When I create those, I will likely put them in an optics or electromagnetics course. This particularly lecture is just an example for my Computational class about uses for the FFT. Here are links to my electromagnetics courses incase you are interested: empossible.net/academics/emp3302/ empossible.net/academics/emp5337/ empossible.net/academics/emp6303/ emlab.utep.edu/eeIntroOpticsPhotonics.htm
Hi, I really love the way you present the physics, everything is clear. It helped me a lot during my phd studies thank you and still helps me. Thank you. I performed the Gerchberg-saxton algorithm on Matlab, however the result I obtained is hard to fabricate. It means that for tow adjacent pixels the phase values can differ a lot, in you figures the phase variation in the near field seems smoother than what I obtained. Do you have any trick about that ?
@@empossible1577 Hello, thank you for sharing this informative guide on the design of kinoforms, it is by far the best step-by-step tutorial I have seen out of all the books and websites on the subject. Thank you very much for sharing. I have made a matlab implementation of kinoform generation according to this tutorial, but ran into a problem regarding the diffraction pattern generated by the computed kinoform: The final diffraction pattern of the kinoform if achieved by using the phase from step 7 (at 3:42), that I understand. As I understand, You use something like this to obtain the output diffraction pattern in step 8: OUT_CEM = abs( AMP_STEP7 .* exp( 1i * PHASE_STEP7 ) ); imagesc( OUT_CEM ) However, when I make the calculated kinoform (the phase pattern from step 7) and shine a light through it instead of one pickaxe at the center of the field (like amplitude of step 8) I get two pickaxes overlaid on each other. I have also noticed that the two pickaxes share a relation - one of the pickaxes seems to be an inverted image of the other (inverted through the center of the overall image (step 8)). If I put the input pattern (pickaxe) in the top-left corner of the input image (amplitude of step 1 at 3:21) I get two pickaxes in the final diffraction pattern (step 8) - one where the input pickaxe was (as expected from amplitude of step 8), and one in the opposing corner (the second pickaxe is inverted - this one should not exist at all). The overall image I get is similar to this (the right part of the image in the link): www.sciencedirect.com/science/article/pii/S1665642315000747#fig0035 I have determined that the diffraction pattern I obtain experimentally can be simulated in matlab by this command: OUT_EXPERIMENT = real( fftshift( fft2( PHASE_STEP7 ) ) ) imagesc( OUT_EXPERIMENT )%
@@ezyxas Thank you! I suspect you missing an fftshift() of the image in the beginning. I have revised the notes to include some MATLAB code. I have also updated the video. I recommend accessing this through the course website since I will likely archive this video. Here is the course website: empossible.net/academics/emp4301_5301/ This particular lecture is in Topic 9. Hope this helps!!
@@empossible1577 Thank you for the quick response. After applying the command you suggested and reviewing the renewed lecture I seemingly to have solved the problem. However, if I try to use some other tool to get the images from produced kinoforms (i.e. bigwww.epfl.ch/demo/ip/demos/FFT/ ) the second image problem still persists. I will try to see what happens experimentally - perhaps the second image is only a software thing. Once again, than You.
Learned something new, great explanation :D
I love your vedio very much.
Could you introduce some optimization methods such as adjoint variable method for device inverse desgin?
When I create those, I will likely put them in an optics or electromagnetics course. This particularly lecture is just an example for my Computational class about uses for the FFT.
Here are links to my electromagnetics courses incase you are interested:
empossible.net/academics/emp3302/
empossible.net/academics/emp5337/
empossible.net/academics/emp6303/
emlab.utep.edu/eeIntroOpticsPhotonics.htm
@@empossible1577 thank you very much, I will study them.
Hi, I really love the way you present the physics, everything is clear. It helped me a lot during my phd studies thank you and still helps me. Thank you.
I performed the Gerchberg-saxton algorithm on Matlab, however the result I obtained is hard to fabricate. It means that for tow adjacent pixels the phase values can differ a lot, in you figures the phase variation in the near field seems smoother than what I obtained. Do you have any trick about that ?
Hmmm...No tricks that I know of. How many points are you using on your grid? Consider padding your final IFFT with zeros.
@@empossible1577
Hello, thank you for sharing this informative guide on the design of kinoforms, it is by far the best step-by-step tutorial I have seen out of all the books and websites on the subject. Thank you very much for sharing.
I have made a matlab implementation of kinoform generation according to this tutorial, but ran into a problem regarding the diffraction pattern generated by the computed kinoform:
The final diffraction pattern of the kinoform if achieved by using the phase from step 7 (at 3:42), that I understand.
As I understand, You use something like this to obtain the output diffraction pattern in step 8:
OUT_CEM = abs( AMP_STEP7 .* exp( 1i * PHASE_STEP7 ) );
imagesc( OUT_CEM )
However, when I make the calculated kinoform (the phase pattern from step 7) and shine a light through it instead of one pickaxe at the center of the field (like amplitude of step 8) I get two pickaxes overlaid on each other.
I have also noticed that the two pickaxes share a relation - one of the pickaxes seems to be an inverted image of the other (inverted through the center of the overall image (step 8)).
If I put the input pattern (pickaxe) in the top-left corner of the input image (amplitude of step 1 at 3:21) I get two pickaxes in the final diffraction pattern (step 8) - one where the input pickaxe was (as expected from amplitude of step 8), and one in the opposing corner (the second pickaxe is inverted - this one should not exist at all).
The overall image I get is similar to this (the right part of the image in the link): www.sciencedirect.com/science/article/pii/S1665642315000747#fig0035
I have determined that the diffraction pattern I obtain experimentally can be simulated in matlab by this command:
OUT_EXPERIMENT = real( fftshift( fft2( PHASE_STEP7 ) ) )
imagesc( OUT_EXPERIMENT )%
@@ezyxas Thank you! I suspect you missing an fftshift() of the image in the beginning. I have revised the notes to include some MATLAB code. I have also updated the video. I recommend accessing this through the course website since I will likely archive this video. Here is the course website:
empossible.net/academics/emp4301_5301/
This particular lecture is in Topic 9. Hope this helps!!
@@empossible1577 Thank you for the quick response. After applying the command you suggested and reviewing the renewed lecture I seemingly to have solved the problem. However, if I try to use some other tool to get the images from produced kinoforms (i.e. bigwww.epfl.ch/demo/ip/demos/FFT/ ) the second image problem still persists. I will try to see what happens experimentally - perhaps the second image is only a software thing.
Once again, than You.
@@ezyxas I have designed and fabricated these using that code. Not sure what you may be missing.