GIFs are an omnipresent part of the internet. They are used on Twitter as a way to react to something, they are used in blog posts to make them more interesting, and they are used for the majority of conversations amongst coworkers on Slack (or maybe that's just me?). With that being said, GIFs are not good for the web.
GIFs are not optimized to be used on the web: their file size is very large and decoding GIFs in the browser is not a very performant operation. However, there is a better alternative to using GIFs and that alternative is to use videos instead.
Video files are much smaller than GIFs and they are more easily decoded in the browsers compared to GIFs. Actually, when you post a GIF on Twitter, Twitter converts the GIF into a video, which dramatically reduces the file size. This is something very important for Twitter, where users may encounter many GIFs while scrolling their timeline (which translates to more data being downloaded on a user's phone). By reducing the amount of data a user needs to download to view content, they are able to load assets faster as well as use less mobile data (which can be very expensive, especially for us Canadians).
Currently, the best article i've found on replacing GIFs with videos is this Google Web Fundamentals article that I suggest you read if you're interested in all this.
Evaluating GIFs in existing web apps
Because Twitter does GIF optimizations, I was curious to see if other web apps also tried to optimized GIFs that are uploaded to their app. As an experiment i've taken a GIF of Fogel from Superbad and uploaded it on three web apps to see how each app handles GIFs and analyze the final file size of the optimized GIFs. The three web apps i've looked at are Twitter, Dev Community, and Medium.
As a reference, the size of the above raw GIF used for this experiment is 1.0 MB.
As mentioned earlier, Twitter converts GIF files to video files (specifically MPEG4).
Twitter reduces the size of the original GIF asset by 94% by converting it to a MPEG4 file!
Medium
On Medium, you have to ability to upload GIFs to the articles you write. As it turns out, Medium doesn't do any sort of optimizations for the uploaded GIFs:
The original files ize of the GIF remains unchanged and the load time for the GIF is around 10 times the load time of the MPEG4 file that Twitter uses. That being said, Medium does do a cool optimization where they show a very blurred image of your GIF until it has fully loaded the GIF, so that the user can at least see something while waiting for the GIF to load.
Dev Community
The Dev Community, just like Medium, doesn't optimize uploaded GIFs for articles written on the platform.
Again, the original file size of the GIF remains unchanged, but the load time is faster compared to Medium (386 ms vs 485 ms). Not sure why this is, but the load time of an asset depends on many factors, network speed being one of them.
One interesting thing to point out is that the Dev Community uses the loading="lazy"
attribute on their <img>
elements. This allows for the GIF to only be loaded when a user scrolls near it. However, at the time of writing, the loading
attribute is only currently supported in the latest version of chrome, and can only be enabled through a feature flag.
One other thing that I've noticed with the Dev Community is that they are using Cloudinary as a CDN for their image and video assets (notice the URL of the GIF source in the HTML <img>
element). Because they are using Cloudinary, the Dev Community could take advantage of Cloudinary's ability to use lossy compression to compress any uploaded GIF assets and reduce the file size by around 50% without losing too much quality. Even better, Cloudinary can automatically convert GIFs to WebM and/or MPEG4 videos which will result in even more savings in file size compared to the lossy GIF compression.
Best video format
There are 2 video formats that are popular choices for videos on the web: WebM and MPEG4. MPEG4 is the format that's been around for the longest since around 1999 and enjoys great browser support. As shown above, Twitter uses MPEG4 videos to replace GIFs. WebM has been around since 2010 and has a slight advantage over MPEG4 in that it can achieve equal quality video at a smaller file size compared to MPEG4. Unfortunately, WebM video browser support isn't as good as that of MPEG4.
Converting GIFs to WebM and MPEG4 using ffmpeg
ffmpeg is a free command line tool that can be used to convert GIFs into WebM or MPEG4 video files. You can do a lot of audio and video manipulations with ffmpeg, but I will admit that knowing the right commands to use is not obvious when reading their documentation or if you are new to audio/video-specific terminology.
In order to run any ffmpeg command, you'll need to first install it on your machine.
Personally, what I end up doing is that I convert a .gif file (or .mov if I do a screen recording on my mac) into both a .webm and .mp4 file. Specifically, the two commands that I use are the following:
Convert .gif to .webm
1ffmpeg -i input-name.gif -c:v libvpx-vp9 -b:v 0 -crf 40 output-name.webm
-i input-name.gif
-> Specify the input file
-c:v libvpx-vp9
-> Selects the VP9 webm video encoder
-b:v 0
-> Specify a bitrate of 0 to enable "Constant Quality" mode.
-crf 40
-> Specify the video quality to be used. A lower number corresponds with a higher quality video, but also a larger file size.
Convert .gif to .mp4
1ffmpeg -i input-name.gif -b:v 0 -crf 25 output.mp4
-i input-name.gif
-> Specify the input file.
-b:v 0
-> Specify a bitrate of 0 to enable "Constant Rate Factor" mode for the MPEG4 encoder.
-crf 25
-> Specify the video quality to be used. A lower number corresponds with a higher quality video, but also a larger file size.
Notice how the -crf
value passed to the MPEG4 command is lower than that of the value passed to the WebM command, which might make you think you're getting a higher quality video with MPEG4. However, since we're using different encoders between both WebM and MPEG4 conversions, the -crf
values are not comparable. You need to evaluate quality by watching the outputted video.
When converting the Fogel GIF that I used earlier in the article to WebM and MPEG4 using the above commands, I get a WebM file size of 59 KB and and MPEG4 file size of 105 KB. To demonstrate the visual quality of both videos, i've added them below, with the WebM video first and the MPEG4 video second. Of course, if you're viewing the WebM video on an unsupported browser, it sadly won't work for you.
Adding videos to HTML
You can add videos to your HTML using the <video>
element along with nested <source>
elements. What's cool is that you can add multiple sources inside your <video>
element in order to specify fallback videos in case a particular video type is not supported. For example, if we want to use WebM videos for browsers that support WebM and MPEG4 videos for browsers that don't support WebM, we could use the following markup:
1<video>2 <source src="my-video.webm" type="video/webm" />3 <source src="my-video.mp4" type="video/mp4" />4</video>
Furthermore, if you'd like to have the video behave similarly to a GIF by having it autoplay and infinitely loop, you can add the boolean attributes autoplay
, loop
, muted
, and playsinline
to the <video>
element.
1<video autoplay loop muted playsinline>2 <source src="my-video.webm" type="video/webm" />3 <source src="my-video.mp4" type="video/mp4" />4</video>
As you can see, GIFs are ill-suited for usage on the web, but videos can replace them. Of course, it does take some work on the developers side in order to convert GIFs to videos, but that can result in a meaningful impact to end users. As alluded to earlier, there are some tools that can help automate the conversion of GIFs to videos, by using services such as Cloudinary to host your image and video assets.