A Quick Analysis of EXIF data from iPad (3rd Gen) Camera Samples
Previously I posted about the EXIF data present in Apple’s untouched, straight-from-the-device iPhone 4S camera samples. Today, I saw that they had done the same thing again by posting original samples of images taken from the iPad. There’s still no original video sample (instead just a compressed one) so unfortunately there’s no telling whether A5X gets a better encoder than the one from A5. I wouldn’t hold my breath, however, as it’ll probably wind up being the same 24 Mbps H.264 baseline with 1 reference frame.
Like before, Apple has left the EXIF data intact on its samples, including the geotagging data for a surprising number. I think it’s interesting to just take a look.
Image 1 – IMG_1610.jpg
ExifTool Version Number : 8.68 File Name : IMG_1610.JPG Directory : /Volumes/Macintosh HDD/nerdtalker/Downloads File Size : 1518 kB File Modification Date/Time : 2012:03:08 12:44:45-07:00 File Permissions : rw-r--r-- File Type : JPEG MIME Type : image/jpeg Exif Byte Order : Big-endian (Motorola, MM) Make : Apple Camera Model Name : iPad Orientation : Rotate 180 X Resolution : 72 Y Resolution : 72 Resolution Unit : inches Software : 5.1 Modify Date : 2012:02:16 11:19:30 Y Cb Cr Positioning : Centered Exposure Time : 1/2160 F Number : 2.4 Exposure Program : Program AE ISO : 80 Exif Version : 0221 Date/Time Original : 2012:02:16 11:19:30 Create Date : 2012:02:16 11:19:30 Components Configuration : Y, Cb, Cr, - Shutter Speed Value : 1/2160 Aperture Value : 2.4 Brightness Value : 10.13573086 Metering Mode : Multi-segment Flash : No flash function Focal Length : 4.3 mm Subject Area : 1295 967 699 696 Flashpix Version : 0100 Color Space : sRGB Exif Image Width : 2592 Exif Image Height : 1936 Sensing Method : One-chip color area Exposure Mode : Auto White Balance : Auto Focal Length In 35mm Format : 35 mm Scene Capture Type : Standard Sharpness : Normal GPS Latitude Ref : North GPS Longitude Ref : West GPS Altitude Ref : Below Sea Level GPS Time Stamp : 19:19:30.6 GPS Img Direction Ref : True North GPS Img Direction : 267.2341772 Compression : JPEG (old-style) Thumbnail Offset : 904 Thumbnail Length : 10049 Image Width : 2592 Image Height : 1936 Encoding Process : Baseline DCT, Huffman coding Bits Per Sample : 8 Color Components : 3 Y Cb Cr Sub Sampling : YCbCr4:2:0 (2 2) Aperture : 2.4 GPS Altitude : 0 m Above Sea Level GPS Latitude : 38 deg 21' 21.60" N GPS Longitude : 123 deg 4' 1.20" W GPS Position : 38 deg 21' 21.60" N, 123 deg 4' 1.20" W Image Size : 2592x1936 Scale Factor To 35 mm Equivalent: 8.2 Shutter Speed : 1/2160 Thumbnail Image : (Binary data 10049 bytes, use -b option to extract) Circle Of Confusion : 0.004 mm Field Of View : 54.4 deg Focal Length : 4.3 mm (35 mm equivalent: 35.0 mm) Hyperfocal Distance : 2.08 m Light Value : 13.9
Like last time I’ve reproduced the entire EXIF output using the ever-awesome exiftool, or you can do the same thing (which uses it as a backend) on exifdata.com. It’s the same data just presented in a nicer fashion online (and with a link to google maps).
So a couple things immediately stand out which I’ve bolded. First, the model of the device reflects Apple’s new naming scheme, and reports simply “iPad.” There’s no 3 or “Late 2012″ or any other moniker, which should further confirm (if it’s even possible to more strongly confirm) that the name of the thing is literally just “iPad.” The software this iPad was running is iOS 5.1, which makes sense. Image size is 2592×1936 which works out to exactly 5.01 MP as well – the original iPhone 4 also produced images 2592×1936 in size, in fact, I wouldn’t be surprised to see the same OmniVision CMOS being shared between the 4 and iPad (3rd Gen).
Moving on we also see that the focal length and field of view reported in EXIF data are exactly the same as those from the iPhone 4S. This does seem to back the claim that the iPad 3rd Gen is indeed using the same optical system/module as the iPhone 4S, at least superficially. Further, this means that the two must be using the same size sensor to achieve the same field of view with the same 4.3mm focal length.
Update: This ended up being the case, as the iPad (3) uses the same CMOS as the iPhone 4 (as predicted), which is OmniVision’s OV5650. That sensor has 1/3.2″ size and 1.75µm pixels. Recall that the iPhone 4S uses Sony’s IMX145 sensor which is 1/3.2″ as well, but with 1.4µm pixels. Using the same optical format is what makes it possible for Apple to reuse the same 5P optical design between the iPhone 4S and iPad (3).
The image is recorded at ISO 80 (which is the lowest I’ve seen the 4S go as well) and at 100% crop looks pretty good, though I wouldn’t go so far as to say it’s mindblowing. There’s some definite noise visible in the sky and some noise-reduction which seems to battle it. Thankfully Apple still isn’t using a sharpening kernel, so there aren’t any halos around the flower’s petals. The lower part of the flower petal also has some oversaturation (100% white). The good part is that it seems to inherit the good optical qualities from the 4S system (same module, different CMOS I guess) and there’s minimal distortion or vignetting, two things that drive me absolutely crazy with most smartphone/tablet camera modules.
Image 2 – IMG_0470.png
Image 2 is pretty surreal. First off, it isn’t a jpg (which is why I’ve added the extensions to these headings) but is rather a png. I’m not sure whether this was just an honest mistake, but something is really weird here and the image obviously didn’t come in this extension or format directly from an iPad.
Pulling out the EXIF data raises more questions than it answers:
ExifTool Version Number : 8.68 File Name : IMG_0470.png Directory : /Volumes/Macintosh HDD/nerdtalker/Downloads File Size : 5.1 MB File Modification Date/Time : 2012:03:08 12:44:50-07:00 File Permissions : rw-r--r-- File Type : PNG MIME Type : image/png Image Width : 2592 Image Height : 1936 Bit Depth : 8 Color Type : RGB Compression : Deflate/Inflate Filter : Adaptive Interlace : Noninterlaced Pixels Per Unit X : 2835 Pixels Per Unit Y : 2835 Pixel Units : Meters Profile CMM Type : Lino Profile Version : 2.1.0 Profile Class : Display Device Profile Color Space Data : RGB Profile Connection Space : XYZ Profile Date Time : 1998:02:09 06:49:00 Profile File Signature : acsp Primary Platform : Microsoft Corporation CMM Flags : Not Embedded, Independent Device Manufacturer : IEC Device Model : sRGB Device Attributes : Reflective, Glossy, Positive, Color Rendering Intent : Perceptual Connection Space Illuminant : 0.9642 1 0.82491 Profile Creator : HP Profile ID : 0 Profile Copyright : Copyright (c) 1998 Hewlett-Packard Company Profile Description : sRGB IEC61966-2.1 Media White Point : 0.95045 1 1.08905 Media Black Point : 0 0 0 Red Matrix Column : 0.43607 0.22249 0.01392 Green Matrix Column : 0.38515 0.71687 0.09708 Blue Matrix Column : 0.14307 0.06061 0.7141 Device Mfg Desc : IEC http://www.iec.ch Device Model Desc : IEC 61966-2.1 Default RGB colour space - sRGB Viewing Cond Desc : Reference Viewing Condition in IEC61966-2.1 Viewing Cond Illuminant : 19.6445 20.3718 16.8089 Viewing Cond Surround : 3.92889 4.07439 3.36179 Viewing Cond Illuminant Type : D50 Luminance : 76.03647 80 87.12462 Measurement Observer : CIE 1931 Measurement Backing : 0 0 0 Measurement Geometry : Unknown (0) Measurement Flare : 0.999% Measurement Illuminant : D65 Technology : Cathode Ray Tube Display Red Tone Reproduction Curve : (Binary data 2060 bytes, use -b option to extract) Green Tone Reproduction Curve : (Binary data 2060 bytes, use -b option to extract) Blue Tone Reproduction Curve : (Binary data 2060 bytes, use -b option to extract) White Point X : 0.31269 White Point Y : 0.32899 Red X : 0.63999 Red Y : 0.33001 Green X : 0.3 Green Y : 0.6 Blue X : 0.15 Blue Y : 0.05999 Image Size : 2592x1936
The Microsoft and HP lines above are just talking about the ICC profile and header attached to the image. It’s an sRGB 1998 ICC profile, and there are CMM type fields (Lino) and others. I would suspect that this image was put through some software with a color management system which saved this information in the headers, and the person saving it chose PNG to not introduce more JPEG artifacts by re-encoding an already lossy-encoded image.
I’m not going to pretend to know exactly what went on here, but it’s fairly obvious the image didn’t come out of camera.app this way…
Image 3 – IMG_1190.jpg
Image 1190 is of a beached boat with “Point Reyes” marked on it. Thankfully this is a JPEG and not a PNG with everything stripped out.
Software : 5.1 Modify Date : 2012:02:14 16:14:55 Y Cb Cr Positioning : Centered Exposure Time : 1/1439 F Number : 2.4 Exposure Program : Program AE ISO : 80 Exif Version : 0221 Date/Time Original : 2012:02:14 16:14:55 Create Date : 2012:02:14 16:14:55 Components Configuration : Y, Cb, Cr, - Shutter Speed Value : 1/1439 Aperture Value : 2.4 Brightness Value : 9.728654971 Metering Mode : Multi-segment Flash : No flash function Focal Length : 4.3 mm Subject Area : 1295 967 699 696 Flashpix Version : 0100 Color Space : sRGB Exif Image Width : 2592 Exif Image Height : 1936 Sensing Method : One-chip color area ... GPS Altitude : 0 m Above Sea Level GPS Latitude : 38 deg 5' 51.60" N GPS Longitude : 122 deg 51' 3.00" W GPS Position : 38 deg 5' 51.60" N, 122 deg 51' 3.00" W Image Size : 2592x1936 Scale Factor To 35 mm Equivalent: 8.2 Shutter Speed : 1/1439 Thumbnail Image : (Binary data 9285 bytes, use -b option to extract) Circle Of Confusion : 0.004 mm Field Of View : 54.4 deg Focal Length : 4.3 mm (35 mm equivalent: 35.0 mm) Hyperfocal Distance : 2.08 m Light Value : 13.3
I didn’t paste the whole output since it’s a lot more of the same as before. Still an iPad running iOS 5.1, same focal length, field of view, all that good stuff. ISO is still 80 as well.
The location is Martinelli Park near Point Reyes (hence the marking), taken on February 14th, 2012. This is two days before the first image in the set, and very close to it (Bodega Bay and Point Reyes are essentially neighbors, at least based on Google Maps)…
Subjectively this image looks pretty decent, though there is definite blurring and loss of high spatial frequencies in the brown grass at left, though this is a challenging subject and great place to look for any camera to start making things a homogenous mess.
Image 4 – IMG_0561.jpg
Software : 5.1 Modify Date : 2012:02:09 12:03:06 Y Cb Cr Positioning : Centered Exposure Time : 1/1890 F Number : 2.4 Exposure Program : Program AE ISO : 80 Exif Version : 0221 Date/Time Original : 2012:02:09 12:03:06 Create Date : 2012:02:09 12:03:06 Components Configuration : Y, Cb, Cr, - Shutter Speed Value : 1/1890 Aperture Value : 2.4 Brightness Value : 10.12332838 GPS Altitude : 0 m Above Sea Level GPS Latitude : 34 deg 1' 14.40" N GPS Longitude : 118 deg 47' 10.80" W GPS Position : 34 deg 1' 14.40" N, 118 deg 47' 10.80" W Image Size : 2592x1936 Scale Factor To 35 mm Equivalent: 8.2 Shutter Speed : 1/1890 Thumbnail Image : (Binary data 8699 bytes, use -b option to extract) Circle Of Confusion : 0.004 mm Field Of View : 54.4 deg Focal Length : 4.3 mm (35 mm equivalent: 35.0 mm) Hyperfocal Distance : 2.08 m Light Value : 13.7
Again there’s no point in republishing all the EXIF data as much of it is the same as before and seems to be valid. ISO is 80 once again.
This photo interestingly enough was captured before all the others, on February 9th, 2012, five days before the third image above. Location this time is Paradise Cove right on the PCH. It’s interesting to me that all the images seem to be of or taken near beaches for some reason.
Of the images we can extract EXIF from, all are taken at ISO 80. This is probably no coincidence, as Apple probably wants to stay away from low light performance where basically everything struggles right now even with BSI CMOSes and fast F/2.2 or F/2.4 optics. I guess the beach is a logical choice of scenery to that end since there’s a lot of light (sand is reflective, after all). The photos are from different dates as well, which seems like a risky thing to do when trying to minimize the chances of some crazed tech-paparazzi catching an Apple engineer/exec with an unreleased iDevice snapping photos. With the iPhone 4S we saw a huge variety of different locations spanning California to Yosemite to Germany, whereas the iPad essentially gets a drive up the coast and a week of photos. Read into that what you may.
Obviously the last point is that the iPad 3rd gen’s camera is much improved from the almost universally-derided iPad 2 camera (which borrowed the iPod touch module). It’s interesting to see a move to using the same optical system as the 4S and likely the same CMOS as the 4, though it does make sense to maximize component cross-compatibility and drive up volume.