How is distortions from 3D to 2D mappings and vice-versa handled in texture mapping.
jayc809
This reminds me of how the edge areas of a panorama are always more stretched out compared to the center (e.g. Google street view). However, it seems like although certain parts of the texture image are warped, these areas are spread out across the image and kind of self-corrects the overall image towards a rectangular shape. In other words, there does not seem to have the problem where with increasing complexity of the 3D model, the 2D texture reaches a point where the local scale of triangles vary so much that we end up either losing detail or have inefficient, large stretched-out components.
brianqch
Is it possible to "lose" data through transformations in one space? I would assume no because the triangles are essentially points that are mapped directly to another point in a different space. Therefore, any changes are reflected across all spaces. Is that right?
omijimo
how are the triangles in texture images properly created to represent a surface? or does the process of converting a texture image onto the world space automatically do that?
yauwilliam69
@brianqch: I think you are right. I believe there exists a bijective function that maps each point on the texture image onto the 3D surface, and vice versa. Otherwise, it would be undefined for one point on the 3D surface if there exists none, or more than one mapping from the texture image.
stang085
Does making a texture image work for any kind of shape? Does the image also warp a lot if the shape is very irregular? where do you start when making texture images?
DestinyLuong
Another thing that is often talked about in 3D modeling is shadow rendering. If we wanted to add a shadow to the model, with consideration to where the light source and other planes are how would we do that? Are shadows drawn on surfaces created in a similar way?
saif-m17
Along the same lines of questioning as @brianqch and @yauwilliam69, one thing talked about in lecture in the case of a texture map onto a sphere (with the globe example) was that you might have multiple points on the texture map go to the same point on the image itself (thinking of the north pole). I wonder then if it's always possible to recover the texture map from the image, since the transformation doesn't seem invertible in that case.
RishSharma7
I am a bit confused when it comes to how texture images start to form for complex 3D polygons and other highly irregular figures. I understand that you have to create corresponding points between the 3D figure and the 2D texture image, but how exactly would you go about doing that if you are dealing with a shape that nobody has ever seen before? Do you just have to create a simple background pattern behind the 3D image, like some sort of grid, and work backwards from there? Or maybe I have the process confused; would love some clarification! Thanks.
gfjvgufkt
I'm wondering how to decide these triangles. Do we need the texture map to be exactly a square? And how can we implement that? Is that also a kind of sampling to this structure? But it seems like the color of each triangle is not pure. How do we decide the color of each triangle?
AlsonC
This is really interesting, never knew how texture could actually be represented on the surface, thought it could always be represented by a difference in colors
How is distortions from 3D to 2D mappings and vice-versa handled in texture mapping.
This reminds me of how the edge areas of a panorama are always more stretched out compared to the center (e.g. Google street view). However, it seems like although certain parts of the texture image are warped, these areas are spread out across the image and kind of self-corrects the overall image towards a rectangular shape. In other words, there does not seem to have the problem where with increasing complexity of the 3D model, the 2D texture reaches a point where the local scale of triangles vary so much that we end up either losing detail or have inefficient, large stretched-out components.
Is it possible to "lose" data through transformations in one space? I would assume no because the triangles are essentially points that are mapped directly to another point in a different space. Therefore, any changes are reflected across all spaces. Is that right?
how are the triangles in texture images properly created to represent a surface? or does the process of converting a texture image onto the world space automatically do that?
@brianqch: I think you are right. I believe there exists a bijective function that maps each point on the texture image onto the 3D surface, and vice versa. Otherwise, it would be undefined for one point on the 3D surface if there exists none, or more than one mapping from the texture image.
Does making a texture image work for any kind of shape? Does the image also warp a lot if the shape is very irregular? where do you start when making texture images?
Another thing that is often talked about in 3D modeling is shadow rendering. If we wanted to add a shadow to the model, with consideration to where the light source and other planes are how would we do that? Are shadows drawn on surfaces created in a similar way?
Along the same lines of questioning as @brianqch and @yauwilliam69, one thing talked about in lecture in the case of a texture map onto a sphere (with the globe example) was that you might have multiple points on the texture map go to the same point on the image itself (thinking of the north pole). I wonder then if it's always possible to recover the texture map from the image, since the transformation doesn't seem invertible in that case.
I am a bit confused when it comes to how texture images start to form for complex 3D polygons and other highly irregular figures. I understand that you have to create corresponding points between the 3D figure and the 2D texture image, but how exactly would you go about doing that if you are dealing with a shape that nobody has ever seen before? Do you just have to create a simple background pattern behind the 3D image, like some sort of grid, and work backwards from there? Or maybe I have the process confused; would love some clarification! Thanks.
I'm wondering how to decide these triangles. Do we need the texture map to be exactly a square? And how can we implement that? Is that also a kind of sampling to this structure? But it seems like the color of each triangle is not pure. How do we decide the color of each triangle?
This is really interesting, never knew how texture could actually be represented on the surface, thought it could always be represented by a difference in colors