Projects ideas

A very disorganized list of projects I want to work on, together with notes about them (mostly links with related info)

Cutting vegetables with blood

Pictures of knives cutting vegetable while drenched in fake blood. No real purpose for this, except that I think it would be cool

Tep as a pirateshipe navigating through places:

Tep has a helm in a room that looks like a pirateship. It would be really cool if it was connected to something online, such that it seemed that tEp was a real pirate ship.

The idea would be to have a website using open street map together with open layers

https://news.ycombinator.com/item?id=23722133 maybe use open layers on toip of open street map

Cli picture tools with gui

There are lots of really cool picture effects that only have a terminal interface. It would be exciting to have a program that could be used for automatically generating previews for those tools by running it with smaller image sizes.

Primitive images:

This project is really cool, but I don't like the loss function used by it, so instead use a laplacian pyramid loss for the gradient descent. https://github.com/fogleman/primitive

So the same project, but instead of using this loss function, use one based on laplacian pyramids contrast as done by macrofusion.

Maybe make it available online, https://github.com/silvia-odwyer/photon

seems like a good option! Specially if combined with: https://github.com/silvia-odwyer/gld

Probably use laplacian pyramid loss for the descent. There are some articles about it online.

this one does

Music lyrics video for danger line

The lyrics are displayed while moving around a "red line" . The "danger line" is the body outline of someone who died (red), but you only discover it is the outline of the body at the end of the song.

Throughout the song the lyrics go around the outline

Until the very end, they are generally outside the outline, but certain words can be on the boundary (the inside of the boundary represents death). Some words might be in the other side of the boundary.

The effect of bullet Hole, is zooming inside O of hole and inside it there is another hole, to give the impression of going through.

Maybe some effect of heartbeats ecg during the solos.

Once you are inside the outline, things get a lot redder, like red border.

Daugther brings you further from the line

3d online game of walking around east campus with the murals

Maybe allow people to upload pictures?

I'd really enjoy to have a 3d representation of the inside of EastCampus, with all the murals and all. https://learn.display.land/capturing-reality

display.land would be ideal, but it was discontinued on July 10 2020.

(Display.land is an interesting 3D scanning and modeling app built around Google’s ARCore augmented reality processing. The app allows you to scan objects in 3D and then edit, export, or share your scans, all for free!) https://developers.google.com/ar/develop/java/quickstart

https://3dscanexpert.com/scann3d-android-photogrammetry-app-review/

https://3dscanexpert.com/free-3d-scanning-video-smartphone/

Route panorama on android

I want to implement an app that performs route panoramas on android.

Here are some link ideas. https://en.wikipedia.org/wiki/Route_panorama https://www.cs.iupui.edu/~jzheng/RP/

And maybe strip/slip photography: https://en.wikipedia.org/wiki/Strip_photography

https://github.com/duanhong169/Camera

https://github.com/CameraKit/camerakit-android

Dating website where you ship your friends

Super resolution photo app by upsampling alignning and enfusing or merging, for astrophotography

https://petapixel.com/2015/02/21/a-practical-guide-to-creating-superresolution-photos-with-photoshop/ https://groups.google.com/g/hugin-ptx/c/ijOCcBX7me8 http://www.astrosurf.com/cidadao/super.htm

selfie with longer focal distance by combining multiple close shots

By scanning your face (similarly to a route panorama), fake your lens to have a longer focal distance, while improving blurring the background

idea: use gyroscope to put a dot on the screen that force the video to stay in the same plane. Use maybe motion tracking: https://source.android.com/devices/camera/motion-tracking

Photogrammetry of street art tutorial

start: https://medium.com/realities-io/getting-started-with-photogrammetry-d0a6ee40cb72 video for the above: https://www.youtube.com/watch?v=doiimSnLOBM also http://azadux.com/everydayascan

Also this guy did a lot of it, but left no instructions: https://www.youtube.com/watch?v=IrgyUFkeJOY

sketchfab seems to be the correct place to share it.

https://news.ycombinator.com/item?id=19685891

https://ww.reddit.com/r/3Dprinting/comments/87tk4j/3d_scanning_with_just_a_phonecamera/

with colmap also https://www.youtube.com/watch?v=bDHJM6nAKtc

cindyjs

https://cindyjs.org/docs/ https://montaga.github.io/spherical/ Just play with some spherical droste effects with pictures from mit https://arxiv.org/pdf/1605.01396.pdf

Also map picture into sphere by using reflections/tiling? Idea: convert rectangle to circle conformally, followed by identifying it with one of the hemispheres of a stereographic transformation https://arxiv.org/pdf/1509.06344.pdf

Or maybe assume it is peirce quincucial https://arxiv.org/pdf/1605.01396.pdf

https://mipav.cit.nih.gov/pubwiki/index.php/Transform:_Conformal_Mapping_Algorithms#Transformation:_Circle_to_Rectangle

https://www.osti.gov/servlets/purl/4800477 https://www.osti.gov/biblio/4800477-conformal-mapping-circle-onto-rectangle

Plausible analytics for MIT/sipb

https://plausible.io/blog/self-hosted-web-analytics-beta

Make pictures boring

Detects what is relevant in a picture and removes it. https://www.immersivelimit.com/tutorials/yolact-with-google-colab https://kharshit.github.io/blog/2019/08/23/quick-intro-to-instance-segmentation

use instance segmentation to detect what is interesting

The use this to fill it up: https://github.com/zavolokas/Inpainting https://github.com/YuanTingHsieh/Image_Completion

Quaternions exploration

Have a simple 3d scene. Apply quaternion functions to the positions of the points. Find an equivalent to droste effect but in 3d space

Similar to this demo https://montaga.github.io/spherical/ So more like this: https://eater.net/quaternions/video/intro

But with a more complex 3d scene and allowing more things than rotations, i.e. allowing complex functions in terms of the quaternion that defines the position.

Using a ray marcher https://github.com/HackerPoet/PySpace might be worth it to have some visualizations that use fractals,

Or maybe just use https://cindyjs.org/docs/cindygltutorial/rendering3d.html.

We will effectively be using conformal transformations https://mathoverflow.net/questions/52209/3d-conformal-mappings

It seems that the only types are inversion, translation, rotation and dilation.

maybe use this for the visualization: https://www.youtube.com/watch?v=ztsi0CLxmjw, same thing used for exploring hyperbolic space.

Also, I want everything to happen on the surface of a 4d sphere.

Just more information about quarenions https://marctenbosch.com/quaternions/ < maybe quaternions are trash

https://arxiv.org/abs/1605.01396

Exploring surface of 4d objects.

You yourself is like a cube, so you see infinite copies of you on the distance

maybe use a non euclidean ray tracer https://www.youtube.com/watch?v=YvU-srHhQxw

https://github.com/Limeth/euclider

https://www.youtube.com/watch?v=tl40xidKF-4

but I kinda wnat it to be in javascript

or maybe like: https://arxiv.org/abs/2002.09533 ->> sounds really really promising / seems to be already doing it. https://www.youtube.com/user/ZenoTheRogue https://ww.reddit.com/r/rust/comments/88b6ix/euclider_noneuclidean_raytracing_prototype/ https://twitter.com/zenorogue/status/1245367263936512001

Just add some posts playing with js here

https://ew.reddit.com/r/orgmode/comments/5bi6ku/tip_for_exporting_javascript_source_block_to/

Make music using this imo problem solution

https://www.youtube.com/watch?v=M64HUIJFTZM

toilet measure weight

You sit on the toilet. It shows you your weight.

Practice retouching face images

https://patdavid.net/2011/12/getting-around-in-gimp-skin-retouching.html https://patdavid.net/2013/08/an-open-source-headshot-ronni_21.html

Pagination for my blog

Pagination for this org mode blog

RSS for this blog

Better theme for this blog, make it not ugly

Use seam carving to split a image into puzzle pieces for cutting

https://www.youtube.com/watch?v=6NcIJXTlugc

Add my emacs packages to melpa

https://github.com/melpa/melpa

Random notes

AR core for android

https://www.youtube.com/watch?v=Mzr3MVzSjSM https://github.com/lvonasek/tango/wiki/3D-Scanner-for-ARcore

https://github.com/googlesamples/arcore-depth-lab https://www.youtube.com/watch?v=Mzr3MVzSjSM

Play with some pre trained CNN

https://www.analyticsvidhya.com/blog/2018/07/top-10-pretrained-models-get-started-deep-learning-part-1-computer-vision/

Click to expand table of contents

https://codepen.io/peternguyen/pen/hICga

From: https://codepen.io/peternguyen/pen/hICga

 
 <input id="toggle" style="display: none; visibility: hidden;" type="checkbox" checked>
  <label for="toggle">Table of contents</label>
  <div id="expand">
 Lorem ipsum dolor sit amet, consectetur adipiscing elit. Maecenas porta non turpis faucibus lobortis. Curabitur non eros rutrum, gravida felis non, luctus velit. Ut commodo congue velit feugiat lobortis. Etiam nec dolor quis nulla bibendum blandit vitae nec enim. Maecenas id dignissim erat. Aenean ac mi nec ante venenatis interdum quis vel lacus.
 </div>
#expand {
  height: auto;
  visibility: hidden;
  display: block;
  height: 0em;
  overflow: hidden;
  transition: max-height 0.3s linear, visibility 0.65s ease;
}
#toggle:checked ~ #expand {
  height: auto;
  display: block;
	overflow: auto; 
  visibility:  visible;
  transition: max-height 0.3s linear, visibility 0.65s ease;
}

(defun org-export-head--custom-toc-input(depth info &optional scope)
  (let ((toc-entries
	 (mapcar (lambda (headline)
		   (cons (org-html--format-toc-headline headline info)
			 (org-export-get-relative-level headline info)))
		 (org-export-collect-headlines info depth scope))))
    (when toc-entries
      (let ((toc (concat "<nav id=\"table-of-contents\">\n" 
                  "<input id=\"toggle-toc\" style=\"display: none; visibility: hidden;\" type=\"checkbox\">\n"
                  "<label for=\"toggle-toc\">\n <h2> <a> Table of Contents </a> </h2>\n </label>\n"
                  "<div id=\"text-table-of-contents\">"
			 (org-html--toc-text toc-entries)
			 "</div>\n"
                         "</nav> \n")))
        toc))))

#table-of-contents #text-table-of-contents{
  visibility: hidden;
  display: block;
  height: 0em;
  overflow: hidden;
}

#table-of-contents #toggle-toc:checked ~ #text-table-of-contents {
  height: auto;
  display: block;
        overflow: auto; 
  visibility:  visible;
}

#table-of-contents h2 b{
    cursor: pointer;
    border-bottom:1px dotted;
}
#table-of-contents h2{
    cursor: pointer;
}

(defun org-export-head--custom-toc(depth info &optional scope)
  (let ((toc-entries
	 (mapcar (lambda (headline)
		   (cons (org-html--format-toc-headline headline info)
			 (org-export-get-relative-level headline info)))
		 (org-export-collect-headlines info depth scope))))
    (when toc-entries
      (let ((toc (concat "<nav id=\"table-of-contents\">\n"
                         "<a href=\"#text-table-of-contents\">\n <h2> <b>Table of Contents </b> </h2>\n </a>\n"
                         "<div id=\"text-table-of-contents\">"
			 (org-html--toc-text toc-entries)
			 "</div>\n"
                         "</nav> \n")))
        toc))))

#something {
  display: none;
}

#something:target {
  display: block;
}

Better TOC for tags page

https://www.lycarter.com/tags/

Keep state about displaying comments

https://developer.mozilla.org/en-US/docs/Web/API/Web_Storage_API