pix-plot VS three.js

Density based clustering with high dimensional data will tend to struggle. This is because, in high enough dimensions, you need a lot of samples to see any density. Also distances start to look very similar (from the curse of dimensionality). To get any traction on such things you need some form of dimension reduction. For something like this non-linear techniques are going to be better. If you want a pipeline of standard parts then something like:

Pretrained-CNN --> UMAP --> HDBSCAN

can turn out relatively reasonable results, especially if the UMAP you use for the clustering is to more than 2 or 3 dimensions (often 5 to 20 is good, depending on the data). You can, of course, still use a 2D UMAP to visualize the results. If you want such a pipeline packaged up then consider the PixPlot package, designed for exactly this use case, from the Yale Digital Humanities Lab: https://github.com/YaleDHLab/pix-plot

* Disclaimer: I am highly biased, as an author of both HDBSCAN and UMAP implementations.

Regarding the optimization of the effect, I found an issue on the three.js GitHub repository: https://github.com/mrdoob/three.js/issues/14104 The alpha channel issue seems to be a difficult problem, which has been present since 2018. At the end of the discussion, a solution proposed by a contributor is not to modify the UnrealBloomPass but to blend the source texture and target texture in the shader.

Here's how we used Three.js to place and find affirmations in the sky.

Three.js is dedicated to generating 3D visuals, offering a powerful suite of features and tools. It is built atop WebGL and optimizes GPU usage for rapid rendering, ensuring smooth performance across devices, including those with limited resources. Three.js's versatility allows for the creation of a vast spectrum of 3D graphics and animations, ranging from simple shapes to intricate models.

7. three.js

Most of frontend libraries are made with Vanilla JS. An example of library that you might frequently use is "Chart.js". But React is not compatible with Chart.js so here it comes "React-chartjs-2" A wrapper library to work with Chart.js in React ecosystem. Oh you want to use "three.js" for some cool 3D? you will need "React-three/fiber". In my case, I need to implement "telegram-web-app", not so fast, I have to create my own wrapper to be able to use it.

For one of my 100-Day Challenges, I chose to dive into Three.js, a JavaScript 3D library. It's easy to start learning, and it becomes difficult to master with the inclusion of shaders.

Nonetheless, it offers nearly infinite possibilities. You can code and create anything, from a mobile app with React Native to a desktop application with Electron.js. Furthermore, JavaScript provides excellent frameworks and libraries that can be applied in various approaches, such as React, Three, Vue, Meteor, and Astro.