Google Super Resolution Zoom: Good Start but not There Yet

Our first testing of Google’s super resolution zoom recently announced in Pixel 3 shows that it indeed can restore some image details, but is still behind the best in class solutions.

Comparing to the “normal” digital zoom, which is basically an upscaling plus edge enhancement, Google’s zoom reveals some details that are indistinguishable in the “normal” image:

However, it’s still not the best of what super resolution can achieve. Below is a comparison of Google’s super resolution zoom to Almalence SuperSensor, a technique based on multi-frame super resolution, running on the same Pixel 3 smartphone:

It looks like Almalence SuperSensor is closer to be delivering on the “optical zoom” promise.

Another nice example captured when taking zoomed images of a book on optical design. Google super resolution makes the text somewhat better readable, however some characters still remain distorted beyond any possibility to recognize them. Also some minor color artefacts are introduced on the originally black text:

And again, if you really want to capture a readable text, Almalence SuperSensor is a solution (note that SuperSensor also got the white color of paper right):

Unfortunately, there is no way to reliably measure the Google’s super resolution processing time. It looks like the processing is performed in background. When trying to quickly open the image right after its icon appears in the camera app, the preview would still show a progress sign for a fraction of a second, probably implying that the image is still being processed. One can feel that the processing takes roughly one second altogether, but there is no way to verify that number. Almalence SuperSensor processing takes 200-500 ms on the same hardware (SnapDragon 845).

The full images used to make the above comparison examples are available in an archive below. Note: for accurate comparison the images of each scene were taken with the same Pixel 3 smartphone, from the same position under the same lighting. The zoom level might slightly differ between images as there is no way to precisely set the zoom factor in the Pixel 3 camera app. The Pixel 3 camera app, updated to the recent version as of Jan 25, 2018, was set to “HDR+” mode.

PS Stay tuned, we will soon show a comparison of super resolution zoom at video frame rates!

Google Super Resolution Zoom: Good Start but Long Way to Go

Please read the updated post.

After publishing this original post we were contacted by Google engineers who pointed to the fact that Super Res was not always on in the images we took. Of course, it happened unintentionally, there’s just no obvious way to make sure that Super Res kicks in when taking an image.

We have re-taken the test images as suggested by Google and we admit the results are much better now. However Almalence SuperSensor is still superior :).

Almalence Digital Lens technology enables high resolution wide FoV VR HMDs

Digital Lens addresses the smear and aberrations inherent to thin and lightweight VR HMD optics.

First objective quality measurements performed with a demo based on HTC Vive showed up to 2.7 times effective resolution improvement and suppression of chromatic aberrations even at the edges of field of view.

The technology opens the door for high resolution displays, solving the problem of optical performance which does not allow to harness the full advantage of extra display resolution. It also enables wide field of view HMDs where any acceptable picture quality at wide angles cannot be achieved with an optical design alone.
Aberration and MTF charts from the report on quality measurement, comparing HTC Vive built-in lens correction to Almalence Digital Lens:

MTF50, HTC Vive with/without Almalence DLVR

MTF50, HTC Vive with/without Almalence DLVR


Almalence DLVR removes chromatic aberrations in HTC Vive even at the edges of field of view

Almalence DLVR removes chromatic aberrations in HTC Vive even at the edges of field of view


HTC Vive with/without Almalence DLVR, slanted edge chart for Imatest measurements

HTC Vive with/without Almalence DLVR, slanted edge chart for Imatest measurements

Almalence’s new technology makes cameras see like human eye

Have you ever noticed that you see much more details when looking at a scene by eye than when seeing a photo of the same scene? If not, it’s just because it’s too common to be noticed. Even having, physically, a lower dynamic range than a camera, human eyes see more.

Almalence Dynamic Range Optimizer optimizes the usage of dynamic range of 8-bit images to make the most of image details visible. DRO uses a technique similar to tone mapping, exploiting the peculiarities of human vision. DRO can be used either for optimal conversion of RAW images into 8-bit RGB/YUV images or for improvement of existing 8-bit images.

Almalence Dynamic Range Optimizer reveals details in highlights and shadows. Automatic processing within milliseconds.

Same data acquired by the camera, different usage of dynamic range. Left: standard image produced by the camera. Right: result of Almalence DRO processing, close to what human eye would see. Details in shadows are revealed due to optimization of dynamic range usage. Noise is not amplified.

DRO is easy-to-integrate, extremely fast solution that automatically improves the images in 100% cases and can substitute HDR in most of cases.

Comparing to existing solutions aiming to provide similar improvements of digital imaging, Almalence DRO:

  • is faster by the order of magnitude, which allows to use it in mobile devices, for both still images and video
  • provides better quality, mainly due to absence of noise amplification in shadows
  • improves both shadows and highlights

Details in shadows are revealed due to optimization of dynamic range usage. Noise is not amplified.

Left: standard image, right: DRO processed (optimized) image. Details in both shadows and highlights are revealed due to optimization of dynamic range usage.

DRO can be integrated in ISP, camera drivers or applications.

Some comparison examples are available on our site: DRO_examples_Almalence_site.zip (33 MB)

Demo application is available upon request.

Almalence makes super resolution available on camera phones

Almalence, the developer of world’s first super resolution technology commercially available on desktop computers, announces that its solution becomes available in the most advanced smartphones in 2013.

Super resolution is now used to provide high quality pictures at high zoom levels in Huawei’s flagship models Ascend P2 and Ascend P6, SHARP’s new devices and emerging devices of several Korean and Taiwanese mobile phone makers.

Nowadays, the resolution of modern mobile phones is quite enough for amateur photography, unless it comes to zooming. With no optical zoom lens, the zoomed images are nothing more than upsized and sharpened low resolution crops. For example, 4x zoom on a 8 Megapixel mobile phone has 0.5 Megapixel effective resolution at best, not taking lens blur into account.

That’s where Almalence’s super resolution can be effectively used to enhance the quality of the images. While further improvement of sensor and lens seems either impossible or very expensive, software solutions allow going beyond the camera’s physical limits.

“Almalence’s Super Resolution Zoom is a pure software solution that does not require change of sensor or lens. This allows an extremely quick and cost-effective integration” – says Eugene Panich, CEO Almalence, Inc., – “in the beginning of 2013 we have integrated Super Resolution Zoom in quite different devices developed by several OEMs in very short timeframes and with almost no efforts required from OEMs.”

Huawei Ascend P2 Super Resolution Zoom Test

Huawei Ascend P2 was the first device with super resolution zoom on the market. These comparison photos were taken at MWC 2013 where P2 was presented. (Left: usual zoom; right: super resolution zoom)

Unlike edge enhancement and noise filtering techniques, super resolution provides real increase of effective resolution, making visible the details that are indistinguishable in normal shots. It virtually doubles the megapixel count of the camera. Capturing more details in zoomed area, it replaces optical zoom lens, for a trifling fraction of cost of such lens, and adding no single atom of weight and size to the device.

Huawei Ascend P6 Super Resolution Zoom test under 150 Lux: Resolution increases 1.9 times

“Spilled coins” (aka “Dead Leaves”) chart taken with Huawei P6 mobile phone (illuminance: 150 lux). Normal 2x zoom vs Super Resolution 2x Zoom. MTF30 measurement with ImaTest shows 1.9x effective resolution increase.

Almalence’s Super Resolution Zoom is a multi-frame technology, which adds benefit of drastic noise reduction in low light conditions.

Super Resolution Zoom under low light (20 Lux). Noise is reduced with no loss of detail. (Test with Huawei Ascend P6)

Images taken with Huawei P6 mobile phone under low light (20 Lux): Normal 2x zoom versus Super Resolution 2x Zoom

Super resolution zoom by Almalence can be used in quite different modes of operation. In some devices it is a background processing while in others it works real time in viewfinder.

Comparing to the technologies that imply physical improvement of the camera, such as using BSI sensors or 41 Megapixel sensors, Super Resolution Zoom is not just a competitor, but also a good addition that is compatible with almost any kind of sensor and lens and improves the image quality no matter how good the camera is.

Almalence’s nearest plan is to implement a hardware super resolution zoom using Tensilica IVP32 image processor, which will result in a solution capable of real time video processing and processing of high resolution still images in tens of milliseconds.

More  information: http://almalence.com/embedded-imaging#superzoom

Almalence Super Resolution vs SONY Clear Image Zoom (single-frame super resolution)

From time to time we are being asked to compare Almalence Super Resolution with other super resolution technologies available on the market, most of them single-frame super resolution. Our previous test was Panasonic Intelligent Zoom.

Today we will test SONY Clear Zoom Image technology, which is used in some SONY cameras for better quality of digital zoom.

We use SONY RX100 camera with Clear Zoom function.

First of all, let’s compare zoomed images taken with Clear Zoom disabled and enabled:

Top: standard zoom, bottom: Clear Image Zoom (SONY RX100)

Top: standard zoom, bottom: Clear Image Zoom (SONY RX100)

We see that Clear Image Zoom produces somewhat sharper image, and somewhere it adds a bit more details (for example, see the second hieroglyph in the center).

Now let’s compare to Almalence Super Resolution. We’ve taken a series of RAW images at the same zoom level and processed with PhotoAcute:

SONY Clear Image Zoom vs Almalence Super Resolution

SONY Clear Image Zoom vs Almalence Super Resolution

It’s easy to see that Almalence Super Resolution adds more details and is far superior to Clear Image Zoom. This is an expected result when comparing a single-frame resolution enhancement technology to a multi-frame super resolution.

More examples:

The web address and phone number unreadable in a Clear Image Zoom photo become readable in Super Resolution image:

Letters in the center of the image and “3F” at the bottom-right corner become readable:

Original images and a copy of the application used for processing are available upon request.

Test of Panasonic LX7 Intelligent Zoom

Panasonic features so-called Intelligent Zoom mode in Panasonic Lumix LX7 camera, claiming that with their “Intelligent Resolution” technology 3.8x optical zoom virtually extends to a 7.5x equivalent.
It sounds as single-frame super resolution, so we decided to take some test images to see how good is it.

Below: 3.8x versus 7.5x zoom

Panasonic Lumix LX7: 3.8x zoom and 7.5x (intelligent) zoom

Panasonic Lumix LX7: 3.8x zoom and 7.5x (intelligent) zoom

100% Crops. Top: 3.8x zoom, bottom: 7.5x zoom (with Intelligent Resolution)

Comparison of 3.8x zoom with 7.5x Intelligent Zoom (Panasonic Lumix LX7)

Top: 3.8x zoom, bottom: 7.5x Intelligent Zoom

As we can see, Intelligent Resolution does not add more details (so the resolution is not increased). The same result can be achieved with extra sharpening. See below a comparison between “Intelligent” zoom and bicubic upsize + unsharp mask filter in Photoshop:

Left: Panasonic Intelligent Resolution, right: generic zoom + bicubic upsize + unsharp mask

Left: Panasonic Intelligent Resolution, right: generic zoom + bicubic upsize + unsharp mask

Conclusion: Panasonic Intelligent Zoom (or Intelligent Resolution) delivers sharper images but does not provide higher resolution and in no way can compete with optical zoom.

Full original images available upon request

Superb quality zoom with newly available Huawei Ascend P2, first tests at MWC 2013

Camera became one of the things (if not the only one) that really make the difference in modern smartphones. Among the leading OEMs such as HTC, Sony and Intel who emphasize their new advanced camera features, Huawei offers something really unique – SuperZoom feature, allowing to take high quality images when zooming in with a mobile phone camera.

We had a chance to test SuperZoom feature with Huawei Ascend P2 phone at Huawei’s stand at Mobile World Congress. In brief, it works incredibly, see the pictures below.

Comparison: SuperZoom vs normal zoom, Huawei Ascend P2

Left: normal zoom, right: Super Zoom

SuperZoom versus ordinary zoom comparison, Huawei Ascend P2

Left: generic digital zoom, right: SuperZoom

Comparison: Super Zoom versus generic zoom, Huawei P2

Left: ordinary zoom, right: Super Zoom

SuperZoom on Huawei P2, sample image, test against usual zoom

Left: usual zoom, low detail; right: Super Zoom, high resolution and lot of details

Conclusion: SuperZoom is a revolutionary technology providing zoom quality comparable to optical zoom, on mobile devices!

Feb 25, 2013, Barcelona.