Print 24 comment(s) - last by RU482.. on Jul 10 at 2:28 PM

Rumored resolutions of camera images leak, along with price

Nokia Oyj.'s (HEX:NOK1V) fresh flagship phone, the Lumia 1020 is set to be unveiled Thursday, but more specs are already trickling out, following the phone's inadvertent outing via Microsoft Corp. (MSFT).

Here's a run-down on what we know about the phone's camera, which is expected to be the best smartphone camera available -- rivaling all but the best point and shoots (other than lacking an optical zoom).
  • 1/1.2" image sensor
  • Dual flash: LED + Xenon
  • OIS (optical image stabilization) via floating lens*
  • 38 megapixel (4:3); 32 MP (16:9); and 5 MP (16:9) (with 7 pixels per "superpixel")
  • F2.2 aperture (some sources say this may be adjustable)
  • Manual internal shutter
  • 2 GB DRAM
  • 32 GB NAND Flash Storage
  • NFC
  • Camera grip with built in battery ($70 USD) (capacity unknown)
  • Wireless charging via backplate*
*This first popped up in the Lumia 920, and is also found in the 925/928.

Nokia Lumia 1020
The Nokia Lumia 1020 [Image Source: WPCentral]

The onboard "Amber" skin, which is Nokia exclusive will provide:
  • FM radio support
  • Flip to silence
  • Pro Camera app (control ISO, white balance, shutter speed, flash, manual focus)
The Pro Camera app is expected to stand in for the more blasé standard Windows Phone Camera app, although automatic settings will be available by default.

The phone will reportedly go on sale at Microsoft Stores on July 22, unlocked, at a price of $602 USD.  The Microsoft stores reportedly will carry three varieties -- yellow, white, and black.  AT&T, Inc. (T) is expected to be the initial carrier for the (locked) version of the device, which is expected to feature additional colors (like cyan).

International release prices and dates are unknown, at this point.

Source: WPCentral

Comments     Threshold

This article is over a month old, voting and posting comments is disabled

RE: I see to recall
By Solandri on 7/9/2013 5:12:42 PM , Rating: 3
It wasn't that long ago people on this site were saying that pixel counts of more than 8 Mp was not able to be supported by optics that can fit into a smartphone (and some saying for anything more than 5 Mp). I disagreed back then as I didn't believe that 5 or 8 was really the limit.

It always amazes me that there is such an outcry when new "limits" are first reached then a while later when those "limits" are breached by a factor of 5 or 8 or more there is no comment at all.

No limits have been broken. The iPhone's lens is a 4.1mm focal length f/2.4, so it's 1.71 mm wide.

The Rayleigh criterion for a 1.71mm diameter lens in the red spectrum (700 nm) is 0.0286 degrees. That's the smallest object you can resolve using that lens. It gives a view equivalent to a 33mm lens (in the 35mm format), which corresponds to a 57x40 degree field of view. So the maximum resolution it supports is 1999x1398 in a 1.5:1 aspect ratio, or 2.8 MP.

The Bayer filter means only one pixel in 4 is red, so the camera's 8 MP is effectively capturing only 2 MP of red image data, which is less than the 2.8 MP limit I just calculated. The extra "data" bumping it up to 8 MP is "made up" by the Bayer filter processing algorithm. Unless they go with a bigger lens or a wider field of view, the camera simply can't resolve more than about 8-10 MP of data (counting each color pixel as separate). You'll just be capturing two blurry pixels instead of one sharp one. You can see this if you compare a cell phone pic with a DSLR pic at 100%. Because more of the data is "made up" by the Bayer algorithm in the cell phone pic, it looks blurrier than the DSLR pic where adjacent pixels are getting truly different optical data.

I haven't seen specs on the Lumina 1020. But its predecessor the 808 uses a 8.02mm f/2.4 lens, which is 3.34mm across - nearly twice as wide as the iPhone's. It has an angular resolution limit of 0.0146 degrees. Its field of view is a 26mm equivalent, or 69.4x49.6 degrees. That puts its maximum capture resolution at 4737x3386 pixels, or 16 MP.

The 41 MP sensor means about 10.2 MP of red data is captured, which again is less than the 16 MP theoretical limit. Again, no limits have been broken here.

The reason cell phone cameras are getting close to these limits before DSLRs is because it's a helluva lot easier to grind a 2mm wide lens to an optically perfect shape, than it is to grind a 77mm wide professional lens to a perfect shape.

RE: I see to recall
By Shadowself on 7/9/2013 6:03:24 PM , Rating: 2
First, I wasn't referring specifically to the iPhone. If I recall correctly those old discussions actually started with another brand of phone, but maybe it was the iPhone.

Second, you've actually supported my point. The real pixel support of this optics is likely for a true 16 Mp (at best) not 38 Mp camera.

The purpose of a Bayer array with one blue, one red and two greens per four pixel set is that the software for the focal plane array can interpolate for the green and red over the blue pixel, similarly for the green and blue over the red pixel and finally for for the blue and green over the red pixel. This way you can get a synthetic red, green, and blue for every single pixel in the array. In order to do that the airy disk must be approximately the size of an individual pixel or smaller. Disks, as you describe, that cover all four pixels, or are larger than that, do not give independent information on each individual pixel.

I'm sure the images that are exported from this camera are the full "38 Mp" implying to the user that they have 38 Mp of independent pieces of pixel information and not the approximately 16 Mp you mention. It is *possible* to do lots of interpolations -- especially if HDR is implemented -- as I mentioned before (especially if you can play the math against the color versus quantum efficiency curves of the focal plane array), but the amount of independent information across each individual pixel is minimal at best.

In theory you can push a 1k x 1k array to about 30k x 30k, but this requires nearly a perfect set of interrelated information:
just the right amount and angular rate of motion and jitter
the optimum cross correlation between the quantum efficiencies as various color bands
the maximum number of independent frames of information observing static or nearly static targets
and so on and so on.

I doubt very much any of this is what Noka is claiming with the phone camera.

Also these calculations assume several not realistic implementations (not realistic in low cost optics):
nearly infinitely thin lenses,
fully achromatic optics,
aspheric lenses,
zero coma on the FPA,
Each of these further degrades the imagery.

Finally, I wasn't talking about physical limits being broken. There are lots of fun things that can be done with multiple rapid frames, FPA jitter, inter frame interpolation (e.g., in HDR mode), multi frame super resolution effects and such. However, implementing those in a phone seems extremely unlikely. If there is a physical limit to discuss it is about the airy disk of a given optics being on the order of -- or smaller than -- a single, physical pixel. When that limit is crossed, either the number of pixels is excessive or a LOT of fancy footwork in the software must be done to recover each modicum of different information obtained.

Thus I'm still at a loss as to why a 38 Mp camera is included unless it is just for bragging rights.

"We can't expect users to use common sense. That would eliminate the need for all sorts of legislation, committees, oversight and lawyers." -- Christopher Jennings

Copyright 2016 DailyTech LLC. - RSS Feed | Advertise | About Us | Ethics | FAQ | Terms, Conditions & Privacy Information | Kristopher Kubicki