Why aren't devices with two cameras more popular?

Question

I mean a device that has two sensor / lens combinations.

I can see several advantages:

• Capturing photos with a higher dynamic range.

• Focus bracketing.

• Capturing multiple photos and merging them to reduce noise (super-resolution), like the excellent iPhone app Cortex Camera, but without needing the user to hold the camera steady for a while, and hoping nothing moves in the scene.

• One of the cameras can be made without a color filter array, to improve low-light performance.

• Two smaller sensors can be thinner than a single bigger one. For example, if you were to increase the sensor size in the iPhone 5s, the camera module would probably not fit in the thickness of the phone. Whereas you could easily put two cameras side by side. This also applies to point-and-shoot cameras with larger sensors, which are hard to fit in your pocket. You could again have two sensors in a pocketable camera.

• Both lenses could be pointed outwards to capture a wider field of view.

• At larger sensor sizes, it would be cheaper to make two smaller sensors rather than one big sensor. Because a defect in one of the small sensors means that you throw away just that sensors, while you can use the other one, while a defect in a bigger sensor means that you have to throw away the entire sensor. So you get better yields, and therefore lower prices.

• If each sensor can capture video at 120FPS, then having two sensors permits video capture at 240FPS.

I can see many advantages with having two sensors. Why aren't such devices more common?

I don't think it will add to the cost significantly. For example, the iPhone 5s cameras are estimated to cost only $13, and that's for both the front- and rear-facing cameras together. Even if it cost $50, that would be a small matter on a device costing $400+, like premium compact cameras.

Answer 1

While a multi-sensor setup might be nice to fancy about, real people (even when genuinely interested, like you) are much more likely to spend their money on an iPhone instead of Fuji FinePix Real 3D. Little interest in market for already existing devices means less incentive for engineering more of them.

From costs side, even if adding another sensor would be only $50 (there's a significant difference between factory-installment cost and cost of final product), a competitor could use this $50 for significantly improving optics; it's going to be $100+ on a double-sensor camera to catch up with that, where the optics have to be added and synchronized to each other. And while hardware might be cheap, the need to redesign camera interface (both hardware and software) for a small market is costly.

Most of the applications listed in the question are impossible to implement straightforward due to parallax, so some stitching (with resulting artifacts) should happen in software, not going to look good in reviews. 3D was a hype in consumer electronics a few years ago, but inconvenience to user and uproar of mobile tech has pushed that back to "toy technology" status (where it IMHO belongs).

To address some of the ideas individually,

• higher dynamic range, focus bracketing, noise reduction, panoramas - already (with some limitations) available using multiple shots from single sensor;
• panorama cameras, high FPS video devices and enhanced dynamic range cameras (Fuji again, SuperCCD) have been around, loved by a few enthusiasts, but not wildly popular among general public;
• skipping the second sensor (and its optics) entirely is even cheaper than having to throw it away only sometimes;
• double FPS means double processing power and data throughput.

Also, two cameras consume double energy.

So, it comes down to

• nothing groundbreaking
• with its drawbacks
• for a significant cost
• appealing to few enthusiasts.

Answer 2

The fundamental problem is that if you have two cameras, they're pointing at two different things. That's just geometry: a line perpendicular to a plane intersects the plane at one point, not two. Not in Euclidian space or anything like that.

Now, granted, the two-camera approach works out pretty well for Homo Sapiens, and it gives you a super-realistic 3D effect, but most photographers and videographers prize the clarity of a 2-D image more (James Cameron notwithstanding) and it's easier to make use of such an image.

And it's not that you couldn't write software to combine two images, but it would be a real challenge to write software that combines two images with really high quality under all or most photography circumstances... using the limited processing power available in a mobile device, in a manner that doesn't interrupt your photo-shoot with a 10-second processing delay.

There are other engineering challenges too. For instance:

If each sensor can capture video at 120FPS, then having two sensors permits video capture at 240FPS.

Not so fast. You could stagger the captures from one sensor to another with enough engineering, but you'd also need a faster bus from the sensors to memory and the encoding hardware. And, of course, both cameras are pointing at subtly different pictures, again, so you might get some freaky flickering between frames.

Both lenses could be pointed outwards to capture a wider field of view.

That would kind of make it hard for them to be used to increase each other's resolution in your Cortex Camera clone, wouldn't it?

And finally, few people actually want what you're asking for. People who actually want lower noise from their camera? They buy a dedicated camera with a bigger lens putting more light on a bigger sensor. Professionals who have a need for super-high-FPS shooting? They shell out the big bucks for a capable video device. James Cameron wants to shoot 3D? Likewise.

The iPhone is not marketed at people who use these capabilities, yet developing for them is an expensive endeavor involving Apple paying many engineers lots of money. It's not even clear that such an investment would break even. Apple would rather develop software that increases their ability to sell cell phones to more people. If the iPhone ever gets two cameras on the same side, it will be a way to record 3D video.

Answer 3

The fundamental problem is lack of mass produced chips to do the processing required for the things you're requesting.

Solving the parallax problem is software is possible, especially when the error is small, as is the case with cameras a few mm apart. The problem is phones are made of generic mass produced consumer chips that cost a few dollars each. You state the iPhone camera costs less than $13, say it's $10. Then including an extra camera would cost $10. But then the chip to merge the images correcting for parallax might cost an awful lot more than that because you would have to amortize the cost of developing the chip. And if you decided to use a general purpose CPU for that, well there goes your 240fps and no-one wants to drain the phone battery for the sake of doubling the field of view or dynamic range.

The iPhone photographer isn't discerning enough to bear the cost. And those who are will be using bridge cameras with larger lenses that can't be placed a few mm apart, so the parallax problem comes back with a vengence, never mind the fact that large lenses and sensors are much more expensive!

Answer 4

You are comparing apples and oranges. The sensor on the iPhone 5 isn't the same sensor that is going to be in a $400 point and shoot. Also, shooting from multiple sensors isn't as simple as you think it is. There is a slight difference in angle from the two different points of view (this is why we can see in 3d) that would render it unsuitable for most uses.

Most of the things you mention as advantages could be accomplished more easily, cheaply and effectively by simply increasing the size of the sensor by the same amount.