How come we adopt modern video formats quickly but images are stuck in ancient formats?


We’ve moved from AVC to HEVC and VP9 relatively quickly, but for some reason images are stuck in JPEG, PNG, and GIF despite newer formats like JPEG2000 and HEIC having been around for years? Videos are just images displayed very quickly, shouldn’t the adoption pace of image/video formats be similar?

In: 40

There is usually not a practical necessity to compress still images as much as possible. We have enough disk space and network bandwidth. Video is almost always visibly degraded out of necessity, and a better format can allow to bump its quality.

It’s also convenient to be able to display a large photo quickly because the compression is simple and doesn’t place much load on the processor. For example, to generate thumbnails for great many files. New formats are many times slower. This was more important when JPEG-2000 was created. New formats usually only target low bitrates, and may not scale to transparent quality, because they have implicit chroma subsampling. Their decoding speed is typically proportional to the bitrate.

Compatibility with old graphics editors is important too. An old version of Photoshop may have cost a lot of money in the past, is free from subscription fees and still fulfills most needs.

New formats are heavy encumbered with licensing, and usually receive an implementation only in new versions of expensive software. Free software may get clunky non-native plugin at best. JPEG-2000 and related wavelet formats usually had paid plugins in free viewers.

iPhones use HEIC by default, so it is probably one of the most widely used image formats out there today.

WebP is quickly gaining adoption on all major platforms.

There’s also a huge amount of progress happening in the professional photography space with all the different RAW formats and support for newer tech like HDR.

Aside from that, old formats are still in use because there’s nothing really wrong with them. The top motivation for all the newer video codecs you mention is better compression to save on storage costs and bandwidth. Static images are *tiny* in comparison and modern computers and internet connections can handle even badly optimized PNGs or JPEGs perfectly fine.

Videos are huge so a theoretical 10% improvement in size is a big deal. Pictures are tiny so nobody cares if you can save a few kb.

Because jpegs can be compressed to almost junk – Facebook (And hence Instagram) compresses the images using their own algorithm to absolute crap but it’s all designed to look good for a brief glance on the phone – and jpeg is pretty much the universal format on all phones.

btw FB have made their horrendous image compression available – it’s called Spectrum

JPEG is simple and fast and good enough for a lot of purposes. By contrast, the sheer size of video files meant that it took a long time to get to “good enough” so there was much more motivation to keep pushing new tech out to consumers.

It has a lot of problems which make it unsatisfactory for fine art, scientific, medical or similar precision work but these are less common uses:
JPEG is always lossy due to rounding errors in the algorithm and it doesn’t support HDR, which is essential when working with wide gamuts.

JPEG2000 fixed these major issues as well as improving the compression ratio and artifact generation and providing more flexibility for multichannel (CMYK or scientific hyperspectral images) as well as handling giant images as tiles. However, at the time of release, the whole technological area of wavelet transformation was swarming with patent sharks so few companies dared to support it and those that did charged a ton of money for the support. It also required huge amounts of CPU time – early versions of the software were as much as 100x slower than regular JPEG, which meant that a file which would take 1 second to load in JPEG format, could take over a minute to load in JPEG2000 format.

Now that we have more storage capacity there is less need for discerning users to use JPEG2000. Camera raw files and TIFF files provide the necessary features.