|Side-by-side: Sony VX1000 & Canon XL-1||
I posted this article on the DV-L mailing list on 22 February 1998. Copyright (c) 1998 Adam J. Wilt.
I've done a few informal tests to compare the Canon XL-1 with the Sony VX1000 at a very cooperative local dealer (Keeble & Shuchat Photography, Palo Alto CA -- knowledgeable people, great service, best non-mail-order prices around), and have found some interesting results.
Disclaimer: I've had a VX1000 since October '95. I have been seriously considering the Canon, especially for its excellent human interface. I went into this comparison as open-mindedly as possible, but I can't discount the possibility of some unconscious pro-VX1000 bias. You have been warned... :-)
Here, I'm comparing image quality in medium-high indoor lighting over a variety of zoom settings. I wasn't looking at low-light capability, sound quality, or how easily I could burn out viewfinders with sunlight. :-) I ran tests on two separate occasions, shooting the same scenes with both cameras, taking the footage into Premiere via the DPS Spark, and cutting A/B and splitscreen comparisons. The output was viewed off disk through a DHR-1000 transcoding to Y/C which was displayed on a PVM-1342Q "super fine pitch" Trinitron rated at some 600 TVL/pph.
K&S has a camera demo area with a large photomural of a cityscape behind an astroturfed area with model trains and cars, similar to the color-sense-challenged demo stages you see at NAB. The cityscape includes skyscrapers with closely spaced vertical lines (a.k.a. "poor man's resolution test chart"). I set both cameras to a telephoto setting of equivalent narrowness, and zoomed 'em out at comparable rates to an equivalent wide-angle setting. I also shot wide shots of the photo-processing desk, and telephoto pans of spare Canon batteries in red boxes. All shooting was handheld.
Overall results: The Canon showed better depth of modulation; images were crisper with the XL-1. The Sony, though a bit "muddier", had better high-frequency definition over all focal lengths, especially on single-color areas such as the red boxes or green astroturf. The Canon's autoexposure setpoint was also about 1/2 to 1 stop brighter than the Sony's. The result was that the Canon looked sharper most of the time, while the Sony was sharper. In general, the Canon's better lens delivers superior pix to the Sony's, but aft of the lens the Sony's higher-resolution chips do a better job of capturing fine detail.
Details: zoomed into the skyscraper mural, the Canon's pix were superior. The alternating light and dark lines were more contrasty, showing 30% more modulation on the waveform monitor than the Sony's images did.
As I zoomed out, wobbling the cameras slightly as handheld shooting will do, the Canon started showing aliasing and moire effects on the vertical lines before the Sony did; they were downright distracting just before they disappeared. They merged into a solid gray mass on the Canon before they did on the Sony, but only by a little bit before. From this, I suspect that the coarser chip resolution on the Canon is the culprit, and that the optical low-pass antialiasing filter may not be attenuating near-Nyquist frequencies smoothly enough, although it does seem to cut off things right at the Nyquist limit.
The Sony, on the other hand, kept the moire at a lower level as the detail limit was approached, but a bit more half-frequency aliasing was visible before the lines merged into a solid mass. Apparently the Sony's optical filter isn't quite as good at brickwalling fine detail before it exceeds Nyquist limits, though it does seem to cause a more linear loss of detail before cutting it off completely.
During the zoom out, fine details elsewhere in the image seemed to disappear slightly sooner from the Canon than from the Sony. However, the different exposure levels of the two cams made this a somewhat difficult call, as did the problem of matching the angle of view from the two shots exactly from frame to frame since the zoom rates differed slightly.
Zoomed wide, detail in the astroturf was noticeably more present in the Sony's pix than in the Canon's. Note that astroturf is pretty much a green mono-color image; pixel-offset chip used in the XL-1 mounting won't boost resolution unless there's luminance information derived from the R&B and G channels alike, so this was not too surprising.
Similar results were found shooting the rows of Canon batteries in their red boxes. The Sony was marginally sharper on areas with full-color (i.e., B&W) detail such as white text on a black background, and the Canon suffered somewhat more on the mono-colored red boxes with white text. However, as the Canon's pix were brighter (again), some of the apparent resolution loss might have been due to slight overexposure of the white text, causing it to "bloom" slightly.
Interestingly, the SteadyShot on the Canon was too good for this slow pan: it was simple to pan the Sony evenly across the scene, but the Canon tended to lock down on an area, then gently swing over to the side and lock down again. I tried to make the panning as smooth as possible; I suspect that the motion-vector feedback from the CCD image caused the XL-1 to attempt to lock down a static shot whereas the rate-sensor-only approach in the Sony was unaffected by this. A three-position setting on the XL-1's Steadyshot -- off, rate sensors, rate + vector feedback -- would help here. (On the other hand, the Canon sits so naturally in the hand and on the shoulder, its images were overall more smooth, stable, and free from annoying motion than the Sony's.)
The wide shot of the photo-processing desk included fine detail, large areas of light and dark color, white text on a brown background, and "real-world" people and racks of photo accessories. Here, The Canon's crispness and brightness paid off in spades; the XL-1's pix were clearly superior. It was hard to tell the images apart in terms of detail (even though the Canon has been reported to go especially soft at wide angles), and the colors appeared cleaner and purer on the XL-1.
However, the difference in exposure and white balance (the Canon autowhites more towards magenta than the Sony does, at least with the units I was using) annoyed me; I went into Premiere and used the "Brightness & Contrast" and "Color Balance" filters to visually match the Sony's pix to the Canon's. When I did this, 90% of the image quality difference went away; it was now difficult to tell the cameras apart on this one scene, although the Canon's image was still slightly "snappier".
Overall, I found that for the most part that the differences between the cameras were real, but usually subtle. The Canon usually gave a more subjectively pleasing image, but suffered on A/B comparisons in terms of fine detail actually rendered and in the "linearity" of detail loss during slow zooms out. I found I could intercut the footage from both cameras without a problem, especially if I matched exposures and color balances. As it was, I nearly went blind watching the tests for long periods of time with my eyeballs mere inches from the screen; I found the pix that close to each other. Most of the readily noticeable differences were due to exposure and color balance; the Canon's pix were brighter (thus often more pleasing) and a bit warmer (autowhite was more magenta; this tended to warm up the scene and was usually more pleasing if a bit less accurate) than the Sony's and this might be one reason that folks have been reporting such a huge subjective difference between the cameras' image qualities. When I controlled for these differences in post (admittedly after the fact), the subjective differences dropped dramatically, resulting in well-matched pix that I could intercut without a problem.
To sum up: both cameras will do the bollocks; you won't go wrong with either one. IMHO the differences between the user interfaces, ease of use, tweakability, shooting styles, ruggedness, size & weight, and the like far outweigh differences in raw resolution figures, and these are much more personal deciding factors that can't be meaningfully captured in dry specification charts. Play with 'em both, then decide for yourself, and don't obsess over the numbers.
Copyright (c) 1998 by Adam J. Wilt.
You are granted a nonexclusive right to duplicate, print, link to, frame, or otherwise repurpose this material,
as long as all authorship, ownership and copyright information is preserved, and a link to this site is displayed.
|Home||SW Engineering||Film & Video Production||Video Tidbits||>DV<|
Contact me via email
Last updated 30 March 1998.