CIFilter.io (Core Image Filter Reference)

March 30, 2019

Today I'm launching CIFilter.io 🎉

Screenshot of the CIFilter.io website

CIFilter.io is a project I've been working on for the last few months, and today it's open source. It has two parts:

  1. A website, https://cifilter.io, which lists all the available CIFilters, their information and examples of applying them
  2. An app which allows you to apply each CIFilter to various inputs, tune their parameters, and apply them to camera and photo library images

The website is completely free and I intend to keep it updated with each new release of iOS and macOS. The app is not free, but your purchase helps support this project and others like it, and you can build it from source if you'd prefer.

In this post I'll go into a bit of detail about the project, how it works, and why I built it.

Background

CIFilter is a powerful part of Core Image that every app developer can take advantage of. There are many (207, as of iOS 12) CIFilters available in Core Image - everything from doing image compositing to photobooth effects.

Application of the CIThermal filter
The CIThermal filter

However, it's hard to find documentation on what filters are available or how they work. Apple has a page listing about 85% of the available filters, but it's "no longer being updated", and no new filters which were made available in iOS 10, 11, or 12 are listed there. If you Google CIThermal, for example, most of what you'll find is iOS header dumps.

For the filters which are listed, many have incomplete documentation - for example, the inputSharpness parameter to CICheckerboardGenerator says:

An NSNumber object whose attribute type is CIAttributeTypeScalar and whose display name is Sharpness.

Good info, but doesn't tell you what the parameter actually does. The examples on this page are also limited - for example, CISourceOutCompositing takes an inputImgage and a backgroundImage. The page shows two images being composited, but doesn't specify which image is which:

Image of CISourceOutCompositing from Apple's documentation

Problem

I ran into this documentation issue at work (in fact, I recently gave a talk detailing more background on what issues I ran into), and I thought that there must be a better way. Turns out, it's possible to get info about every filter and their parameters by using CIFilter's attributes var.

The CIFilter.io app takes this filter attributes data, parses and normalizes it, and renders it to a json (which you can view here). The CIFilter.io website parses this json and renders it using Gatsby.

The Website

The CIFilter.io website lists all filters and allows searching through the list:

Gif showing CIFilter.io's search bar

Each filter lists the description and parameters, pulled directly from the filter's attributes dict, which are generally more descriptive. For example, the inputSharpness parameter discussed above has the following description on CIFilter.io:

The sharpness of the edges in pattern. The smaller the value, the more blurry the pattern. Values range from 0.0 to 1.0.

And the example for CISourceOutCompositing shows the input parameter names:

Example of applying the CISourceOutCompositing filter

The App

The CIFilter.io app also shows documentation for all filters, but allows you to interactively apply them as well. Here's a gif of applying CICheckerboardGenerator:

Gif showing CICheckerboardGenerator in the CIFilter.io app
The color wheels don't transfer very well to GIFs

You can also apply filters to photos from your camera or photo library (this is how the examples on the website are generated, and how I generated the cat thermal from earlier in this post). Here's CIComicEffect, for example:

Gif showing CICheckerboardGenerator in the CIFilter.io app

Next Steps

There are several improvements tracked on GitHub that would be good to implement:

  1. Compatibility for Dash docsets
  2. More examples for more filters
  3. Use filters.json to generate a Swift library for applying CIFilters in a type-safe manner

I'll be working on some of these going forward, but I'm hoping the community can get involved too 🙂

Until then, you can check out CIFilter.io on the website, app, and on Twitter.