The HP 812 is the first camera I've used (for some time) which doesn't allow you to select white balance. This means that the normal array of nine white balance samples we provide has had to be reduced to just three, Auto White balance in our three different lighting conditions.
|Outdoors, Auto WB||Incandescent, Auto WB||Fluorescent, Auto WB|
The 812 did remarkably well in both natural (outdoors) and incandescent light. Indeed the incandescent automatic white balance is among the best I've seen of any level of digital camera. It's a shame the same can't be said of fluorescent light white balance, here is a good example of where manually selectable white balance would have been very useful.
Low Light Focus
This test measures the minimum amount of light under which the camera can still focus. The focus target is our lens distortion test chart (shown here on the right), camera is positioned exactly 2 m (6.6 ft) away.
Light levels are gradually dropped until the camera can no longer focus. This is carried out at both wide angle and telephoto zoom positions (as more light reaches the focusing systems with a larger aperture).
This test target is the optimum type of subject for most "contrast detect" AF systems (as it has a vertical line at its center), you should consider the results below the best you could expect to achieve.
|Lens position||Aperture||Lowest light focus|
|Wide angle (37 mm)||F2.6||2.5 EV (14.1 Lux, 1.3 foot-candle)|
|Telephoto (111 mm)||F4.8||2.5 EV (14.1 Lux, 1.3 foot-candle)|
Light intensity (Lux) = 2.5 x 2^EV (@ ISO 100), 10.76391 Lux = 1 foot-candle (fc)
Despite repeating our test several times I continued to get the same result, the amount of light required for an AF lock was the same at both wide angle and telephoto. This isn't what we'd expect as more light should be getting through (wider max aperture) at wide angle. Anyway, 2.5 EV isn't too bad, but it's not particularly stellar either.
The 812's flash performance was a fairly mixed bunch. Just like in the white balance tests the 812 does remarkably (almost artificially) well on the colour patch flash shot test. However, there's noticeable drop off in the 2 m wall test. It does appear as though HP has addressed the blue cast problem which occured in the original firmware.
Barrel and Pincushion Distortion
As we would expect the 812's distortion performance was virtually identical to that seen in Pentax's Optio 430 (because they share the same lens mechanism). Slightly higher than average barrel distortion at wide angle and no measurable pincushion distortion at telephoto.
|Barrel Distortion, 1.1% @ wide angle||Pincushion Distortion, 0% @ telephoto|
Purple Fringing (Chromatic Aberrations)
The 812 did fairly well on our standard chromatic aberration test chart, just a hint of purple fringing but nothing to be really worried about. In every day shots with overexposed details the tendancy was towards a thin blue or green fringe.
|You can see a purple fringing below the overexposed detail and some blue/green above it.|
|Our now standard chromatic aberration test shot - mostly blooming|
Overall Image Quality / Specific Issues
When I first reviewed the HP 812 I didn't like the image quality. The primary issue was very soft images which had a 'video like' look and sometimes had noticeable noise reduction. Since the new firmware things have improved, although I'd still say that the 812 is softer than many other four megapixel digital cameras and it's resolution isn't up to its four megapixel billing.
The other point to remember is my disappointment with the camera's JPEG encoding algorithm, this means that if image quality is important to you then you will avoid anything under maximum resolution (thus use either or ).
New firmware improves sharpness
As you can see from the side-by-side samples below the new v1.00.11 firmware does go a long way to improving the 812's sharpness (although it's still softer than others). While resolution hasn't visibly increased the overall look of images is much sharper and more acceptable. It also appears that the camera now has an improved moiré removal algorithm. It's good to see a big manufacturer taking fast action to fix a problem and get the fix into the hands of the users. Kudos.
|Old firmware||New firmware (v1.00.11)|
Again, from the same sample as above the orange colour you can see below should actually be bright red. You can see hints of the true colour around the edge of the first crop, the rest of the orange colour seen was not in the original shot.
Dynamic Brightness / Contrast Control
Although HP hasn't publicized this the 812 does have a rather interesting feature which automatically alters the S-curve (used for translating the linear data from the CCD into a 'gamma correct' sRGB image). This automatic selection means that although the 812 doesn't have an exposure compensation control it does do a very good job of correcting the image before it's written away. This can clearly be seen in the two examples below.
Two very different shots, but both have similar levels of "brightness", that is the camera's use of the entire range of greyscale (whites come out white, blacks not washed out) while maintaining as much of the source dynamic range as possible.