Meet Guetzli, the image compression algorithm that might change the shape of the Internet

Guetzli

Google developed Guetzli, a new JPEG encoding algorithm. What’s special with it? The algorithm, which btw is open source, is able to compress a JPG file without loss of image quality and to cut its size by 35% (again: it’s lossless).

Guetzli is a JPEG encoder that aims for excellent compression density at high visual quality. Guetzli-generated images are typically 20-30% smaller than images of equivalent quality generated by libjpeg. Guetzli generates only sequential (nonprogressive) JPEGs due to faster decompression speeds they offer.

From the Google Research Blog:

Guetzli [guɛtsli] — cookie in Swiss German — is a JPEG encoder for digital images and web graphics that can enable faster online experiences by producing smaller JPEG files while still maintaining compatibility with existing browsers, image processing applications and the JPEG standard. From the practical viewpoint this is very similar to our Zopfli algorithm, which produces smaller PNG and gzip files without needing to introduce a new format, and different than the techniques used in RNN-based image compression, RAISR, and WebP, which all need client changes for compression gains at internet scale.

The visual quality of JPEG images is directly correlated to its multi-stage compression process: color space transform, discrete cosine transform, and quantization. Guetzli specifically targets the quantization stage in which the more visual quality loss is introduced, the smaller the resulting file. Guetzli strikes a balance between minimal loss and file size by employing a search algorithm that tries to overcome the difference between the psychovisual modeling of JPEG’s format, and Guetzli’s psychovisual model, which approximates color perception and visual masking in a more thorough and detailed way than what is achievable by simpler color transforms and the discrete cosine transform. However, while Guetzli creates smaller image file sizes, the tradeoff is that these search algorithms take significantly longer to create compressed images than currently available methods.

It’s easy to figure out how big of an impact this will make for the Internet. The algorithm may literally reduce the size of the Internet, and that’s no joke.

Guetzli
20×24 pixel zoomed areas from a picture of a cat’s eye. Uncompressed original on the left. Guetzli (on the right) shows less ringing artefacts than libjpeg (middle) without requiring a larger file size (Image © Google)

You may try out Guetzli on your own, it’s free and available on GitHub. It’s compatible with all browsers and image processing applications, and obviously it’s compatible with the JPEG standard.

What about the name? You’ve seen above that it is Swiss German for cookie. The project was born out of Google Research’s Zurich office.

[via Google Research Blog]

Another mention of Canon’s 2/3″ global shutter sensor with wide dynamic range

canon rumors

I reported previously about a global shutter sensor Canon is researching and developing. This sensor made it into the news again. Nikkei Technology reports about the Canon 2/3″ sensor with global shutter and wide dynamic range, which was first presented at ISSCC 2017.

From the ISSCC 2017 lecture, Nikkei reports:

The size, pixel count, pixel pitch and power consumption of the sensor are 2/3 inches, 2,592 x 2,054, 3.4μm and 450mW, respectively. In a demonstration after the lecture, the company used the sensor to take a picture of an electric fan whose blades are turning […]

Through […] improvements, Canon doubled the saturated amount of electric charge and widened dynamic range. Specifically, it increased the saturated amount of electric charge from 8,100 electrons with a frame rate of 120fps to 16,200 electrons with a frame rate of 60fps.

The dark temporal noise of the new sensor is 1.8 electronsrms. As a result, its dynamic range is 79dB with a frame rate of 60fps and reaches 111dB when HDR is applied.

Well let’s hope we see such a sensor in a Canon DSLR soon. Besides this utopian wishes, this is another report that shows Canon’s commitment to develop innovative sensor technology.

The slides below are from the ISSCC 2017 presentation.

[via Nikkei Technology]

Dual Pixel AF has become a Canon standard, here’s how it works

Dual Pixel Af Panasonic Dfd

It’s safe to assume that Dual Pixel AF (DPAF) has now become Canon’s standard on-sensor auto-focusing technology. Since it made it into a Rebel, it will be featured on almost any future Canon DSLR and MILC (and who knows…maybe also in some high-end PowerShots).

Dual Pixel AF debuted in July 2013 (press release), on the still noteworthy Canon EOS 70D. It took Canon a long time to feature it on other cameras, and even longer to put it into a mirrorless camera. The EOS M5 and EOS M6 are the first Canon MILCs to feature it. And hell if it makes a difference on a mirrorless system. If only Canon had used it before on their mirrorless offering.

Click here to open the rest of the article

Canon research and develop a 2/3″ sensor with global shutter

canon rumors

Interesting Canon research paper spotted by Image Sensors World. Canon is working on a 2/3″ sensor with a global shutter.

The paper is named

A 1.8e-rms Temporal Noise Over 110dB Dynamic Range 3.4μm Pixel Pitch Global Shutter CMOS Image Sensor with Dual-Gain Amplifiers, SS-ADC and Multiple-Accumulation Shutter, by Masahiro Kobayashi, Yusuke Onuki, Kazunari Kawabata, Hiroshi Sekine,Toshiki Tsuboi, Yasushi Matsuno, Hidekazu Takahashi, Toru Koizumi, Katsuhito Sakurai, Hiroshi Yuzurihara, Shunsuke Inoue, Takeshi Ichikawa at ISSCC 2017 on Feb. 6, 2017.

A more in-depth analysis of the technology discussed in the paper can be read at Harvest Imaging:

[…] the presented sensor has a funnel-shaped light guide structure above the pixels, an optimized light shield to keep the PLS low.  To enhance the dynamic range of the sensor, the columns are provided with a gain stage that automatically choses between a gain of 1x or 4x.  With some clever timing of the transfer of the PPD and with an increased readout speed of the sensor, extra new option can be added, such as wider dynamic range and in-pixel coded exposure.

Below are some slides from the paper presentation.

[via Image Sensors World]

Look what you can do with a hacked Canon A650 & Helios 44M-5 lens

canon a650
Hacked Canon A650 with Helios 44M-5 lens

These are the hacks I love most. Alexey Kljatov took a Canon A650 (2007) and did his own thing mounting a Helios 44M-5 58mm f/2 lens on the camera. To make this hack really fit for macro photography, the lens was attached backwards.

The hack appears to be made quickly and doesn’t look neat. However, it produces amazing images like the snowflake photo below. That’s the real hacker spirit: take gear, modify it following an idea, realise something awesome.

Kudos Alexey Kljatov for the cool hack! Be sure to have a look at his website for more amazing macro shots done with the hacked Canon A650, and to learn the how to of his macro setup plus a ton of information about macro photography.

canon a650
Shot with the hacked Canon A650
[via Mirrorless Rumors]

Canon EF 400mm f/2.8L IS II lenses used to discover a new “Dark Galaxy” with Dragonfly Telephoto Array

Dragonfly Telephoto Array

Do you remember the Dragonfly Telephoto Array developed by the University of Toronto’s Dunlap Institute for Astronomy and Astrophysics? Well, there are some news.

First things first. What is the Dragonfly Telephoto Array?

Dragonfly is an innovative, multi-lens array designed for ultra-low surface brightness astronomy at visible wavelengths. Commissioned in 2013 with only three lenses, the array is growing in size and proving capable of detecting extremely faint, complex structure around galaxies. The most recent upgrade—completed in 2016—saw Dragonfly grow to 48 lenses in two clusters.

Last time I reported about Dragonfly it had 10 lenses mounted, now the lenses are 48. To build the Dragonfly, scientists used Canon EF 400mm f/2.8L IS II lenses, because of “unprecedented nano-fabricated coatings with sub-wavelength structure on optical glasses“. I guess the lenses were modified according to the scientists’ needs.

Canon EF 400mm f/2.8L IS II
Canon EF 400mm f/2.8L IS II, yours for $10,000.

Next, what is the Dragonfly Telephoto Array good for?

Dragonfly is designed to reveal the faint structure [of the universe] by greatly reducing scattered light and internal reflections within its optics. It achieves this using commercially available Canon 400mm lenses with unprecedented nano-fabricated coatings with sub-wavelength structure on optical glasses.

Also, Dragonfly images a galaxy through multiple lenses simultaneously—akin to a dragonfly’s compound eye—enabling further removal of unwanted light. The result is an image in which extremely faint galaxy structure is visible.

Well, scientists discovered a previously unknown Dark Galaxy by using the Dragonfly Telephoto Array. The galaxy was named after the array. As The Atlantic reports:

Dragonfly 44 is a dim galaxy, with one star for every hundred in our Milky Way. But it spans roughly as much space as the Milky Way. In addition, it’s heavy enough to rival our own galaxy in mass, according to results published in the Astrophysical Journal Letters at the end of August. That odd combination is crucial: Dragonfly 44 is so dark, so fluffy, and so heavy that some astronomers believe it will either force a revision of our theories of galaxy formation or help us understand the properties of dark matter, the mysterious stuff that interacts with normal matter via gravity and not much else.

The discovery was made by astronomers Pieter van Dokkum of Yale University and Roberto Abraham of the University of Toronto. They did not use Canon sensors, tough. The lenses are mounted on SBIG STF-8300M CCD cameras. The array began imaging targets in 2013 from its home at the New Mexico Skies hosting facility.

If you are interested and want to know more, go for the videos below.

About the Dragonfly Telephoto Array

Below: about the newly discovered galaxy

[via The Atlantic]