. This is explained in the paper in section 3.2 in the paragraph on depth-guided sampling. — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID: ***@***.***>