Question about rasterization function.

[InFistLee]

Thanks for the great work! I have two questions about the rasterization function.

1. What is the difference between "expected depth" and "accumulated depth", as illustrated in the description of argument render_mode.
2. The shape of render_colors is described as **render_colors**: The rendered colors. [C, width, height, X]. in rendering.py. However, what we actually get seems like [C, height, width, X]?

   render_colors, render_alphas, info = rasterization(
       means=means,
       quats=quats,
       scales=scales,
       opacities=opacities,
       colors=colors,
       Ks=Ks,  
       viewmats=w2c_mat,
       width=width,
       height=height,
       sh_degree=sh_degree,
       packed=False,
       absgrad=False,
       sparse_grad=False,
       rasterize_mode=rasterize_mode,
       render_mode='RGB',
   )
   print("width:", width, "\nheight:", height, "\nrender_colors:", render_colors.shape)
   # width: 1600 
   # height: 662 
   # render_colors: torch.Size([1, 662, 1600, 3]) 

[Atticuszz]

hi here the official explanation gsplat

[InFistLee]

hi here the official explanation gsplat

Thanks！That solves all my questions.