Digital Cameras - Toshiba PDR-M4 Test Images
portrait: (820k) Good color, tonal
range, and detail overall. The default
shot (828k) produced a surprisingly
good exposure, with the PDR-M4's exposure system handling the
"high key" subject quite well. We preferred the slightly
brighter shot taken with +0.3EV (820k),
and chose it as the main shot for this category, although it
lost more of the highlight detail. The color is good, with a
slightly "cool" cast overall. (An ideal candidate for
the PhotoGenetics software
we review elsewhere on this site!) However, we observed in our
"Far Field" shots below that the "daylight"
white balance appeared to entirely correct this coolness - Unfortunately,
we didn't have an opportunity to reshoot the Outdoor Portrait
shot with that setting... Skin tones are good, as are the colors
in the flowers. The PDR-M4 has the tendency we've noticed in
many digicams to push the blues to a slightly purplish color.
(We're not sure why, but this seems to be a very common weakness
among the cameras we've tested.) Resolution and sharpness is
very good, easily on a par with other 2 megapixel cameras we've
tested. For those interested, the table below contains links
to shots taken with a range of exposure compensations, from 0
to +0.9 EV.
portrait: (788k) For the closer
shot, we chose a bit more exposure compensation, with a setting
of +0.6EV (820k). This sacrificed detail
in the shirt for better skin tones. Detail, color, and resolution
are all excellent, although the relatively short fixed focal
length of the PDR-M4 means it probably wouldn't be your first
choice for portraiture, due to its tendency to enlarge the subject's
nose... Overal color cast is very slightly cool, but more accurate
than most cameras on this shot. As before, the table below holds
a series of exposure, ranging from 0 to +0.9 EV.
||Indoor portrait, flash: (620k) The indoor flash pictures we shot with the PDR-M4 were a bit on the dark site (probably due to the large white wall behind the model), and had an overall magenta cast, probably due to the heavy incandescent room lighting. - This shot is generally very difficult for digicams to achieve proper color balance on, due to the huge difference between the color temperature of the flash and the room lighting. To its credit, the PDR-M4 produced a very uniform color cast, with none of the blue highlights we frequently find. As a result, the images "clean up" remarkably well in Photoshop, as shown in this image (556k), which had an "auto levels" operation performed on it. Like other Toshiba digicams we've tested, the PDR-M4 is unusual in that it allows exposure adjustments on the flash, as well as on the ambient-light exposure. In the case of the 'M4 though, you can only adjust the flash down, not up. Given the rather dark overall images here, we definitely would have liked having the option to boost flash exposure somewhat. The ability to adjust both flash and ambient exposure is nonetheless very nice, allowing subtle variations such as this (624k), wherein we boosted the ambient exposure by 0.6EV, while dropping the flash by 0.3EV. This shifted the balance of the lighting somewhat, an effect more evident (but still very subtle) after a subsequent Photoshop Levels adjustment (820k).|
portrait, no flash: (548k) This
shot is a very difficult test of the white-balance algorithms
of most cameras, given the strong yellowish cast of the household
incandescent lighting used to illuminate it. Still, it's a situation
likely to be encountered by users, so we think is a valid test.
Like so many cameras we've tested, the PDR-M4's automatic white
balance system wasn't quite up to the task of getting rid of
so much yellow, producing shots with a strong yellow cast, as
shown here (820k). The "incandescent"
setting fared quite a bit better, producing the result shown
here (820k), although there's still too
much yellow for our tastes. (Like most digicams, the "incandescent"
setting seems to be adjusted for professional tungsten lighting,
rather than household bulbs.) The good news though, is that the
color cast cleans up beautifully in Photoshop (tm), with an "auto
levels" adjustment, as shown here
(820k). (This would be another good application
for the PhotoGenetics software,
mentioned earlier.) The table below shows a range of exposure
compensation applied to this shot, in both automatic and incandescent
white-balance modes. (The main shot
(820k) was taken with the incandescent
setting, and an exposure compensation of +0.6EV.)
House shot: (936k) Our standard House poster is one of our strongest tests of detail and resolution. The PDR-M4 performed very well, easily keeping pace with the highest-resolution cameras we've tested to date (August, 1999). A slight softness is evident in the corners of the frame, partially due to the lens used to shoot the original for the poster, but some also due to corner softness in the M4's lens. Color was very good, with virtually no color cast when using auto white balance. A shot taken with the "daylight" white balance setting (240k) resulted in a more yellowish cast to the picture as a whole. For this test, we shot a series of images, showing the camera's performance at each of its resolution/image quality settings. We observed that the low-resolution pictures the PDR-M4 takes (800x600 pixels) are of rather poor quality, compared to competing units: If you plan to shoot a lot of smaller pictures, either choose a different camera, or shoot at the larger resolution with a lower quality setting, and resize the images in the computer. Our results are arranged in the table below for your perusal.
We also tested the three "sharpness" settings of the PDR-M4, as shown in the table below. We found the default in-camera sharpening of the PDR-M4 quite acceptable, neither too strong, nor too weak. The "Hard" setting resulted in a more pronounced sharpening, with strong edges in high-contrast areas, but wasn't as severe an effect as we've seen on many cameras. Likewise the "Soft" setting produces images apparently having no sharpening applied in-camera, but no deliberate blurring either. We'd recommend this setting for images that are going to be enlarged significantly, with subsequent sharpening done in Photoshop or other imaging program.
Far-Field shot: (940k)
This image is shot at infinity to test far-field lens performance.
NOTE that this image cannot be directly compared
to the other "house" shot, which is a poster, shot
in the studio. The rendering of detail in the poster will be
very different than in this shot, and color values (and even
the presence or absence of leaves on the trees!) will vary in
this subject as the seasons progress. In general though, you
can evaluate detail in the bricks, shingles, and window detail,
and in the tree branches against the sky. Compression artifacts
are most likely to show in the trim along the edge of the roof,
in the bricks, or in the relatively "flat" areas in
"Musicians" poster: (892k) While this is just a picture of a poster, the color values for the various skin tones are pretty representative of the three ethnic groups represented. Skin tones are tough for digital cameras, both because the Caucasian skin color is so sensitive to over-saturation, and because all of the tones are "memory colors:" People are so familiar with the range of "correct" colors that any deviation is immediately obvious. The PDR-M4 did very well again here, producing very natural colors, and with an excellent overall color balance. (Some cameras get fooled by the bluish cast of the background and the strong blue of the Oriental model's robe, and overcompensate with a yellowish cast overall, but not the M4.) We used the auto white balance for our main shot (892k) here, as it gave slightly warmer tones than did the daylight setting, as shown in this low-resolution sample (196k). Detail is also very good, about mid-range of the 2-megapixel cameras we've tested to date, but not far off the best we've seen. As before, a full set of resolution/image quality samples are arranged in the table below, as well as a set of sharpness variations in the second table following.
|Macro shot: (896k) - Initially, we were perplexed in our macro shooting with the PDR-M4: The manual specified a minimum macro distance of 4 inches (10cm), but we couldn't get closer than about 6 inches. A quick check with Toshiba revealed that a small percentage of their first production run (apparently 5 or 6 cameras out of a set of 200) had a lens defect that resulted in this problem. A full-production unit was shipped to us that didn't exhibit this problem, with the result you see here (820k), although we were still obviously pushing the limits at just a shade over 4 inches, as witness the fuzziness of the brooch, which stands up above the background by about 0.2 inches. The minimum capture area measured 3.6 x 2.7 inches (91 x 68 mm). This is reasonably close, about middle-of-the-road for digicams we've tested. The flash does a fairly good job (924k) of throttling-down this close , although the perfectly perpendicular subject reflected the flash quite strongly into the lens. While sharpness was quite good overall, we noticed some softness toward the right-hand edge of the frame. Normally, we'd attribute this to a misalignment of the camera with the subject, but in this case we adopted a trick suggested by a reader for squaring-up copystand shots like this: Lay a mirror on the baseboard, and align the camera so its image is perfectly centered in the mirror. (Handy trick!) Overall, macro performance is good, but not in the same league as some current cameras.|
"Davebox" test target: (768k) The PDR-M4 turned in an good performance on this test, showing very good color and tonal range. Colors were bright yet natural, but we found the red swatch of the MacBeth chart a little under-saturated relative both to the original, and to other top-rated cameras. On the other hand, shadow detail was better than most cameras we've tested, and the camera did an excellent job of using the available tonal range. Noise in the shadows is somewhat higher than average though, not only in the always-difficult blue channel, but in the red as well. The difficult red/magenta separation in the small horizontal color-separation target is handled well, although the magenta is shown as more of a somewhat cooler red than as a true magenta. Tonal separation in the delicate pastels of the Q60 target is particularly good. Our main shot (768k) here was taken with the auto white balance setting: This low-resolution shot (156k) shows the more yellowish tone produced by the "daylight" option.
As before, we have a full resolution/image-quality series in the table below, as well as a series showing the effect of various "sharpening" settings.
Low-Light Tests (New)
The PDR-M4 had a real surprise in store for us in the low-light department: It uses significantly more-sophisticated in-camera processing to reduce noise levels and eliminate "stuck" pixels than we've found in any other camera to date (August, 1999). We initially discovered this when we noticed that the camera was taking a *really long* time to take the shots in "Bulb" exposure mode. In fact, it seemed to be taking exactly twice as long as the exposure time we'd selected. We're proud that we guessed correctly what was going on, as subsequently confirmed by the Toshiba engineering team: The PDR-M4 automatically takes a "black current calibration" shot immediately after every Bulb-mode exposure, and then applies the results of that calibration to the captured image "on the fly," before even writing it to the memory card! Very slick trick, but one that requires a lot of on-board horsepower, something the PDR-M4 apparently has in spades... We expect other manufacturers will adopt this approach as onboard processors generally become more powerful.
The table below holds links to the shots we took in our low-light
testing. Overall, the basic exposure system of the camera significantly
outperformed the specifications Toshiba gave for it, reaching
as low as EV 9 (4 foot-candles, 44 lux) quite easily, and producing
a usable image at EV 8, (2 fc, 22 lux) even though the "official"
spec would have indicated EV 10 (8 fc, 88 lux) as the lower limit.
Using the "Bulb" exposure mode, we were able to obtain
good images as low as EV 6 (0.5 fc, 5.5 lux)(!), although the
resulting image had a rather pronounced
reddish cast, as seen here (704k).
However, this cast was quite uniform, and cleaned up very
nicely in Photoshop, as shown here
(588k). (This is *really* dark: 0.5 fc
is about equivalent to the light of one candle at a distance
of about 1.4 feet!) Even at the longest exposure times, there
was no sign of "stuck" pixels, and image noise was
quite low. Below 0.5 fc/5.5 lux, the autofocus began to seriously
lose its ability to function, but the results were nonetheless
better than other more-expensive cameras we'd tested previously.
Really, about the best low-light performance we've seen to date
(August 1999) from any camera, a very impressive performance!
Note though, that the camera does in fact have a fairly low ISO
rating, meaning that you'll be forced to very long exposures
to obtain these results. Also, the reddish tinge is not to be
ignored, although as noted it is quite easy to "clean up."
Herewith a collection of our low-light test results:
Flash Range Test (NEW!)
In response to several reader requests, we've begun testing the usable working range of the on-board flash units of digicams. We don't have a real formal procedure for this, but mainly want to determine whether the manufacturer's ratings are reasonable or not. In the case of the PDR-M4, we're happy to report that the flash unit appears to be fairly conservatively rated: Toshiba rates it to only 8.2 feet (2.5 m), as shown in this digital tele-enabled shot (160k). We felt that all of the flash test shots were a little underexposed, perhaps because of the specular reflection from the vertical grayscale strip on the test target. Beyond that overall exposure issue though, the flash fell off relatively little as we backed away from the subject. The table below shows shots taken at a range of distances, from 6 to 11 feet, in 1-foot increments. You can judge for yourself how far away the flash is useful: We'd like to see it be brighter even up close, but the falloff with distance is quite moderate, and we felt that even the 11-foot sample was quite usable.
ISO 12233 ("WG-18") resolution target: (776k) (Technoids only) Resolution on the PDR-M4 is a little tough to call, due to somewhat higher-than average color aliasing on the fine parallel lines of our test target. If we ignore the color artifacts, we see a visual resolution of 750-800 lines/picture height horizontally, and 650-700 vertically. When the color artifacts are taken into account though, the resolution numbers drop to 700 and 650, respectively. (To determine the resolution without the color aliasing, we converted the test image to monochrome in Photoshop.) When only the monochrome image is considered, resolution is on a par with other 2 megapixel cameras we've tested, but the color version tests somewhat lower. Although color aliasing will only show up in the face of such very fine repeating patterns, a the conservative interpretation would be that the PDR-M4's resolution is on the order of 700 lines per picture height, a bit below other 2 megapixel cameras tested to date (August, 1999). As noted earlier, the PDR-M4 does not do a particularly good job at the 800x600 resolution level (200k): Even for web work, we'd advise shooting at the higher resolution, and downsampling in an image-editing application after the fact. Interestingly though, the "digital tele" function (188k) gave better 800x600 results than the 800x600mode itself!
As usual, the tables below show samples of all resolution and image size, as well as the effects of the various options for in-camera sharpening).
Viewfinder accuracy/flash uniformity target: The optical viewfinder on our preproduction test sample was rotated about one degree relative to the CCD's field of view, but the production sample we received subsequently was dead-on in this respect. Both optical and LCD viewfinders feel somewhat looser than those of other cameras, with coverage of 79.5% for the optical (212k) and 91% for the LCD (208k). Flash coverage is more uniform than most, with some falloff in the corners, but far from the worst we've seen. (Worst-case corner brightness was 55% of that of the center of the target: We haven't quantified this parameter previously, but will begin measuring and reporting on it in all future reviews.)
We've recently begun testing cameras for optical distortions, such as barrel/pincushion distortion, and chromatic aberration. The PDR-M4 did well in both these respects, with only 0.3% barrel distortion, and a slight fringing due to chromatic aberration at the edges of the frame of about 2-3 pixels (0.2% maximum measured).