RayTracing?

[nutball]

Oh, while i'm at it, i would like to ask a related question...
I have observed some phenomenon in the real world defying what i believe to know about physics and optics.
What i see hints that light travels along curvy paths, not straight lines.
Yes, i'm serious. Here's what i saw and still see:

I was out the house to smoke a cigarette. While i look at the cigarette 30cm away from my eye, i see neighbors roof in the background. The structure of the roof forms a regular grid, which is why i noticed the phenomenon.
Here is an illustration, ignoring the angle of the roof, which does not matter:

View attachment 9382
Left is the near cigarette, and right is the grid formed by the roof, maybe 20m behind.
The grid is regular. But near the cigarette, it's compressed.
It's like the light travels around the cig. obstacle. The light actually bends.
And due to that, i can actually see grid lines which should be occluded by the cig.

Try it yourself. It's not related to temperature of the cigarette. I have closed one eye as well. Light still bends around it. I can go out and do it again. It fucking bends.
The effect is limited to a small area around the obstacle, i would say few millimeters around the cig. But it's easy to see if you focus.
(Use a pencil or whatever instead. And move it slowly a little bit left and right. Then you should notice a kind of lens effect, making it easy to focus and concentrate on the compression.)

Why is this?
I know black holes can bend light. But the effect of a cig. should not be noticeable.
I lack any explanation.

Refraction. The refractive index of air depends on temperature and pressure. Cigarettes are hot, right?

See also: caustics. And turbulence.

You also seem to have re-invented parallax for some reason. I don't know why.

 

[JoeJ]

Refraction. The refractive index of air depends on temperature and pressure. Cigarettes are hot, right?

That's what i thought first too. But i can use a new and unlit cigarette - it still bends the same way.

I could even assume the cig. pushes away the air around it, creating some sort of lens from differing air pressure. But such compressed air would diffuse out quickly, all air pressure becomes equal, and the lens effect should be only temporary.
But it's not temporary. It's constant over time in an already converged uniform air pressure around it.

Maybe i should try to observe similar effects from different situations. Maybe something very specific conflicting with obvious assumptions causes confusion.

Oh, i just remember: It seemed the effect was noticeable only on the right side of the cig. And the sun light came from the left. But this makes even less sense and i don't trust this observation.

 

[JoeJ]

You also seem to have re-invented parallax for some reason. I don't know why.

I must admit i do not really know what parallax exactly means. You'd need to help me out... ;D

 

[homerdog]

Oh, while i'm at it, i would like to ask a related question...
I have observed some phenomenon in the real world defying what i believe to know about physics and optics.
What i see hints that light travels along curvy paths, not straight lines.
Yes, i'm serious. Here's what i saw and still see:

I was out the house to smoke a cigarette. While i look at the cigarette 30cm away from my eye, i see neighbors roof in the background. The structure of the roof forms a regular grid, which is why i noticed the phenomenon.
Here is an illustration, ignoring the angle of the roof, which does not matter:

View attachment 9382
Left is the near cigarette, and right is the grid formed by the roof, maybe 20m behind.
The grid is regular. But near the cigarette, it's compressed.
It's like the light travels around the cig. obstacle. The light actually bends.
And due to that, i can actually see grid lines which should be occluded by the cig.

Try it yourself. It's not related to temperature of the cigarette. I have closed one eye as well. Light still bends around it. I can go out and do it again. It fucking bends.
The effect is limited to a small area around the obstacle, i would say few millimeters around the cig. But it's easy to see if you focus.
(Use a pencil or whatever instead. And move it slowly a little bit left and right. Then you should notice a kind of lens effect, making it easy to focus and concentrate on the compression.)

Why is this?
I know black holes can bend light. But the effect of a cig. should not be noticeable.
I lack any explanation.

I assume this phenomenon goes away if you hold the cig out at arm's length. So that should give a hint as to why this is happening. It could be because your retina has surface area and is not single small point. Perhaps something about an object partially obscuring your retina confuses your brain. The effect would be completely invisible unless the occluder is very close to your face. Try closing one eye and see if you observe the same distortion around your nose.

On the other hand if this is an optical effect then you could perhaps recreate it with a camera.

Third option is that there is something other than tobacco in your smokes

Oh and 4th option is that the roof is reflecting off the cig at such an extreme angle and the distortion is a kind of mirage. Now that I think about it, this seems most likely.

 

[hughJ]

I assume that's just diffraction? Fresnel diffraction? If you look through a small aperture/pinhole or knife-edge you can see distortion at the edges, yeah...?

 

[JoeJ]

I assume this phenomenon goes away if you hold the cig out at arm's length. So that should give a hint as to why this is happening. It could be because your retina has surface area and is not single small point. Perhaps something about an object partially obscuring your retina confuses your brain. The effect would be completely invisible unless the occluder is very close to your face. Try closing one eye and see if you observe the same distortion around your nose.

That's it. I'm pretty sure. : )
The cig was closer then remembered, just tried it again. The farther away it gets, the lesser the distortion.
And the nose test feels very similar, just the nose is a bit too close so it blurs and darkens the background, the range of distortion may be much larger, but it's too subtle to tell.
The explanation of retina having area works for me. I can imagine the brain creates the distortion along its processing.

Third option is that there is something other than tobacco in your smokes

Reminds me on the talking disco ball. But i did not trust this observation either. : )

Thanks for all the inputs. I'm enlightened and happy.

 

[homerdog]

Actually I think hughj is correct, though why this happens is above my paygrade.

 

[hughJ]

I remember it being a huge pain in the ass to replicate the diffraction effect for sound propagation; you not only have to trace between the source and receiver, but also against any edges of occluding surfaces and then based on the angles you had a differing amount of frequency-dependent attenuation (low-pass filters applied to different frequency bands). Thankfully the effect is subtle enough for light propagation that I think RT/PT can ignore that wave-like interaction.

 

[JoeJ]

Thankfully the effect is subtle enough for light propagation that I think RT/PT can ignore that wave-like interaction.

But i always think that GI data structures could help with audio too, since it's a similar problem.
If we have volume or surface probes with depth, on could look up those probes instead raytracing the scene again.
Instead of walking on meshes to find edges forming a closest path between source and receiver, one could e.g. walk in image space of the probes, descending depth gradient.
Both should be much faster, so you could have a lot more paths if it helps.

 

[Cyan]

Intel wants to make Path Tracing usable on Integrated Graphics.

Intel Strives to Make Path Tracing Usable on Integrated GPUs

Is mainstream ray tracing around the corner?

www.tomshardware.com