This is so cool, I wonder how they keep the clothes attached to the figure
draynr
Seeing clothes being rendered so realistically is incredible, but how I wonder how far the boundaries of cloth simulation and hair rendering are being pushed onto technologies like AR. While these scenes look incredible, I think there's possibility for a higher degree of realism, especially for hair, that could almost replace the physical presence of both.
RishSharma7
In response to draynr, I also find the clothing rendering to be pretty cool, but looking at how this current rendering looks, and even how some of the best possible renderings look online, I still think we're quite a while away from being able to replace the physical presence of clothes. For hair, I think we're even further. Seeing the intricate textures and details of all different types of hair, and seeing how movement affects those details and textures, makes me think that we have a very long way to go before neither hair nor clothes are needed for these renderings. But maybe I'm misinformed.
sebzhao
From my understanding, if there's a physical object in reality, then using camera views you can construct 3d objects with pretty good quality, but it isn't necessarily super fast and as optimizable, though a lot of work is being done to make it faster.
yykkcc
I'm wondering if we can put sensors on the cloth while doing the motion capture to collect the movements of the cloth. That sounds feasible to me, but people usually don't do that. Maybe the cost of doing that is much higher than the cloth simulation.
508312
I am also interested in how to keep clothes attached to a person. In case of a t-shirt it is trivial I believe. However, the pants are a little bit more tricky: In real life they push onto a person, but simulating person deformation plus all the forces is probably too much, so do you just stick it to the vertices of a person?
SadhikaA
I was wondering how the motion capture worked for this. I assume that they have certain markers on the person as they moved and then transferred It over to the model. As for the clothes, I think if each part of the cloth mapped to a specific part of the original human model then this would prevent any incorrect folding.
theflyingpie
This makes me wonder how modeling different types of fabrics works -- the t-shirt looks more soft and flowy, whereas the pants and perhaps vest have a stiffer feel. Does this have to do with the granularity of the meshes or some other properties?
muuncakez
After completing HW4... I really like the idea of springs to simulate cloth. for example, changing the spring constant or density to give the cloth different fabric looks. In HW4, using high constant (k_s) makes for a more stiff looking material like a rug that resisted folding creating a "heavier" and "denser" look bc the folds were much bigger due to the increased stiffness.
other cloth rendering techniques include using coils instead of springs. or taking it a step further and simulating particle energies (like two particles from the cloth should repel each other as to not clip through itself)
with spring, keeping the pants up I think requires the pants to be "pinned" around the hips of the model. This way the pant fabric stills "falls" around the legs but is held in place s.t. gravity doesn't just pull the pants down.
This is so cool, I wonder how they keep the clothes attached to the figure
Seeing clothes being rendered so realistically is incredible, but how I wonder how far the boundaries of cloth simulation and hair rendering are being pushed onto technologies like AR. While these scenes look incredible, I think there's possibility for a higher degree of realism, especially for hair, that could almost replace the physical presence of both.
In response to draynr, I also find the clothing rendering to be pretty cool, but looking at how this current rendering looks, and even how some of the best possible renderings look online, I still think we're quite a while away from being able to replace the physical presence of clothes. For hair, I think we're even further. Seeing the intricate textures and details of all different types of hair, and seeing how movement affects those details and textures, makes me think that we have a very long way to go before neither hair nor clothes are needed for these renderings. But maybe I'm misinformed.
From my understanding, if there's a physical object in reality, then using camera views you can construct 3d objects with pretty good quality, but it isn't necessarily super fast and as optimizable, though a lot of work is being done to make it faster.
I'm wondering if we can put sensors on the cloth while doing the motion capture to collect the movements of the cloth. That sounds feasible to me, but people usually don't do that. Maybe the cost of doing that is much higher than the cloth simulation.
I am also interested in how to keep clothes attached to a person. In case of a t-shirt it is trivial I believe. However, the pants are a little bit more tricky: In real life they push onto a person, but simulating person deformation plus all the forces is probably too much, so do you just stick it to the vertices of a person?
I was wondering how the motion capture worked for this. I assume that they have certain markers on the person as they moved and then transferred It over to the model. As for the clothes, I think if each part of the cloth mapped to a specific part of the original human model then this would prevent any incorrect folding.
This makes me wonder how modeling different types of fabrics works -- the t-shirt looks more soft and flowy, whereas the pants and perhaps vest have a stiffer feel. Does this have to do with the granularity of the meshes or some other properties?
After completing HW4... I really like the idea of springs to simulate cloth. for example, changing the spring constant or density to give the cloth different fabric looks. In HW4, using high constant (k_s) makes for a more stiff looking material like a rug that resisted folding creating a "heavier" and "denser" look bc the folds were much bigger due to the increased stiffness.
other cloth rendering techniques include using coils instead of springs. or taking it a step further and simulating particle energies (like two particles from the cloth should repel each other as to not clip through itself)
with spring, keeping the pants up I think requires the pants to be "pinned" around the hips of the model. This way the pant fabric stills "falls" around the legs but is held in place s.t. gravity doesn't just pull the pants down.