I’ve written about the benefits of oversampling to improve image quality: cameras today ideally would be be utilizing ultra high resolution sensors and delivering 3/4, 2/3 and 1/2 linear resolution images all but free of digital artifacts. For example, a full-frame Nikon or Canon 96 megapixel sensor delivering 72 or 48 or 36 megapixel raw images.
We can put the oversampling theory to a test with camera phones.
iPhone 5s: 3264 X 2448, 8.0 megapixels, ~2.6 MB JPEG
Nokia Lumia 1020: 7136 X 5360, 38.2 megapixels, ~11.6 MB JPEG
I don’t have a Nokia Lumia 1020 and my iPhone 5s is on order, but over at dpreview.com is a studio comparison of the iPhone 5s versus the Nokia Lumia 1020. I was curious to compare the image quality (I have asked Nokia for a review camera, but I don’t know if they will provide one).
It’s a shame that raw format is not available from both cameras, though Martin D points out that Digital Negative for iPhone / iPad offers a DNG capability.
Examining each JPEG:
- The Nokia image shows poor contrast on the left side. Whether this is lighting I cannot say, but it looks suspiciously like lighting that changed, perhaps due to camera distance to the target. Or it might just be a camera phone weakness.
- Color and contrast are better with the iPhone 5s than the Nokia 1020.
- The Nokia Lumia 1020 has more than 4X as many megapixels which is more than 2X the resolving power (in pure numbers terms). It is obviously superior in detail resolution but only in the central 3/4 of the frame; it degrades markedly outside that area, so much so that it becomes inferior to the iPhone from about frame edges to corners.
- Downsampling the Nokia Lumia 1020 image to the same resolution as the iPhone 5s image produces markedly superior definition that trounces the iPhone, but only in the central 3/4 of the frame.
- While strong centrally, the Nokia Lumia 1020 has blurred and smeared corners. The iPhone 5s offers consistent quality corner to corner, but much lower peak sharpness due to its modest 8 megapixel sensor resolution that produces moiré and other unpleasant digital artifacts.
- Both cameras show pincushion distortion.
A small crop (fair use) from from the dpreview test shots is shown below, the Nokia Lumia 1020 image being downsampled to match the iPhone 5s resolution. Observe that the Nokia Lumia 1020 offers a clean and crisp image on the concentric rings. It is a strong demonstration of the benefits of downsampling a high resolution sensor to a lower resolution: the digital artifacts disappear. With the iPhone 5s, very strong moiré effects develop, and the color-on-color areas fare badly.
I printed both images at full resolution to a 13 X 19 inch print, using the Canon PIXMA Pro-100 on Semi Glossy paper.
My wife immediately gravitated to the iPhone 5s image because of its superior contrast and color. I am in agreement, except that the Nokia 1020 image shows considerably better definition over the central area with very fine lines and text crisply rendered, as shown above.
What is quite impressive: the 13 X 19 print quality I will summarize as “99% of the public will be dazzled”. While camera phone usability is horrible (no grip, no control over exposure, no raw, no nothin'), the market for low end cameras is clearly doomed, with sales in 2013 already tanked and that part of the market to be obliterated by camera phones within a year.
The open question is whether the contrast and color seen with the Nokia Lumia 1020 are due to the test conditions, or whether this is a general weakness, meaning a strength in favor of the iPhone. Even after making some adjustments, the iPhone image looks better overall. And since most camera phone shooters are not going to be doing Levels and Curves and sharpening in Photoshop, it’s dubious whether such adjustments are 'proper' for comparing. Bottom line: in spite of the lower resolution, the iPhone 5s camera is going to make a lot of people very, very happy.