Tags: files

I’ve been playing around with the newly-released Squoosh, the spiritual successor to Jake’s SVGOMG. You can drag images into the browser window, and eyeball the changes that any optimisations might make.

On a project that Cassie is working on, it worked really well for optimising some JPEGs. But there were a few images that would require a bit more fine-grained control of the optimisations. Specifically, pictures with human faces in them.

I’ve written about this before. If there’s a human face in image, I open that image in a graphics editing tool like Photoshop, select everything but the face, and add a bit of blur. Because humans are hard-wired to focus on faces, we’ll notice any jaggy artifacts on a face, but we’re far less likely to notice jagginess in background imagery: walls, materials, clothing, etc.

On the face of it (hah!), a browser-based tool like Squoosh wouldn’t be able to optimise for faces, but then Cassie pointed out something really interesting…

When we were both at FFConf on Friday, there was a great talk by Eleanor Haproff on machine learning with JavaScript. It turns out there are plenty of smart toolkits out there, and one of them is facial recognition. So I wonder if it’s possible to build an in-browser tool with this workflow:

• Drag or upload an image into the browser window,
• A facial recognition algorithm finds any faces in the image,
• Those portions of the image remain crisp,
• The rest of the image gets a slight blur,
• Download the optimised image.

Maybe the selecting/blurring part would need canvas? I don’t know.

Anyway, I thought this was a brilliant bit of synthesis from Cassie, and now I’ve got two questions:

1. Does this exist yet? And, if not,
2. Does anyone want to try building it?

In Going Offline, I dive into the many different ways you can use a service worker to handle requests. You can filter by the URL, for example; treating requests for pages under /blog or /articles differently from other requests. Or you can filter by file type. That way, you can treat requests for, say, images very differently to requests for HTML pages.

One of the ways to check what kind of request you’re dealing with is to see what’s in the accept header. Here’s how I show the test for HTML pages:

if (request.headers.get('Accept').includes('text/html')) {
    // Handle your page requests here.
}

So, logically enough, I show the same technique for detecting image requests:

if (request.headers.get('Accept').includes('image')) {
    // Handle your image requests here.
}

That should catch any files that have image in the request’s accept header, like image/png or image/jpeg or image/svg+xml and so on.

But there’s a problem. Both Safari and Firefox now use a much broader accept header: */*

My if statement evaluates to false in those browsers. Sebastian Eberlein wrote about his workaround for this issue, which involves looking at file extensions instead:

if (request.url.match(/\.(jpe?g|png|gif|svg)$/)) {
    // Handle your image requests here.
}

So consider this post a patch for chapter five of Going Offline (page 68 specifically). Wherever you see:

if (request.headers.get('Accept').includes('image'))

Swap it out for:

if (request.url.match(/\.(jpe?g|png|gif|svg)$/))

And feel to add any other image file extensions (like webp) in there too.

So, remember when I posted all those episodes of Simon Singh’s Five Numbers radio series on Pownce so that they’d have permanent URLs? Yeah, well, so much for that.

Fortunately Brian had saved all the MP3s. I’ve posted them on S3 and huffduffed them all. I can be fairly confident that Huffduffer won’t be going the way of Pownce, Magnolia, Geocities, and so many more.

Anyway, if you want to listen to the fifteen episodes of the three radio series’ on mathematics, you can subscribe to the podcast at https://huffduffer.com/adactio/tags/five+numbers/rss.

Or you can listen to each episode at these permanent URLs:

1. Five Numbers

   1. A Countdown to Zero
   2. Simple as Pi
   3. The Golden Ratio
   4. The Imaginary Number
   5. Infinity

2. Another Five Numbers

   1. The Number Four
   2. The Number Seven
   3. The Largest Prime Number
   4. Kepler’s Conjecture
   5. Game Theory

3. A Further Five Numbers

   1. 1 — The Most Popular Number
   2. 2 — At The Double
   3. 6 Degrees of Separation
   4. 6.67 x 10^-11 – The Number That Defines the Universe
   5. 1729 — The First Taxicab Number

I like Tumblr. I like Pownce. They both make it very quick and easy to post discrete quanta of information. I use Pownce for posting audio files and links to videos. I use Tumblr to post quotations. But both services suffer from the same problem: refindability.

Magnolia and Delicious encourage tagging. Those tags can then surface some pretty interesting aggregate behaviour but first and foremost, they’re useful for the individual doing the tagging. It’s pretty easy for me to track down something I bookmarked on Magnolia even if it was quite a while back. I don’t need to keep a list of all the tags I’ve ever used: I just need to search for a word that I think I might have used when I was tagging a bookmark. While it would be very difficult for me to try to second-guess how someone else might describe something, it’s usually pretty easy to put myself in the shoes of my past self.

As my store of data on Pownce and Tumblr increases, I’m starting to miss tagging (or any kind of search) more and more. Then again, I can understand why both services would resist that kind of scope creep. Both services rely on their simplicity. Adding another field to fill in could potentially be a road block between the user and the task they want to accomplish (although it doesn’t feel that way with Delicious or Magnolia). Update: it turns out that you can tag in Tumblr but it’s hidden behind the “advanced” link. Thanks to Keith Bell for pointing that out.

Here’s a case in point. Over time I’ve been posting MP3 files to Pownce of a series of radio programmes by Simon Singh, author of The Code Book — a superb piece of work. The audio from the radio programmes is available from the BBC website but only in Real Audio which, let’s face it, is complete pants. I originally got the MP3 files from Brian but after a catastrophic hard drive crash, I realised that it would be better to store them at an addressable URL. Besides, I wanted to geek out with my mathematically-minded friends. Pownce’s raison d’être is sharing stuff with friends so it seemed like the perfect home for the Singh files.

But without any kind of tagging or search, there’s no easy way for me or anyone else to revisit just those files at a later date. As a temporary patch, I’m listing the URLs for the Pownce posts that correspond to each episode. If you want to download the files, you’ll need to log in to Pownce.

1. Five Numbers

   1. A Countdown to Zero
   2. Simple as Pi
   3. The Golden Ratio
   4. The Imaginary Number
   5. Infinity

2. Another Five Numbers

   1. The Number Four
   2. The Number Seven
   3. The Largest Prime Number
   4. Kepler’s Conjecture
   5. Game Theory

3. A Further Five Numbers

   1. 1 — The Most Popular Number!
   2. 2 — At the Double
   3. 6 Degrees of Separation
   4. 6.67 x 10^-11 — The Number that Defines the Universe
   5. 1729 — The First Taxicab Number