LONDON, England - Over the last few months, I have been carefully monitoring a bunch of photography and mobile websites and forums in order to gauge reactions to the Nokia N8′s camera specification and learn about what people were excited and concerned about. That left me with a mammoth list of questions – to which I present the answers here. Hopefully, alongside the two posts last week, it will give you a much clearer picture of the device’s capabilities and an understanding of our design decisions.
- Why doesn’t the Nokia N8 include a mechanical lens cover?
- The Nokia N8 uses the largest image sensor ever in a mobile, bigger than many digital compacts, does this mean the noise levels will be much lower than any other competitor product?
- Sharpness vs. detail, what’s the difference?
- What improvements have you made in response times in the camera?
- Why did you choose 12-megapixel rather than 8-megapixel or 5-megapixel?
- Is there a mechanical shutter in the N8’s camera?
- When shooting video, sometimes the scene seems distorted, almost jelly-like, why is that?
- Does the Nokia N8 include image and video stabilization?
- Why didn’t Nokia implement a continuous autofocus system for video?
- There appears to be an LED by the Xenon flash. Can that be used as a video light?
- The N8 uses Xenon flash for stills but why didn’t you include an LED solution for video?
- What’s the flash range?
- Given you have the largest sensor ever in a mobile phone I am expecting the noise to be extremely low especially in low light?
- At a glance, images from other devices tend to look sharper, why is that given the N8’s Carl Zeiss optics?
- Why did you choose a 28mm lens rather than the more typical 35mm?
- With the 28mm wide-angle optics, surely there’ll be distorted images which will mean I can’t capture portraits?
- The Nokia N8 provides only 2x digital zoom in stills and 3x zoom in video, can you explain why?
- Can you comment on the actual measured performance of the optics?
- How close will I be able to shoot still images?
- The Nokia N86 8MP used a variable aperture lens design but the Nokia N8 doesn’t, why?
- I’ve heard the sensor used in the Nokia N8 is the largest ever used in a mobile. Could you confirm the size?
- Can you please describe the audio recording capability?
- What resolution does the HDMI provide?
- How well does the AMOLED display perform in bright light?
- How does the shutter key feel?
- In previous products, Nokia has used a large area for autofocus. Why the change to a more conventional small centre spot area?
- Why didn’t you include features such as smile detection?
- Are the red-eye removal and face detection systems the same as in the Nokia N86 8MP?
- Can you comment on the recording capability in video?
- Nokia have tended to provide very small file sizes. Looking at the file sizes of some of the images on the web, it looks as though Nokia are continuing in this way?
- I noticed that in some images, even in bright light according to the EXIF information, the sensitivity appears to be slightly above 100 e.g. ISO 105. It seems strange to not use ISO 100 all of the time in bright light?
- Are you able to comment on your approach to auto white balance?
- Do you use any dynamic range booster with the Nokia N8?
- Will there be a RAW output mode for still images?
We take lens protection very seriously, while also needing to balance the capability of the entire product, not just the camera but also the size and design. Adding a lens cover would have increased the size of the raised area around the camera, especially with regard to thickness (something in the range of 2 – 2.5mm). There would also have been an impact to the footprint of this area as there are other components located here, such as the flash capacitor, loudspeaker and one of the stereo digital microphones.
The cover glass itself is multi-coated and is heavily scratch resistant. It passes Nokia’s tumble test and other associated durability tests. Furthermore, there are more and more applications becoming available which are integrated with the main camera, such as Facebook and augmented reality applications. These allow the camera to be launched from the application itself. A manual design would not allow this. You need to separately open the lens cover. An automatic lens cover would overcome that issue but such mechanisms thus far have required even larger footprints.
From a personal point of view, I am accustomed to wiping the lens in a circular motion with a clean soft item of clothing. The scratch-resistant properties allow this and the design makes it easy to do without it feeling a chore. I have seen many camera phones even with physical lens covers still showing finger prints. Granted, the risk is lower but there is no perfect solution, I’m afraid. One issue which can occur with mechanical mechanisms is a build-up of dust which collects in the recessed areas and because the opening is so small it’s very hard to clean. We find the designs such as that used in the Nokia N8 are much easier to simply wipe clean.
This is a complex area. As I said in my article last week, our target was to create a more natural reproduction of the original. From the still images through to video and the audio. There are over 300 different parameters we can tune. Many of these are interlinked so as you adjust one parameter, it has a knock-on effect with another and so on. Our belief was that we could create more natural images than have been possible before due to the unique combination of large image sensor and exclusive Carl Zeiss optics. We could have tuned the N8 with noise, detail or sharpness in mind only. In the end, we reduced artificial sharpening to very low levels yet have excellent natural detail, retain relatively low noise but which typically has a more film like look to it other than in very low light conditions.
Detail is how much information the image/video actually contains. On the other hand, sharpness in digital imaging is more about the processing that’s been used to create a perception that the image is sharper than the optics were able to resolve it – effectively fooling your eye to think it’s better than it is. It’s a technique used in virtually all mobile devices and many, many digital cameras. It’s the detail that’s important. If you like edges to have a crisp look to them you can create that in many image editing programs.
There is also a sharpening setting in the N8’s still camera, should you prefer this look. But you’re not actually creating any new information. The benefit of the new Carl Zeiss optics is that they naturally resolve very high levels of detail without the need for edge enhancement (sharpening). In fact, we’ve almost disabled the sharpening, it’s set so low. The result is no black or white lines around areas of high contrast as you will see in other broadly comparable devices, for example grass. I’ve seen grass from digital cameras with thin black edges. Grass is green.
If you prefer the high contrast/super sharp look, use the sharpness setting. You can also configure this setting as your preferred default setting when the camera starts.
The key differences you’ll note are as follows:
- Shutter lag is now down to only 150ms. The average human reaction time is 250ms, so this lag is hard for us to perceive.
- Viewfinder lag – the image you use to see on the viewfinder has, since the first devices, lagged behind the ‘live’ action by around 100ms. In the case of the N8, this is virtually zero. What this means is that – with the very fast shutter lag and almost zero viewfinder lag – it’s much easier to capture the precise moment.
- Autofocus time is super fast – typically focusing time is around 350ms (depends on subject distance, contrast and lighting levels).
- Despite the 12-megapixel image resolution, camera start-up time is slightly improved and shot-to-shot times are typically around just two seconds now.
- Around 90 per cent of usage of phone cameras is for still capture. To reflect this, we’re trying something new. The camera always starts in still capture mode. In most cases, where people want to grab a spontaneous moment, they will use stills. We therefore made still capture the default. Of course, there is just one touch of the display to switch between video and still capture.
This is an excellent question. We compete in an incredibly complex industry. We have to balance the beliefs of the majority with what we believe is right for our customers.
With the N8, we didn’t want to go backwards from the N86 8MP, only forwards [I’m referring to performance rather than megapixels], but we also need to be competitive in the industry. For these reasons, we concluded the only way forward was to do this properly and that meant using the biggest sensor ever in a mobile. This may come as a surprise to some, given Nokia’s early track record in mobile camera development, but it’s never been our intent to lead on megapixels. This is why we have consistently adopted larger pixels than those considered to be at the forefront. As an example of that, most manufactures have switched to using the potentially less sensitive 1.4 micron pixels in their 8 and 12-megapixel equipped devices. Whereas Nokia are still using the larger and therefore more sensitive 1.75 micron pixel sensors.
Until such time that there are sufficient improvements in smaller pixel sensors. We’ll be sticking with this approach for now, but sensor technology is constantly improving.
There are some benefits which we’ve taken advantage of with using this resolution sensor and one which you’ll be able to take advantage of:
- When recording video – due to the high resolution sensor – we take advantage of pixel binning to reduce noise. By combining multiple pixels and combining one new pixel you effectively filter out much of the noise that would normally otherwise need to be filtered out resulting in a reduction of resolution. In reasonable lighting conditions, we’re able to completely disable noise reduction for video as a result which allows us to retain maximum sharpness but with very low noise. I’ll come back to this point later when discussing digital zoom
- High performance digital zoom up to 3x in video with almost lossless performance (virtually no upscaling).
- When viewing images on a large screen at normal viewing magnifications, because with a higher resolution image you don’t zoom in to the image as far, any noise that may be visible is reduced, as the pixels of the monitor display more than one pixel of the image. The more pixels they account for, the noise is less apparent. Factoring this in, it’s theoretically possible to shoot in roughly half the light level with a 12-megapixel sensor vs. a 5-megapixel sensor with similar pixel size and characteristics. The appearance of noise will be roughly the same.
Longer term, we continue to debate and discuss all kinds of interesting views of possible directions we could take. I still see lots of opportunities.
We use a mechanical shutter for stills, so you won’t see motion distortion effects. However, in video you will see this in some situations as you do even with high-end DSLR’s with HD video. But for most situations I believe it will be OK. It’s a limitation of the readout speed of the sensor. I expect – at some point in the future with new sensors – this issue will be eliminated.
When shooting video, like all mobile phones and the vast majority of digital cameras including some digital SLR cameras, a rolling shutter is used. For various reasons, mechanical shutters can’t be used for video. Furthermore, usually due to bandwidth issues, it’s not possible to read each pixel at the same time. A rolling shutter overcomes this by allowing each pixel to be read in sequence. For example, starting with the top left pixel of the frame and then reading the pixels to the right until reaching the end of the row before moving to the left pixel on the next row down. So, as you move the device, pixels which are in the bottom right corner of the image are being read later in the movement path than those that were in the top left.
Smoothness is one of the most important aspects when filming video. Avoid sudden movements. Avoid moving at all, if you can. This isn’t as crazy as it may first sound. Next time you’re watching a movie, take notice of the camera moves. In most cases, you’ll see the camera doesn’t move at all. Instead, the director cuts to another view of the same subject/scene. If the camera does move they use Steadicams or a rig moving on rails with a dedicated operator just to move the rig. Of course, for a device which is always with you and fits in your pocket, such additional equipment and manpower is hardly practical. My recommendation is to concentrate on keeping everything smooth. Avoid sharp movements. If you can, think about what you’re going to film before you start filming. This will allow you to choose a framing and filming distance that works better for the entire scene rather than just the beginning.
The Nokia N8 includes video stabilization but not image stabilization. For now, it’s not possible to integrate optical image stabilization into such a small module (compared to digital cameras). Digital stabilization methods typically rely on at least two frames being combined to create one clearer one. If the subject moves, you can get alignment problems. These also don’t work with xenon flash. Having said that, if you prefer to use the xenon flash it has a firing duration of between 1/3,333 and 1/200,000 of a second! Which will freeze virtually any subject you’re likely to want to photograph.
Video stabilization works best in situations where you are trying to hold the device still or when the motion is more predictable. In situations where the movement is more erratic, it may be better to turn off video stabilization.
The Nokia N8 is our first product to use an Active Hyper Focal Distance system. We have used hyper focal distance before but in the case of the Nokia N8 where the HD video demands greater focus precision, we needed to improve the dependability.
Many mobile phones which include autofocus use driving systems which rely on friction to a certain degree. However, in some cases after the lens has been driven to the desired position, it can ‘slip’ causing a shift in focus.
For this reason, we use an ‘Active’ system where we constantly measure the position of the lens and readjust it to maintain the hyperfocal distance if required. The use of this system allows us to provide sharp and highly detailed videos from approximately 60cm to infinity, without the need for focus adjustment.
Other systems may provide benefits for shooting at closer distances but in our evaluation, we found this to be less dependable in the vast majority of recording situations which are typically at distances greater than 60cm. On this basis, we concluded that – at this stage – it is better to improve the dependability of the video performance across a wide range of conditions regardless of light and contrast levels and subject/scene changes.
However, we are continuing to investigate the opportunity of providing a focusing system which provides the best of both worlds for possible implementation in the future.
The LED next to the xenon flash is optimized for autofocus assistance. It’s amber-red in colour and therefore is not practical for use with video. It also serves as a privacy indicator and tally lamp when shooting video.
To get sufficient illumination from an LED solution over a usable distance today, you need to use dual LED. Current generation single LEDs provide little more than a tick box on a spec sheet. Unfortunately, we simply had insufficient space for Dual LED as well as Xenon, the area around the camera is packed very tightly already.
I know some of you will say you’d rather the device was a few mm’s thicker, etc. but it’s very important to balance the design against the functionality, otherwise people will say it’s ugly which will influence its appeal. But we we’re always listening to market feedback and therefore continue to try and develop the perfect solution. But for some, it’s just not possible. In a world of over 6.5 billion people that’s hardly surprising.
The actual shooting conditions can dramatically influence the effective flash range. For example, shooting in an average-sized home with light furnishings will provide an increased effective flash range as the walls and ceilings act as reflectors, therefore increasing the amount of light reaching the subject. In typical conditions, 3-3.5m is achievable. In open night scenes you may find this to be a little less. Indoors, in an average home, you may find 4.5-5m achievable. This is pretty impressive, considering we were able to make the flash module approximately 30 per cent smaller than the one used in the N82. The N8 has virtually the same flash power as the N82. However, the N8’s sensor is more sensitive and the noise reduction routines are much more sophisticated today. As a result of testing people’s preferences, we’ve decided to tune low light differently to the N82 and N86. Both of those products prioritized low noise at the expense of details. In the case of the N8, due to popular demand, we’ve prioritized detail over noise. Does this mean the N8’s images are noisy? No. No, what it means is that compared to the N86, noise is a little more visible but you gain a lot more detail and the images look natural. Compared to the N82, noise (when present) is much finer but the detail is significantly higher
We use a pre-flash metering system to calculate how much flash is required for the main exposure. This system helps prevent overexposure, for example, in close-up flash situations. In auto flash mode, the time between the small pre-flash and the main flash has been shortened in comparison with the N82 to provide faster shooting capability. To provide this faster shooting means the flash may not quite reach its full flash charge in time for the main exposure. However, in auto red-eye reduction mode the time delay between pre-flash and main flash is greater. The result is that that the flash will be 100 per cent charged by the time the main flash fires, resulting in a small additional amount of flash power. Up to 3.5-4m you probably won’t be able to see this difference. But if you’re shooting at longer distances, it may be worth selecting the red-eye reduction flash mode to grab a little additional flash range.
As mentioned earlier, there are over 300 different parameters to set and tune. Of course, it’s possible to use aggressive noise reduction to reduce noise, but this tends to reduce image detail. The combination of the new Carl Zeiss optics and large image sensor gave us the best platform we’ve ever had to develop and tune image and video quality. We used this to create what we believe is the closest challenge to digital cameras. We know that people want low noise, detail and vibrant (but still true to life) colour. Rather than optimizing for just one of these and compromising one or more of the others we believe we’ve created the best balance of these elements. But we’d love to hear your opinions on this.
We’re always listening and continuously trying to create even better performance. Generally, while you’ll see a little noise in the images, it’s pretty low. In low light we prioritized detail and fine noise. Many previous products including some very popular ones such as the N86 8MP and N82 offered relatively low noise but this ate into image detail. Now we have lower noise than the N82 but much more detail than both the N86 and N82 in low light. The noise levels in N86 are lower because we were using more noise reduction. In the case of the N8 we’ve reduced noise reduction significantly to maintain more natural detail. We’ve also been able to use the sensitivity to extend the range of the flash to around 4.5m+, depending on shooting conditions.
There’s a subtle but important difference between sharpness and detail. Perceived sharpness can be created through edge enhancement image processing. But – as I said before – we wanted images which were truer than ever to real life and therefore more natural than before. This meant reducing edge enhancement to a point where it’s almost disabled. There’s a number of reasons for this:
- You’ll be able to get some fantastic prints from this. Much more film like.
- You can set a higher sharpness setting if that’s your preference and the new user scene mode allows you to set this as your default setting amongst others.
- We’re finding more and more interest from those who have an increasing passion for photography. For these users, this reduced image processing (whilst not RAW) does provide much greater flexibility for the more advanced user. Over sharpened or too aggressive noise reduction significantly hampers such flexibility and creativity. Of course, we’d like to do a lot more in this regard, but that’s a topic for another day!
When we were originally developing the camera module for the Nokia N86 8MP, our aim was to create something which would be more suited to spontaneous, 24/7, modern photography. Given the wide range of environments these devices are used, it was our belief that the wide-angle capabilities of a 28mm lens would provide greater creativity but also greater convenience. Furthermore, there are a few additional benefits when using wide-angle lenses. For example, greater depth of field. We take full advantage of this when recording HD video.
No, the new Carl Zeiss optics provide extremely low levels of distortion (in the order of 0.1 per cent. Far, far lower than high-end compact digital cameras and without the need for angle of view depriving image processing. Generally speaking, telephoto lenses are preferred over wide-angle lenses for providing extreme close-up portraits, not just because of the perspective but also for allowing you to shoot further away from your subject. In this regard whilst perhaps not to quite the same degree, 35mm suffers similar constraints to 28mm. When shooting at normal shooting distances with the low geometric distortion, we believe it’s still possible to take natural shots of people.
In the case of stills, we capped the zoom at a point where we felt the image quality was still usable. But honestly speaking, going much beyond this only reduces image quality. Of course, though, there are on-device editing tools for cropping, etc. which you may prefer to use.
In video, it’s a slightly different story and this deserves further explanation for you to be able to get the most from it.
At the starting point (zero zoom), as mentioned earlier, we scale the resolution from the full-resolution sensor to the resolution needed for 720p video, 1280×720. Through binning, we’re able to reduce noise in video to such a level that noise reduction is actually disabled in normal lighting. This is great for all-round, daily use. Digital zoom between 1x and 2x is reasonably consistent. There is a little fall off as you approach 2x zoom. However, around this point, there is a mode switch. So, rather than the normal method, we reduce the resolution by the same factor and then scale according to the amount of zoom. There’s virtually no upscaling. However, when you extend slightly above this point, the scaling changes from binning to direct read out from the sensor to the scaling. The result is possibly a little visible noise in some situations, but there’s a significant increase in image detail.
We have a sweet spot around 2x -2.5x where the performance is at its best in bright light. The performance here is perhaps better than you would get with an equivalent optical zoom, because as you zoom, there would normally be a fall off in resolution – depending on the optical design, of course.
To explain a little more.
At 1x, we are using binning as do many products. The benefit we get as a result of the Carl Zeiss optics and large sensor is that we don’t need to add any noise reduction in good lighting in video so we retain a large amount of detail that is usual lost in the binning and noise reduction process.
The binning process is essential in handling the full frame sensor view down to the required resolution for 720p video at 25fps. The benefit of the binning process, though, is that it removes any noise that would be otherwise be present.
At 2x zoom, however, because we are only using a cropped section of the sensor, we effectively have more processing power available which allows us to process images with less detail loss, so the images at around 2.5x have more detail than at 1x. As the zoom then approaches 3x we then reach the limit of scaling. More than 3x would mean upscaling rather than 1:1 or downscaling, as we are doing up to this point. The downside at 2x above though is that we get a little more noise because we lose the benefit of the binning process to wipe out the noise.
In terms of stills, many camera optics (including dedicated cameras) often don’t resolve anywhere near their sensor resolution. For example, some 10-12-megapixel sensor equipped cameras can barely provide true 5-megapixel resolution. In the case of the Carl Zeiss optics we’re using here, the theoretical output resolution is so close to 12-megapixels that it’s not worth mentioning. However, in the real world after balancing the various multiple parameters it can come out a little different.
In the case of video, the limitation here is the binning process which most cameras do in video. In daylight situations, as commented earlier, we’re not using ANY noise reduction, yet in normal viewing I don’t expect you to see any noise either. The reason for this is that the binning process is effectively wiping out any noise we would have in the system allowing us to maintain as much detail as possible. So it’s the combination of the sensor and optics which allow this.
Zoom to around 2x zoom in video. The cropping and scaling process works a little differently resulting in perhaps a little visibility of noise but enhanced detail. This can also work well in low light depending on your preference and the actual conditions. It’s worth experimenting with to see what provides you with the best performance based on your own preferences.
All the details: Nokia Lumia 800
All about our stunning new smartphone.The specs; the price; the facts
Approximately 10cm from the subject in close-up scene mode. Around 20-30cm in auto.
We wanted to improve the performance of the optics but were constrained by the size of the module. We therefore traded the variable aperture for superior optics. We compensated for this trade-off by incorporating an ND filter to handle bright lighting conditions. The newer, more sensitive silicon used in the Nokia N8′s sensor completely compensates for the difference in the maximum aperture. (f/2.8 in the case of the Nokia N8 and f/2.4 in the case of the Nokia N86 8MP).
The 1/1.83” sensor used in the N8 uses 1.75 micron pixels which, with its 12-megapixel resolving capability, results in a sensor larger than most compact digital cameras. What’s the ISO range of the new sensor? We’ve capped the sensitivity to ISO800. The default setting is 100. There are 3 manual settings which relate to ISO 100, 400 and 800.
The N8 records audio in stereo and encodes it using the AAC audio format at 128kbps with 48kHz sampling, it sounds really great! What really makes the difference is the placement of the mics and the new algorithms which define how we capture the sound. All of these elements combine to provide a far greater sense of the original environments ambience. One of the benefits is the ability to handle harsh environments where the sound volume is very high and relatively constant. I’ve seen/heard videos tested in a bar environment where there was live music. In this particular example, I was amazed at how the N8 handled the range of sounds. Firstly, the music was clear and free of distortion and second, there was someone behind the person filming who asked them what they wanted to drink and that was also perfectly clear!
This is made possible through the placement of the mics. One of the stereo digital mics is at the front of the product (just below the display) and the 2nd is placed at the back just by the camera. Another situation this works well for is when recording video with narration. You can speak normally. Your speech will be crystal clear whilst still being able to capture virtually all of the different audio elements around you. It works really well. Finally, we’ve introduced some new wind reduction algorithms. These won’t eliminate wind sound altogether, depending on the actual conditions, but will make a significant improvement.
The digital mics we use are incredibly sensitive. It’s worth spending a little time perfecting a comfortable and stable grip on the N8 which allows you to start and stop recording the video without moving your hands across the surface of the handset, etc. as those mics can pick up such sounds. But with a little practice this isn’t a problem. You can either start and stop video with the dedicated capture key or you can use a touchscreen-based start key. After starting recording, touch based keys appear for pause and stop. Personally I tend to use the dedicated capture key. But if you’re using the touch screen zoom control, you may find it better to use all touch screen controls. We’ll come back to zoom later…
Photos and videos are presented over HDMI at 1280×720 resolution. Approximately four times the resolution of previous TV connection options. It makes an incredible difference. I’ve seen comments from people expressing the view that if I didn’t show my images before like this why am I going to do it now? I think it’s down to one key aspect. In the past, your video and images still looked much better when viewed on your PC. Now that situation is completely different. Now I can connect my Nokia N8 to my HD TV which is the largest screen in the house in the most comfortable room in the house for everyone to enjoy the experience. I can even use a Bluetooth keyboard and/or mouse to make it even more convenient. Furthermore, many people don’t have their PCs connected to the TV, so this is the easiest and best way to watch that content in all its glory on a huge screen. Not to mention enjoy the rich audio too. I’ve personally found myself doing this much more when visiting friends now too, because it simply works so well.
In my own personal experience, I have had no problems shooting video and stills in even very bright conditions. Many of you have asked, “But how have you done that when OLED has been not-so-usable in bright light?“ There are two key contributing elements here. First, this newer display is much brighter than previous OLED displays. In certain situations, it’s up to almost 60 per cent brighter! And second, there is an anti-reflection film applied to the display.
OK, again a personal view on this one. Two comments. 1. Some of those who have had the opportunity to handle the N8 have commented that this has the best feel thus far of any mobile. Specifically it’s easy to feel a click relating to half pressure [Autofocus/exposure], as well as the full press, where the amount of effort has been reduced but not to the point that you accidentally take images. 2. There is also a touch based shutter key which, for some, may be a better means of avoiding camera shake in low light without flash. I personally prefer to use the physical shutter key, but I’ll let you decide, once you’re able to get your hands on one.
We had a number of requests for this. I have to say the jury is out right now what’s the best system to use and we’re discussing options for the future too. Having been quite used to the wide area system, which I think has some benefits to offer, over the months testing the Nokia N8, I’ve become a fan again of this system for its speed and ability to pinpoint small objects. Of course, you can lock focus and exposure with a half press. The autoexposure is also heavily biased to this small area so if you do lock focus/exposure, the results are more predictable.
I’m not really a fan of some of the more gimmicky features and rather we focused our efforts on providing better pictures more of the time, which is exactly what we’ve done.
They’re similar. However, there has been further development in both of these areas. Red-eye removal works in both auto and red-eye reduction modes. In red-eye reduction mode, it runs a more intensive version. This may provide a small increase in shot-to-shot time. The face detection system is now linked to the flash system. It compensates automatically now if there is a face detected which is backlit but within flash range.
Update (August 2010) : It was our intent to have a ‘light’ version of the red-eye removal system running in Auto flash mode. However, due to a small number of occasions where the performance was outside our target performance we have decided to continue to develop this for inclusion in a sw update. The ‘full’ red-eye removal system remains as planned in the red-eye reduction flash mode and in the vast majority of situations works extremely well.
Initially, the Nokia N8 will record and encode video as follows: 1280×720 HD @25fps using H.264 codec with up to 12mpbs bit rate. Stereo audio is encoded using 128kbps bit rate, using AAC codec sampled at 48kHz. We will be looking for opportunities to further improve the video performance in later software updates. Can’t say anymore than that at this stage ;-)
There are additional options for VGA 30fps and a mobile-sharing-friendly 3gp optimized setting. In addition to these, you can also reduce the frame rate to either 15fps or 5fps. This is to improve low light video recording capability when you use low light (15fps) and night video (5fps) modes.
This has been a contentious topic many times in the past and I can say this. There has been no product that I have worked on that has compression-based artifacts. Very recently, I completed verification of this. I captured images in various conditions using various JPEG compression settings up to and including 100 per cent JPEG quality setting, (which gives massive files, BTW!) Myself and colleagues then reviewed this at magnifications up to 300 per cent and concluded that based on visual inspection there was not even the slightest justification for changing from the current JPEG quality setting. The principle reasons for this are, I believe, as follows: 1. The human eye can’t see such differences. 2. Because we’re providing images with relatively low noise, our image compression is more efficient.
I noticed that in some images, even in bright light according to the EXIF information, the sensitivity appears to be slightly above 100 e.g. ISO 105. It seems strange to not use ISO 100 all of the time in bright light?
On the surface, we understand this behavior may seem slightly out of the ordinary. However, it’s perfectly normal and is a result of some new image processing algorithms we’re using for the first time with the Nokia N8. There is no adverse effect on noise with such small sensitivity adjustments.
We have actively decided not to follow the technically correct approach here, which some purists won’t like. For example, when at the ends of the colour temperature spectrum, we don’t try to make things technically perfect. The result would be something that doesn’t match your ‘mind’s eye’. For these products, we believe this approach is much more important than being technically correct. However, there are a few manual white balance presets if you prefer to override the system still.
An example of this is when shooting under candlelight. The warm ambience created by the candle light, you want to retain. So we recognise these situations and then adjust the AWB accordingly. A red rose on a cloudy day is another example. You’ll remember it more vividly than reality.
We do apply a little to preserve details in the shadows as much as possible. However, this is effectively a gain control so if you apply too much, you increase the appearance of noise in those areas too.
The need is understood. However, sorry to disappoint but the N8 will not include this capability. We need JPEG+RAW mode to make this work, so people can still easily send images direct from the device. RAW on it’s own is no good IMHO in such devices. This is a lot more complex than it seems on the surface because you need to take into account the whole ecosystem with devices such as phones. You can’t just presume everyone has Adobe Photoshop. We would also need to provide some additional manual controls for you to really benefit from RAW.