Fun with Frequencies and Gradients

Harry Ho

Here is the writeup for the computational photography project on using frequencies and gradients!

Unsharp Images

Our original gorilla-rhino
Minor detail added to the legs

Hybrid Images

Original Nina from Fullmetal Alchemist
Frequency spectrum
Nina with a blurred convolution
Frequency spectrum of Nina
Original Alexander from Fullmetal Alchemist
Frequency spectrum
Alex with higher frequency filtered out
Frequncy spectrum
Hybrid image of Alexander and Nina from FMA
Frequency spectrum
Professor Quirrell
Low pass of Professor Quirrell
Frequency Spectrum
He Who Must Not Be Named
High band
High band spectrum
A repeat of the first image.
FFT of both

Gaussian Stacks

stack level 5
stack level 4
stack level 3
stack level 2
stack level 1
stack level 0
stack level 5
stack level 4
stack level 3
stack level 2
stack level 1

Multiresolution Blending

Same as above except for multiresolution blending
Quirrell
Note that since the image is darker, Voldemort can still be visible
Eye
Mask
Hand
Horror

My terrible attempt at poison gradients

This portion of the project referred to images by relative difference in pixel intensity rather than using actual color. To compute the differences, we must literally get the difference from every pixel to all 4 neighbors. Fortunately there is a faster way to do this than usingbrute force techniques. With a large matrix composed only of zeros, you can compute the total differences in a matrix. I used a sparse matrix, which is a very large matrix composed mostly of zeros.