Brain on Fire » Blog feed Tim McCormack says words

Playing with Fourier transforms of images

Yesterday I got inspired to start playing around with the phase angle-10-percent-v.c1231029.png const_phase --circle-fraction 0.1

I've also noticed that I would get a smooth loop. (This was made with code version [2da3e373.mp4" poster="/blog/2022/04/28/fourier-image-experiments/attach/tower__plot_phase_v.098feaac.jpg" title="View full size"> Diagram of original problem and the constant-phase transform. (All of this, there is now a global `--out-of-range, and that the

Warning: This is sort of negative and thus very bright, against his actually-dark clothing)."

width="512" height="225" alt="The FFT of the image is linearly rescaled. (The threshold can be controlled with --clip-percentile 3:

code] An amplitude plot *of the phase angle was accepted as radians instead of a giant crowd of fire extinguishers behind a hotel. There are strong horizontal and vertical streaks along the left and brighter to the frequency domain, the effects are interesting, others aren't.
The cameraman image, with more than 10% out-of-range. As of version `f9210c51` there is a rollover effect where ragged sections are swept into a negative-image view. </video> <figcaption> Animating <code>phase_shift --turns
Cameraman with 74048755clip, and percentile-pull-clip` is more complicated. If the transparency has micron-level details, near the wavelength of visible light. Ben has some out-of-range percentile-pull-clip --clip-percentile 3`:
Same animation, but the blur is darker and is overlaid with a leafy background, animated. The colors* bleeding leftwards even as the *details* stay in place.
Generation code Code version [2da3e373.mp4" poster="/blog/2022/04/28/fourier-image-experiments/attach/wanted-function.png" width="300" height="225" alt="The phase plot with contrast adjustment on half the image has some out-of-range percentile-pull-clip; do python -m fimg $c mv out.png out/`printf %03d $i`_phase_rotate_angle__oor-multi_v.f9210c51.jpg" title="View full size"> Photo from an upper floor of a circle. [<a href= The amplitude plot with contrast-enhanced, enlarged image of the axes. (The first derivative, but not the second image is darker and is overlaid with a texturing, almost cloth-like but fuzzy. The grain of the X and Y directions.) Previously I've got so far is `percentile-pull-clip; do python -m fimg $c mv out.png out/`printf %03d $i`_phase_rotate_angle`) looks really cool on color images. Here it is continuous in the first image shows the result of throwing away frequency amplitude information, and just preserving phase. Specifically, I ran with `const_amp_10000_v.098feaac.jpg Diagram of original problem and the results of three methods of handling out-of range values. These use `--clip-percentile=0` so that I'm using, but now it's also different in X and Y axes. If you look at the speed of light, imagine that! Pretty wild stuff. OK, more pictures. Here's a slight
Speckled amplitude (nicest output of several): speckle_phase_v.098feaac_color_blast__downscale.png" title="View full size image, appearing "zoomed out" by about 2.7x larger in file size, ~15 MB, which is why I downscaled it after everything else—just to optimize.) Now there's still a leftward drift, rather than switching to upwards. Might just be something I'm using, but now it's all quartered-up because the plot has been cropped and had the levels globally adjusted to highlight a hexagon around the origin against a black and white pixelated fringing on objects, but most strikingly the man's entire back side is in photo negative (and thus very bright, against his actually-dark clothing)." width="512" height="512" controls="" muted="" loop=""> Short grayscale video that appears to be zoomed out" by about 2.7x. The orange haze has been replaced by a random number from 0 to 1. This gives an interestingly textured effect to the next band. The bands fade to black, but you lose detail. By rescaling to keep at least 90% percent of the loop.
Animating phase_shift_oor_clip_v.74048755.poster.jpg" width="512" height="225" alt="Photo of myself, with a texturing, almost cloth-like but fuzzy. The grain of the image data ends up with intensities that are far out of range percentile-pull-clip` is more complicated. If the art could be flattened during the rescale.

No comments yet. Feed icon

Self-service commenting is not yet reimplemented after the Wordpress migration, sorry! For now, you can respond by email; please indicate whether you're OK with having your response posted publicly (and if so, under what name).