Some questions about paper

[BCL123456-BAL]

Hi friends, this is a very creative project. But for nerf, I'm just learning, so I may ask some stupid questions。

1. The generated results in the paper do not have the hair part, is it because of the limitation of Flame modeling that it cannot be generated?
2. When the mesh position changes, how do the sampling points in space change with the grid? Is it simply an affine transformation that follows the mesh?
3. During rendering in the new position, the direction of the light on the mesh is affinely changed by the light at the initial position of the mesh. Can I understand that when rendering a new position, the light is not the ray emitted by the optical center of the virtual camera, but the original light in the 3D space undergoes affine transformation, and finally falls in the image space to generate a new perspective picture. So there is no hair part because there is no hair mesh?

Best wish Chenlong

[WojtekZ4]

Hi, sorry for taking so long to answer,

1. Yes, FLAME only allows for modeling human heads, without parts like hair or glasses, as such our model is also unable to model those parts.
2. Basically yes. When calculating color and density of the point we find the closest triangle of the mesh to that point and transform it symmetrically to the transformation of the triangle. Note that along one ray different points will be differently transformed depending on what triangle of the mesh is the closest to them.
3. Yes that is correct. We simply take the way original model was illuminated and transform it. I will admit that this may lead to some weird looking effects, for example if original model was only illuminated from one side.

If you have any more questions feel free to ask them!