I'm a simple man. I see a new ritvikmath video and I upvote.
@weissProduction9 ай бұрын
Very nice explanation. Thanks👍🏻
@nickayres5310 Жыл бұрын
It's ever more fascinating to look at this stuff in the frequency domain, or considering sharpening/debluring as an inverse problem and using things like the Wiener filter to understand it.
@moravskyvrabec Жыл бұрын
Love it. Just came back from a long hike photographing birds and this was the perfect topic to find in my feed.
@nitinkapoor4752 Жыл бұрын
Well…this has been on my mind since 1999/2000 the first time I got my pictures scanned into digital form(I use to shoot slides and even invested in a rather expensive Nikon slide scanner) I had some inkling of what could have been going on(without thinking of matrix mathematics)… and also the inherent limitation … avg of 10 and 20 is 15… but 15 could have come from 5 and 25 too… Finally I have a better understanding of what is being done!
@Deacc Жыл бұрын
Another amazing video, love the diversity of the topics covered recently. Quite eye opening to see the powers of data science in various fields . Would be interesting to see how well a computer vision model could do predicting stocks “visually”
@ritvikmath Жыл бұрын
thanks and interesting idea!
@thomashirtz Жыл бұрын
Men the setup with the white wall and the window is so slick, very nice job To be perfect would be just find a way to not see the microphone as it is just a tiny bit distracting Love the explanation too :) Cheers !
@zenu903 Жыл бұрын
I'm currently taking CS50's computer science course and was stuck on the "edge detection filter" problem for a while. Really cool to understand exactly what is being talked about here.
@ImolaS3 Жыл бұрын
Edge detector kernels are slightly different in how they work. Roberts and Sobel are numerical approximations of 1st order differential operators and Laplacian of Gaussian is a numerical approximation of a second order operator where the zero crossing signifies the edge. They are applied in a raster fashion in the same was as the sharpening filter though
@pradyumnsrivastava3845 Жыл бұрын
Explained seamlessly how sharpening filter is constructed. Could you also do for Edge detection Filter? Thank you
@hkanything Жыл бұрын
Stable diffusion sharpen everything
@tomoki-v6o Жыл бұрын
before,I was thinking that they mimic the lens optics for blurring ,never thought they use easy trick like convolution, I tried the sharpening kernel for a normalized pixel range 0-1 , things becomes different, what s so special about 255?
@ImolaS3 Жыл бұрын
Images are generally comp[rised of either one (greyscale) or three (RGB) 8-bit numbers (binary bits) and 8 bits all set to one is 255. So, the values in an actual image are really never 0-1 but are 0-255 where in a greyscale image (what we call a black and white photo) 255 is white and 0 is black, with 128 a mid grey