It is interesting to see the floor as a bunch of pixels and looking very grainy, or "noisy" as the slide calls it. It makes sense that the render looks this way as at every pixel, we are estimating whether or not it gets blocked by the light by sampling random directions. The nature of every pixel being discrete ensures that pixel right next to each other may have drastically different light profiles due to the variability. Overall though, it will still represent the light profile that we want
ericlu28
The current algorithm of randomly assigning directions at each pixel, and seeing if that vector will hit the light source or not has clearly created the very noisy picture on the floor. This is connected to the fact that we are looking at every pixel point (integrating), rather than just integrating over the area of the light source-- this is the alternative algorithm, light area sampling, which makes more sense, and is similarily easy to reason through
saif-m17
We mentioned in lecture that the reason the picture looks so noisy is because there is a low chance our ray hits the light source. Why exactly is this the case, and I'm also a bit unclear about how this causes the noisy look.
It is interesting to see the floor as a bunch of pixels and looking very grainy, or "noisy" as the slide calls it. It makes sense that the render looks this way as at every pixel, we are estimating whether or not it gets blocked by the light by sampling random directions. The nature of every pixel being discrete ensures that pixel right next to each other may have drastically different light profiles due to the variability. Overall though, it will still represent the light profile that we want
The current algorithm of randomly assigning directions at each pixel, and seeing if that vector will hit the light source or not has clearly created the very noisy picture on the floor. This is connected to the fact that we are looking at every pixel point (integrating), rather than just integrating over the area of the light source-- this is the alternative algorithm, light area sampling, which makes more sense, and is similarily easy to reason through
We mentioned in lecture that the reason the picture looks so noisy is because there is a low chance our ray hits the light source. Why exactly is this the case, and I'm also a bit unclear about how this causes the noisy look.