Was the audio speed variation intentional throughout? It is amazing how much more efficient Fourier filtering is compared to kernel real space filtering.
I think it's a combination of me talking faster plus editing out some parts that make the audio speed seem different. It is pretty amazing that Fourier transformations speed up filtering!
I didn't learn that when i learned about Fourier at university. Mindblowing to see it used for images. Keep up the good work, I'm becoming a bigger fan each new vid ❤
Thank you! It is amazing what what Fourier transforms can do. I first found out about the orbits thing from a Mathologer video and my mind was blown. It was too cool not to use when introducing deconvolution.
Thanks, Doug! It's a complicated topic, and I'm not sure I explained it as well as I could have. But I like my spaceship too! It makes it easier to breathe in space :)
It’s really excellent work! I'm a fan! I'm waiting for the sequel with the associated deconvolution! I found an article which talks about lucky fourier, a hybrid technique to lucky imaging which would allow more image to be kept to improve the SNR. Is this your final idea?
Thank you!! I haven't actually heard of lucky fourier. I'll have to check it out :) I've been thinking a lot about this for my new deconvolution tool. And possibly a new noise reduction tool too...
you're welcome :) And thank you! Glad you got it! The circles helped me wrap my head around the Fourier transform when I first heard about "high" and "low" frequencies in images. It's not perfect of course (there are no pictures of circles in Fourier space). But I thought it'd be interesting!
Damn, I went through a physics degree, but I never fully grasped the usefulness of Fourier transforms until now! It reminds me of how we transmit radio signals, since they are basically high frequencies piggybacking on low frequencies signal, so I assume you _could_ use a Fourier transform to decode a radio signal, but you obviously don't _have_ to..
Yeah! It's so crazy how Fourier pops up seemingly everywhere. I was just watching a video by Fermilab where they talk about Fourier transforms and the Heisenberg Uncertainty Principle.
duuuude that analogy was brilliant!!! You have such good way to mix math and astro, congratz, add more code to it too (as Software engineer in computer vision myself I love to dill into this hahaha)
Thanks!! Maybe I'll start a new series where I just go over the code. Or maybe it should be a new YT channel, because the YT algorithm can get confused when a channel's topics are too diverse.
@@deepskydetail I understand, my suggestion was just to maybe pseudo code, this can be very helpful to engineers like me understand pretty fast the behind the scenes. You create a lot of smart applications and code, it is in the core of the channel 😁 keep the great work man and get yourself an Instagram to talk with the new digital friends hahaha
It's a hard concept, and I didn't explain it very well. I guess maybe it would be better to not think of the circles as analogies. Rather, they are actual representations of the Fourier transformation itself. Maybe this will help: The circles are the Fourier transform applied to a complex wave. Instead of going in a straight line, like at 8:29, the wave travels in a loop. The circles are then the individual sine/cosine waves that make up the looped wave. Think of the function in 8:29 as wrapping around itself. What's nice, is that you can make it look like any outline that you want, from a square to an outline of a galaxy or whatever. Fourier transforms of real images though are just more complicated, and you can't visualize them like the examples. But the same ideas apply: high frequencies = noise and detail. Low frequencies = general structure.
Who needs Brilliant when we have Deep Sky Detail. 😁
Lol! Thanks, but I might get cocky if you say that . Also, Brilliant can be good for practicing what you learn :)
Was the audio speed variation intentional throughout?
It is amazing how much more efficient Fourier filtering is compared to kernel real space filtering.
I think it's a combination of me talking faster plus editing out some parts that make the audio speed seem different.
It is pretty amazing that Fourier transformations speed up filtering!
I didn't learn that when i learned about Fourier at university. Mindblowing to see it used for images. Keep up the good work, I'm becoming a bigger fan each new vid ❤
Thank you! It is amazing what what Fourier transforms can do. I first found out about the orbits thing from a Mathologer video and my mind was blown. It was too cool not to use when introducing deconvolution.
Whew Dr Detail, I’m experiencing 🤯 I might need a brilliant subscription to wrap my head around this!!
Cool video Mark, I’m digging your spaceship!!
Thanks, Doug! It's a complicated topic, and I'm not sure I explained it as well as I could have.
But I like my spaceship too! It makes it easier to breathe in space :)
@@deepskydetail 😂 You2 apparently does not have that requirement. I always wondered if stick beings required oxygen…
@@AstroAF Yeah, apparently, they can breathe anywhere lol!
It’s really excellent work! I'm a fan! I'm waiting for the sequel with the associated deconvolution! I found an article which talks about lucky fourier, a hybrid technique to lucky imaging which would allow more image to be kept to improve the SNR. Is this your final idea?
Thank you!! I haven't actually heard of lucky fourier. I'll have to check it out :) I've been thinking a lot about this for my new deconvolution tool. And possibly a new noise reduction tool too...
When you talk about circles... you mean wavelets?... I ophthalmologist and have no Idea what you mean with circles.
Excellent videos
All About high and low frequency understood... thank you.
you're welcome :) And thank you! Glad you got it! The circles helped me wrap my head around the Fourier transform when I first heard about "high" and "low" frequencies in images. It's not perfect of course (there are no pictures of circles in Fourier space). But I thought it'd be interesting!
Damn, I went through a physics degree, but I never fully grasped the usefulness of Fourier transforms until now!
It reminds me of how we transmit radio signals, since they are basically high frequencies piggybacking on low frequencies signal, so I assume you _could_ use a Fourier transform to decode a radio signal, but you obviously don't _have_ to..
Yeah! It's so crazy how Fourier pops up seemingly everywhere. I was just watching a video by Fermilab where they talk about Fourier transforms and the Heisenberg Uncertainty Principle.
duuuude that analogy was brilliant!!! You have such good way to mix math and astro, congratz, add more code to it too (as Software engineer in computer vision myself I love to dill into this hahaha)
Thanks!! Maybe I'll start a new series where I just go over the code. Or maybe it should be a new YT channel, because the YT algorithm can get confused when a channel's topics are too diverse.
@@deepskydetail I understand, my suggestion was just to maybe pseudo code, this can be very helpful to engineers like me understand pretty fast the behind the scenes. You create a lot of smart applications and code, it is in the core of the channel 😁 keep the great work man and get yourself an Instagram to talk with the new digital friends hahaha
That's a good idea :)
The analogies were too highly abstracted for me to understand.
It's a hard concept, and I didn't explain it very well. I guess maybe it would be better to not think of the circles as analogies. Rather, they are actual representations of the Fourier transformation itself. Maybe this will help:
The circles are the Fourier transform applied to a complex wave. Instead of going in a straight line, like at 8:29, the wave travels in a loop. The circles are then the individual sine/cosine waves that make up the looped wave. Think of the function in 8:29 as wrapping around itself. What's nice, is that you can make it look like any outline that you want, from a square to an outline of a galaxy or whatever.
Fourier transforms of real images though are just more complicated, and you can't visualize them like the examples. But the same ideas apply: high frequencies = noise and detail. Low frequencies = general structure.