If we visualize the light as rays, it is obvious to see that why camera with lenses is so much better than a pinhole camera. The pinhole camera as shown in this slide clearly illustrates where the rays are from, which is essentially just "one" ray. However, for the lensed camera, a point in the imaging plane can have rays from much larger volume, which makes it better for imaging at a faster shutter speed given that it has more light.
stephanie-fu
This is one of my favorite references to a pinhole camera: http://people.csail.mit.edu/torralba/research/accidentalcameras/
The need for a camera lens as the number of incoming light rays increases is also fascinatingly similar to this theory for how eyes have evolved, where some organisms increased aperture without a lens and ended up with very blurry vision: https://royalsocietypublishing.org/doi/10.1098/rspb.1994.0048
muuncakez
based on the arrows here, I was wondering if the image of a pinhole camera end up inverse? So I did a lil internet sleuth and it does! Apparently almost all cameras do until you manually flip it, but most do this as part of processing the image. Human eyes also see the img upside down (our eyes are like pinholes as well) but our brains flip the image as part of "processing".
However based on where we look through the hole we will see different parts of the object, this is because light travels in a small wave that nearly create straight lines of travel. So if I laid flat and looked up through a pinhole, I see the top of the object on the otherwise, stand tall to see the bottom of the object, left to see right, right to see left.
the forum gives a great examples/explanations:
I was also wondering if the pinhole camera is subject to the same results found from a double slit setup, do the light waves also interfere and interact with each other in a pinhole as well? or is the flipped img the pinhole version of the striped pattern from the double slit experiment?
danielhsu021202
How accurately would a pinhole camera sample radiance as compared to a lens from, say, a DSLR camera?
yykkcc
@muuncakez In my opinion, pinhole camera has only one opening, which means there's no opportunity for two sets of waves to interfere with each other as they do in the double slit experiment. The pinhole ensures that light from each part of the scene is projected to a specific part of the image plane without interacting with light from other parts of the scene.
s3kim2018
Connecting this slide with a previous lecture on perspective and parallel projection, we know that during perspective projection (pin hole camera model) objects further away form the scene are proportionally smaller than objects closer to the scene. But the objects further away does not have blurry color nor the scene looks less intense. Does radiance not decrease with distance?
stephanie-fu
I think Prof. Ng's comment about assuming a vacuum where radiance is constant along a ray might be relevant here - unless we're in a foggy room where there are volumetric interferences, it's a good assumption for light traveling in normal air.
If we visualize the light as rays, it is obvious to see that why camera with lenses is so much better than a pinhole camera. The pinhole camera as shown in this slide clearly illustrates where the rays are from, which is essentially just "one" ray. However, for the lensed camera, a point in the imaging plane can have rays from much larger volume, which makes it better for imaging at a faster shutter speed given that it has more light.
This is one of my favorite references to a pinhole camera: http://people.csail.mit.edu/torralba/research/accidentalcameras/
The need for a camera lens as the number of incoming light rays increases is also fascinatingly similar to this theory for how eyes have evolved, where some organisms increased aperture without a lens and ended up with very blurry vision: https://royalsocietypublishing.org/doi/10.1098/rspb.1994.0048
based on the arrows here, I was wondering if the image of a pinhole camera end up inverse? So I did a lil internet sleuth and it does! Apparently almost all cameras do until you manually flip it, but most do this as part of processing the image. Human eyes also see the img upside down (our eyes are like pinholes as well) but our brains flip the image as part of "processing".
However based on where we look through the hole we will see different parts of the object, this is because light travels in a small wave that nearly create straight lines of travel. So if I laid flat and looked up through a pinhole, I see the top of the object on the otherwise, stand tall to see the bottom of the object, left to see right, right to see left. the forum gives a great examples/explanations:
https://physics.stackexchange.com/questions/628424/why-do-images-not-appear-inverted-when-looking-directly-through-a-pinhole-camera#:~:text=Pinholes%20don't%20flip%20the%20image%2C%20but%20filter%20light%20rays.&text=The%20image%20is%20mirrored%20about,was%20just%20a%20%22hole%22.
I was also wondering if the pinhole camera is subject to the same results found from a double slit setup, do the light waves also interfere and interact with each other in a pinhole as well? or is the flipped img the pinhole version of the striped pattern from the double slit experiment?
How accurately would a pinhole camera sample radiance as compared to a lens from, say, a DSLR camera?
@muuncakez In my opinion, pinhole camera has only one opening, which means there's no opportunity for two sets of waves to interfere with each other as they do in the double slit experiment. The pinhole ensures that light from each part of the scene is projected to a specific part of the image plane without interacting with light from other parts of the scene.
Connecting this slide with a previous lecture on perspective and parallel projection, we know that during perspective projection (pin hole camera model) objects further away form the scene are proportionally smaller than objects closer to the scene. But the objects further away does not have blurry color nor the scene looks less intense. Does radiance not decrease with distance?
I think Prof. Ng's comment about assuming a vacuum where radiance is constant along a ray might be relevant here - unless we're in a foggy room where there are volumetric interferences, it's a good assumption for light traveling in normal air.