Software (contd.) - RAW conversion
As is normal in our digital SLR reviews I like to compare the supplied RAW conversion software, any optional manufacturer RAW conversion software and some third party RAW converter. There's been quite a lot of talk recently about Nikon's move to encrypt the recorded white balance in RAW files on its newer digital SLR's. We won't get into that debate here but you should be aware that of the third party converters here only Bibble Pro decodes the white balance, Adobe Camera RAW can not.
- JPEG - Large/Fine, Default settings
- PictureProject - Nikon PictureProject 1.1
- Capture - Nikon Capture 4.2.1
- Adobe Camera RAW - Adobe Camera RAW 3.1 (Photoshop CS2) *1
- Bibble Pro - Bibble Pro 4.2.6
*1 Because of the encryption issue ACR can not read the recorded white balance from D2X RAW files, in this case we used the eyedropper to achieve a manual white balance
Place your mouse over the label below the image to see the GretagMacbeth ColorChecker chart converted using each RAW converter. There's absolutely no difference in color reproduction between PictureProject and Nikon Capture, this can probably be attributed to PictureProject using the image processing 'engine' from Nikon Capture to perform its RAW conversion. Between the two Nikon RAW converters and JPEG straight from the camera there's a slight difference in contrast but not actual color.
|Adobe Camera RAW||Bibble Pro 4.2.6|
Sharpness and Detail
The most detailed results from PictureProject and Nikon Capture closely followed by Adobe Camera RAW and in-camera JPEG. The image with the least immediate detail is that from Bibble Pro, although some tweaking of the sharpening setting would probably draw a little more texture detail out of the watch face. There's a tiny amount of color moire visible down the arc of the edge of the watch in all but the Adobe Camera RAW image.
Resolution (double distance)
Because of the D2X's twelve megapixels we need to shoot our chart from double distance. This means that the numbers indicated on the chart must be multiplied by 200 to return the resolution in LPH (Lines per Pixel Height). Interesting results too, both PictureProject and Nikon Capture extend visible detail further than a JPEG although actual measurable maximum resolution is about the same at 2400 LPH. Both Adobe Camera RAW and Bibble Pro exhibit some noticeable moire artifacts at these high resolutions.
|JPEG from camera||PictureProject|
|Nikon Capture||Adobe Camera RAW|
RAW converters performance compared
One thing which became apparent when using these different RAW converters was that they performed very differently and made different requirements of the computer hardware. We devised a simple test to allow us to measure and time the conversion of two RAW images through each converter.
- Start RAW converter
- Open two RAW images (timed)
- Note memory usage with images displayed
- Save image one as JPEG (timed)
- Save image two as TIFF (timed)
The computer used for this test had the following specifications: AMD Athlon 64 3800+ (2.4 Ghz), 1.0 GB RAM, Seagate Barracuda 7200.7 SATA 160 GB HDD, Windows XP service pack 2. All other applications where shutdown to ensure maximum possible memory and CPU cycles. AMD Cool 'n' Quiet disabled.
|Nikon Capture 4.2.1||Adobe Photoshop CS2||Bibble Pro 4.2.6|
|Open RAW files||12.2 sec||19.3 sec *||5.4 sec *|
|Memory (private bytes)||792,300 KB||187,476 KB||519,296 KB|
|Memory (virtual size)||1,136,668 KB||316,880 KB||599,925 KB|
|Save as JPEG||15.3 sec||2.6 sec||8.4 sec|
|Save as TIFF||18.1 sec||4.1 sec||4.3 sec|
* For Photoshop this time is the sum of the time to display the RAW plug-in and then to convert the image (twice), for Bibble Pro this is the sum of the two open procedures.
As you can see the best performance from a time taken point of view is Bibble Pro which opened files quickly and felt snappy in use, save time for JPEG was slower than Photoshop but faster after the first save. Memory usage for Nikon Capture is frankly unbelievable, it grabbed huge amounts of RAM just to display an image which caused our test machine to slow noticeably.