See our launch story for the Lumia 1020, our previous analysis and our review part 1 for some background to this feature, but in summary:
| Optics | Processing | Extras | |
| PureView original (Nokia 808) | 1/1.2" | Dedicated Broadcom chip, no artificial enhancement | None, zoom/framing is set at capture time |
| PureView 2013 (Nokia Lumia 1020) | 1/1.5", Back Side Illuminated | Handled in the main SoC (2GB of RAM), processing tuned to boost colours and for sharpened detail | Optical Image Stabilisation, 'live' photo reframing |
With a little help from some fancy javascript (if you're viewing this on a mobile screen then you're probably advised to stop and go find a laptop or desktop - you'll need the screen width and a mouse or trackpad will help!), here are direct comparisons of a number of test photos and subjects. And don't worry, I'll hold your hand all the way in terms of drawing conclusions.
Each of these tests, although I was varying the subjects and light conditions, were aimed at answering the question above, i.e. looking at the amount of detail that the two camera phone flagships can discern and reproduce.
Test 1: Hazy sun, pond scene, sign with gull(!)
Here's a full scene from our local park, centred on that very worn sign and perched gull out in the water in the middle distance, use your mouse or pointer to compare the results from both devices directly:
Download original file: Nokia 808 PureView | Nokia Lumia 1020
Discerning which photo is 'better' here is tricky - the 808's image looks too muted, while the 1020's looks too vivid, though I admit that the latter looks more attractive overall.
In order to discern 'oversampled' detail/'pureness', let's crop in to the centre of each shot and we can compare directly again, down at the pixel level:
I'd say that the above comparison demonstrates the main differences between the two PureView sensors/systems - there's very clear artificial sharpening on the 1020 image, while the 808's looks more 'natural'. However, whereas lesser (e.g. Samsung Galaxy S4, iPhone) camera phones sharpen to disguise the fact that they haven't got enough raw detail in the first place, it should be emphasised that the Lumia 1020 sensor has got the detail, it just chooses to exagerate it. Whether or not you prefer extra sharpening is entirely subjective, of course.
One aspect of the PureView 'system' is, of course, that 'lossless zoom', whereby each device can zoom up to 3x, down to the 1:1 pixel limit on the sensor, without having to 'make up' detail or interpolate (as other devices do). Here's the same scene, but using the PureView zoom in each case:
Download original file: Nokia 808 PureView | Nokia Lumia 1020
Results when at 3x zoom on the Nokia 808 always seemed a little disappointing - I think I get too used to the oversampling benefits and then, when they're taken away by virtue of being down at 1:1 on the sensor, suddenly I'm back in the realm of 'normal' sensors! The Lumia 1020's sensor and processing algorithms are better optimised to work well when zoomed in like this. Yes, the results are slightly artificial, but when there's no oversampling going on, a little processing is perhaps needed.
Let's look closer, though. Again, let's crop into the centre of each photo to see pixel-level differences:
Yes, the sharpening artefacts are slightly unplesant to the eye, but you can't deny that the text on the very worn sign is easier to read in the 1020's crop.
Test 2: Hazy sun, fairground ride in middle distance
Here we have a fairground ride about to be full assembled. Some pretty and colourful graphics though - the ride was a good 100 metres away, so this is an extreme test for each device. There's no full scene shot here, I was only interested in using the PureView zoom in each case:
Download original file: Nokia 808 PureView | Nokia Lumia 1020
Typically muted colours from the 808 again and rich, over-vivid colours from the 1020. Let's look closer with a crop. As before, use your mouse or pointer to compare them:
This is a fascinating example because the 1020's crop looks, immediately 'better', yet all of the improvements are due to the image processing in terms of colour and sharpening. Again, the jury's out - would you prefer ultra-natural, muted, or enhanced, colourful, eye-popping?
Test 3: Murky, overcast, cafe stall and shops beyond
With plenty of rain about and still spitting as I shot this, and with murky, indistinct light, here's an outdoor shot of a Costa cafe eating area and plenty of shop detail beyond it, again use your mouse or pointer to compare the results from both devices directly:
Download original file: Nokia 808 PureView | Nokia Lumia 1020
Both devices did a good job of capturing the scene, though both made it look lighter than it actually was! In order to discern 'oversampled' detail/'pureness', let's crop in to the centre of each shot and we can compare directly again, down at the pixel level:
A tough call again - I think I prefer the Nokia 808's 5MP version of the scene. My gut feel is that non-zoomed oversampled scenes like this are best on the 808 - the detail gathered with almost zero artefacts or colour distortions is staggering.
For interest sake, here's the same scene, but using the (3x) PureView zoom in each case:
Download original file: Nokia 808 PureView | Nokia Lumia 1020
Not a lot in it again, perhaps the 1020 has the edge in this 'zoomed' state? Again, let's crop into the centre of each photo to see pixel-level differences:
The same story here - some fine judged algorithms in the 1020 to bring out detail from 1:1 pixel data, though I do find the noise and artefacts just a little distracting. These zoomed shots are asking a lot of both devices though, and no other competing smartphone can get remotely close to the results here. The Samsung Galaxy S4 Zoom is the only phone which can not only match, but exceed the PureView devices in some of these tests, and that's because it's essentially a standalone camera with optical zoom, with a smartphone grafted into the body. Watch this space for tests of the Lumia 1020 and 808 against the S4 Zoom and other current smartphones in due course.
Test 4: Low light, dusk, street scene
Nokia has made a big thing out of the improvements in noise handling and detail in low light, so I started to fully exercise the devices by ramping down the light levels. This was just after sunset, though the light wasn't sufficient to budge the Lumia 1020 from its default ISO 100, interestingly. Here's the full scene, use your mouse or pointer to compare the results from both devices directly:
Download original file: Nokia 808 PureView | Nokia Lumia 1020
This is perhaps the Lumia 1020's first misstep in my handling. The shot from the 1020 has a blue 'cast' which simply wasn't there in real life. Note that these photos were all taken on full 'auto' - I dare say that some fiddling with white balance could have corrected it, had the 1020 been my only camera phone and had I had the desire, etc.
In order to discern 'oversampled' detail/'pureness', let's crop in to the centre of each shot and we can compare directly again, down at the pixel level:
The sharpening algorithms in the Lumia 1020's camera also come a cropper here, look at the brickwork of the house in these 5MP images - although the 1020 version looks immediately clearer, you don't have to examine the image for too long to realise that the clarity is a visual effect and not 'real'.
Here's the same scene, but using the PureView zoom in each case:
Download original file: Nokia 808 PureView | Nokia Lumia 1020
Not much in it, apart from the aforementioned colour cast, so let's crop into the centre of each photo to see pixel-level differences:
We're deep in image processing subjectivity here, but note a few things in the images. The central car looks much better in the 1020's version, with details nicely sharpened and made visible - look at the wheel trim in particular. However, the brickwork again looks too artificial in the 1020 image, it seems that sharpening algorithms don't like bricks! I'm also curious about the greenery behind the house - it has a different character in the 808 and 1020 versions - which do you prefer?
Test 5: Night time, street lights only
Here's the full scene, taken two hours after sunset, with just artificial lights - the Nokia 808 went to ISO 800, the Lumia 1020 seems to have set itself at ISO 4000 (which I didn't know was possible!). Use your mouse or pointer to compare the results from both devices directly:
Download original file: Nokia 808 PureView | Nokia Lumia 1020
As usual withe the OIS-equipped Lumia range, the longer shutter time in the 1020's case, plus the increased ISO means a much brighter image. In real life, I'd put the perceived brightness of the scene between the two, so no winner here. But in order to discern 'oversampled' detail/'pureness', let's crop in to the centre of each shot and we can compare directly again, down at the pixel level:
The Lumia 1020's image is revealed, at this crop level, as being pretty stunning. Yes, there's digital noise in the image and yes, it all looks brighter than it did in real life (to my eyes) but the level of detail and information that's being pulled out of such an extreme low light shot is very impressive.
PureView conclusions
There's a very definite theme running through my tests here. I can answer the starting question easily enough. "Can the Lumia 1020, with its slightly smaller (though BSI) sensor and image processing differences, deliver images that are as good as those from the existing Symbian-based Nokia 808?" On the whole, yes. Though there are huge subjectivity caveats here because of the differences in how images are processed. Remember from the postscript on one of my recent editorials:
With all these comments about image processing, it's a fair question to ask why any of it is needed? After all, surely a photo is a photo and the camera application should just present what it 'sees', pixel by pixel?
Sadly, things aren't quite that simple (they never are). Down at the pixel level, each only picks up (thanks to a conveniently placed filter) one colour, either Red, Green or Blue, along with a measure of illuminance. What then happens is that the camera's electronics combines the data from these raw coloured pixels using a Bayer filter, to produce a somewhat blocky, blotchy 'pixellated'(!) image. In addition, there's also random noise and uncertainty from the sensor pixels, especially in low light, where even individual pixels make a difference and where we're firmly in the realm of quantum mechanics.
So, we've got a pixellated image with sensor data that's a little random in places as well. Taking this and producing a photo that looks good to the human eye is the realm of 'image processing'. The aforementioned N8 and 808 partly achieved their results by having large sensors and thereby reducing the amount of 'after' processing needed. Smaller camera units such as that on the Lumia 920 and 925 require a little more help - mainly edge enhancement to restore fine detail that has been lost in the Bayer grid pixellation, plus noise reduction to smooth out the 'blockiness' and random pixel variances.
Thus, even the Nokia 808's 'natural' look still represents some processing, albeit very neutral in nature. Whereas a conscious effort has been made in handling the data from the Lumia 1020's sensor to emphasise colours and detail (through sharpening). Possibly this is response to industry trends and devices like the iPhone and Galaxy S4, both of which tend to emphasise the same things.
The end result though is that the Lumia 1020 does live up to the Nokia 808's heritage. It just interprets the (captured) world in a slightly different way. Nokia is betting that more people prefer the extra processed look than prefer the colder, more neutral colours and detail from the N8/808 generation. Do you think it's right? Comments welcome!
Notes / PS
1. I should also emphasise again, in light of any criticisms of images from both devices above, that it's all relative. These two smartphone cameras from Nokia are (the 'freak' S4 Zoom apart) significantly better than almost all the competition. It's true that, for macro shots, devices like the Galaxy S4 can excel, but for general purpose shots (and note that I haven't even tested the Xenon flash yet) the two PureView devices are effectively in a class of their own.
2. One benefit of the processing in the Lumia 1020 being 100% processor-based is that it's fairly trivial for Nokia to tweak the algorithms used in a simple update to its various camera applications.
3. All 1020 shots here were taken in the Nokia Pro Camera application. Watch this space for the next part of our Lumia 1020 review, focussing(!) on the stills camera in more detail.
4. Many thanks to Rafe for his help with managing the images and integrating them into the page in such interactive fashion!
See also
Review: Nokia Lumia 1020 (focus on 1020 as a smartphone)
Review: Nokia Lumia 1020 camera - photo capture
Windows Phone generations: Imaging differences between the Lumia 920 and 925
4 way flagship phone camera shootout: Nokia 808, Lumia 1020 and 925, Galaxy S4 Zoom