October 2012

Wednesday 10 October 2012

Camera megapixels: Why more isn't always better (Smartphones Unlocked)




It's time to forget megapixels as the measure of smartphone camera performance and pick a new yardstick.

Just days ago, Samsung announced the Samsung Galaxy S III, the global, quad-core, Android Ice Cream Sandwich successor to its best-selling smartphone ever, the Galaxy S II.

CNET readers' reactions were mixed, with several comments that the 8-megapixel camera didn't seem too hot.

Rumors of a 12-megapixel camera leading up to the announcement were partly to blame. It's no wonder that some felt that a perfectly good 8-megapixel spec was taking a step back, especially with the 16-megapixel shooter on the HTC Titan II out in the wild, and Nokia's 41-megapixel 808 PureView, a Mobile World Congress stunner.

Despite the fact that 8 megapixels is pretty standard for a high-end smartphone camera these days, one CNET reader described the Samsung Galaxy S III's camera as "so last year." Never mind that at least one high-end phone, like the Samsung Galaxy Nexus, still touts a 5-megapixel camera.

It isn't that 5-megapixel cameras can't be good, even better than phones with an 8-megapixel count lens; or that we're due for another bump along the megapixel scale. It's that to many shoppers, 5 megapixels just doesn't sound as good as 8, even if the camera produces terrific, knock-your-socks-off shots. And well, if 8 is good, then 12 is better.

The dirty secret lurking behind today's 8-megapixel yard stick for high-end status (and what any photography nut will tell you) is that the megapixel number alone is a poor way to predict photographic performance.

For instance, the original Samsung Focus took some lovely shots on its 5-megapixel camera, while the Motorola Droid Razr's 8-megapixel lens creates disappointing pictures. And the 5-megapixel camera on Apple's iPhone 4 beat out some 8-megapixel cameras on the market and delivered good low-light results.

Of course, that's not to say that bigger can't also be sometimes better. For instance, HTC's One X high-performance 8-megapixel smartphone camera boasts rapid shot-to-shot action, and its Titan II takes 16-megapixel shots of solid quality.

So what's the formula for fantastic photos? It involves an entire camera module that includes not just the size and material of the main camera lens, but also the light sensor behind it, the image processor, and the software that ties it all together.

Internet Giant Social Media
Enhanced by Zemanta

The fallacy of megapixel - megapixels camera


Internet Giant Social Media - The fallacy of megapixels

You can start to see that cramming more pixels onto a sensor may not be the best way to increase pixel resolution.

Jon Erensen, a Gartner analyst who has covered camera sensors, remembers when the cell phone industry jumped from 1-megapixel to 2-megapixel sensors.

"They would make the pixel sizes smaller [to fit in more pixels]," Erensen told me over the phone, "But keep the image sensor the same." Erensen similarly used the water analogy, this time swapping "buckets" for "wells."

What ended up happening is that the light would go into the well and hit the photo-sensitive part of the image sensor capturing the light. So if you make the wells smaller, the light has a harder time getting to the photo-sensitive part of the sensor. In the end, increased resolution wasn't worth very much. Noise increased.

The relationship between the number of pixels and the physical size of the sensor is why some 5-megapixel cameras can outperform some 8-megapixel cameras, and why we may not see, or want, a 12-megapixel camera on a smartphone. A slim smartphone limits the sensor size for one, and moving up the megapixel ladder without increasing the sensor size can unnecessarily degrade the photo quality by letting in less light than you could get with slightly fewer megapixels.

Then again, drastically shrunken pixel sizes aren't always the case when you increase your megapixels. HTC's Bjorn Kilburn, vice president of portfolio strategy, did share that the pixel size on the 16-megapixel Titan II measures 1.12 microns whereas it measures 1.4 microns on the One X's 8-megapixel camera. CNET's Josh Goldman points out that this is a small pixel size; however the take-away in terms of this discussion is that the two similar sizes mean that photo quality should be comparable at a pixel-by-pixel comparison.

Unfortunately, most smartphone-makers don't share granular detail about their camera components and sensor size, so until we test them, the quality is largely up in the air. Even if smartphone makers did release the details, I'm not sure how scrutable those specs would be to the majority of smartphone shoppers.

For more information on the interplay between megapixels and sensors, check out the excellent description in CNET's digital camera buying guide.

What about Nokia's 41-megapixel PureView?

Nokia's story behind its 808 PureView smartphone is really interesting. CNET Senior Editor Josh Goldman has written one of the best explanations of the Nokia 808 Pureview's 41-megapixel camera that I've seen. I strongly suggest you read it.

In the meantime, here's a short summary of what's going on.

Juha Alakarhu (pronounce his first name YOO-hah), is head of camera technologies at Nokia, where he works within the Smart Devices team. Alakarhu explained to me that although Nokia has engineered the PureView to capture up to 41-megapixels, most users will view photos as the 5-megapixels default.
Enhanced by Zemanta

Crytek Rasmus Hojengaard on Crysis 3 - Interview at Gameplanet New Zealand


Gameplanet: Is it necessary for Crysis to be synonymous with pushing the limits of hardware technology?

Rasmus Hojengaard I think we would probably be challenging some of Crytek's pillars if we didn't do that. The value you get from pushing these things is different now from what it was ten years ago. But it's part of our DNA. Even if people aren't conscious about doing it, it'll just happen anyway. There is benefit in it, and there needs to be companies that try to push things, because it sets an example for others to follow and benchmark from, and that rewards the whole industry.

One of the things we're trying to do now is push things we weren't pushing before, but that's not to say we wouldn't push things that haven't part of Crytek's DNA from the start.

Gameplanet: The original Crysis was fairly brutal on PC hardware, are you saying that won't occur with Crysis 3?

Crysis 3's senior creative director Rasmus Hojengaard. Hojengaard: The hardware and software today is a little more structured. DirectX 10 and 11 certainly give us possibilities to streamline things, and scale things in a way that wasn't possible in the original. So yes, we definitely want to have stuff in the game that you need the biggest beast to support fully, but we really want to have medium specifications that support maybe 85, 90 percent of that fidelity, that visual quality. So part of that is making sure that the artwork itself, the art direction itself, is great. That means that even if you can't use all the features, you're still going to get something that looks awesome. It's something that means a lot more to us in this game than it did in previous games.

The answer is yes, we want to do that, but also no, we don't want it to be an elitist type of game where you can't get a good experience without a nuclear power plant.

Gameplanet: The art direction appears pretty diverse, how did you manage to find seven different types of rainforest to depict?

Hojengaard: What we did was we sat down and researched the concepts of the rainforests, and there were a lot more than seven. There were Amazon-type rivers, misty mountain tops, canyons that maybe slope in a different way. Then we'd have to look at which ones potentially could support our gameplay formula, and which ones fit the architectural and geometrical layout of New York City, and that's how we picked the seven different diverse areas.

All this allows changes to gameplay that we didn't have before. We didn't have the ability to almost artificially approach this because we didn't have this setting of being underneath a dome that has its own ecosystem that would affect the world in a way that the sun doesn't. So it's a great marriage of high-tech and low-tech; the low-tech in mother nature taking effect, but it's controlled by high-tech. 

We believe the amount we're pushing New York will really resonate with people because it's going to be very clear that we're not cutting any corners by going back to New York again. We're actually putting it in a different context for you, so that people will go, "Wow, I have not seen this with Times Square", or, "Wow, I did not recognise that church, because I didn't realise those spikes aren't actually trees, they're church spires". So that kind of recognition is something that really gives us a unique visual language, and we haven't explored like that before. We've explored those elements before, but not combined them. And it's a lot of research work!

Gameplanet: So how does the nanodome make the interior grow faster?
Hojengaard: Basically this nanotechnology is somehow filtering the light to the surface from the sun, and it's manipulating it and changing the way that it works so that the ecosystem within the dome is very different to what you find outside. It basically just exaggerates what is already there. I can't give you an engineering explanation as to how this works; we'll probably explain in more detail how it works but we're not doing that just yet. But in theory what you get is a potentially naturally occurring thing on speed, if you will. Overgrowth that would normally take hundreds of years happening in 20 years.

This only affects vegetation, so it's not going to change animals, but the distribution of animals will change. Maybe you'll have a hundred frogs instead of five frogs, but it's important for us that it doesn't feel like a fantasy setting. It needs to have tactical elements and a tactical feel to it. So that's kind of how we're defining how far we can push this stuff.

Gameplanet: You were quick to point out the new water features, what kind of work have you done in this area?

Hojengaard: Well, the engine helps, but it's also about picking out very smart ways of creating water drama, like things that flow, and waterfalls. One side of that is technical, of course, but another is just being smart about how you process things. Let me give you an example – I know this isn't Crysis but when you have waterfalls in Skyrim, it's the same technology, so you need to have equal investment in the assets you use to produce this as well as the technology that drives it. So we're focussing on both things, and to be fair, we do have a lot of experience building tropical environments, so we know what we're doing. We're just iterating on that further, and pushing it even more. Later on in the year you'll see examples of that particular thing that are much more evident of what we've done exactly.

Gameplanet: We've seen the new addition to the suit: hacking. Can you give us an example of how this works later in the game?

Hojengaard: I can't give you a complete example, but what I can say though is that the idea is that the complexity should not be in the hacking itself, the complexity should be in what stuff you're hacking, and at what point. It's about the diversity of this and how you use it, rather than how you use each individual thing. We don't want to have a weird mini-game where you have to connect pipes and type a secret number, we want it to be the strategy of hacking. The suit is so technical it would obviously do that for you if you told it what to hack.

It's going to scale a lot, what you saw now is only a small-scale of what this will eventually grow to over the course of the game.

Gameplanet: Being able to shoot out of stealth is also new, what have you added to counteract this ability?

Hojengaard: Well we obviously need to balance this, but one of the ways to do so is to limit the number of arrows you get. We have all the ways to tweak this, one being countermeasures in the enemies' arsenal, another being the location itself, how enemy placement is, or a combination of that in conjunction with countermeasures then obviously how many of these arrows you're going to have to shoot. Plus finding the right balance at each point in the game.

Internet Giant Social Media
Enhanced by Zemanta