Like all digital SLR's and more recently prosumer level digital cameras Sony has decided to provide a RAW format option on the DSC-F828. RAW simply means data direct from the sensor (in the case of the F828 14 bits per pixel) which hasn't been processed in any way. Additionally the current camera settings (such as parameters, exposure etc.) are recorded in the header of the RAW file. Sony uses the .SRF extension for RAW files.
As mentioned earlier it appears as though Sony are packing the RAW data to 16-bits per pixel, this makes the RAW files about 2 MB larger than they need to be.
RAW vs. JPEG: Resolution
Provided with the DSC-F828 is the Image Data Converter application for converting RAW files to JPEG or TIFF. As you can see the RAW conversion engine in Image Data Converter appears to be almost totally identical to that in the camera, output from IDC is almost pixel for pixel identical to the camera. With one exception, IDC appears to apply slightly harder sharpening and hence stronger 'white halo' sharpening artifacts can be seen around the black detail of the RAW image.
|RAW (saved as highest quality JPEG)||JPEG|
RAW Latitude (dynamic range)
The sequence of images below were created by applying a negative digital exposure compensation to a RAW file at levels of 0, -0.5, -1.0 and -2.0 EV. As you can see at -0.5 EV some detail is recovered from the overexposed areas of the image, at -1.0 EV a little more detail but also a darkening of previous highlight detail to an almost gray appearance. To conclude the F828 appears to have approximately 0.5 EV (half a stop) of RAW latitude above the clipping point of the 'normal' exposure in an average shot. Note that the images below have been half-sized (downsampled in Photoshop CS; 1632 x 1224) to reduce file size and bandwidth requirements.
|-0.5 EV digital exposure compensation|
|-1.0 EV digital exposure compensation|
|-2.0 digital exposure compensation|
In natural light (outdoors) the DSC-F828's white balance was good, although did have a very slight cyan cast (not as bad as we experienced with the DSC-V1). In incandescent and fluorescent light the camera exhibited a slight pinkish cast. As expected manual preset white balance was as perfect.
|Outdoors, Auto||Outdoors, Sunny, Cloudy||Outdoors, Manual|
|Incandescent, Auto||Incandescent, Incandescent||Incandescent, Manual|
|Fluorescent, Auto||Fluorescent, Fluorescent||Fluorescent, Manual|
In macro mode and at full wide angle the DSC-F828 can focus as close as 2 cm from the subject, this however leads to images with very high distortion and low corner softness. We then attempted various positions between wide and telephoto for the optimum frame coverage with minimum distortion / softness. This was harder than I had expected and so we ended up with two possibilities, one at 62 mm equiv. and one at 70 mm equiv. Neither of which are 'the answer' as such, and I had hoped for quite good macro performance without distortion from this lens.
The DSC-F828 delivered mixed flash test results. Firstly with our skin tone test (which I repeated a dozen times to verify) had a slight cyan cast and was slightly underexposed, the wall behind the hand should be pure white. Secondly against our color patches chart no obvious color cast and a fairly good exposure.
|Skin tone - Slight cyan color cast, moderate exposure (slightly underexposed but acceptable)||Color patches - Good color balance, no color cast, good exposure|
The DSC-F828 uses a dark frame subtraction noise reduction method for long exposures, the camera records a second shot with the shutter closed which is used to subtract fixed pattern 'hot pixel' noise from the main image. This means that if you shoot a ten second exposure the entire process will take twenty seconds. Overall the DSC-F828 performed fairly well with moderately long exposures (the samples below are ten seconds each). We did note visible noise as well as some remaining hot pixels and some 'black hole' marks from the noise reduction.
|Manual exposure, ISO 64, 10 sec, F5.0|
|Manual exposure, ISO 64, 10 sec, F8.0|
Barrel and Pincushion Distortion
Interestingly the DSC-F828 produced almost identical barrel distortion results as the only other 28 - 200 mm prosumer lens, that on Minolta's DiMAGE A1. As we would expect we measured over 1% barrel distortion at wide angle (this is a 28 mm equiv. focal length) and what is actually quite good pincushion distortion of around half a percent at telephoto (200 mm equiv.)
|Barrel Distortion, 1.2% @ wide angle||Pincushion Distortion, 0.4% @ telephoto|
Vignetting / Lens Shading
Our vignetting / lens shading test is very simple, a shot of a blank wall from two meters away, vignetting will always be most visible at wide angle and maximum aperture and will start to disappear at smaller apertures and/or further zoom. We did observe a very slight amount of lens shading at wide angle and telephoto on the DSC-F828, it's on the borderline of whether it would be actually visible in everyday shots.
|Slight lens shading at wide angle, F2.0||Slight lens shading at telephoto, F2.8|