Subject: Comparing Sony TRV900 vs. PC-1, PC-2, Panasonic NV-DX110
Date: Thu, 16 Sep 1999 02:10:40 -0700
From: "jerome maro"

I tried to do some resolution tests (using test patterns printed from your
site... thanks!) and here are my observations with the PAL version of the
TRV900, compared with my PC-1 (also PAL, 810K pixels CCD):

1. Resolution: I don't have a firewire card for my PC, so all I could do is
examine the output on my video projector (which can resolve over 550
lines...) and I found out that it is surprisingly difficult to estimate
resolution this way... Still: the TRV900 is a clear winner over the
PC-1. The PC-1 won't resolve over about 430 lines, while the TRV900 goes
about 100 lines more.

2. General pictures: I also try to record more normal images with the two
cameras and here are my observations.

 - Indeed the PC-1 is more soft (as predicted by the resolution test), but
the difference only shows on very fine details. The PC-1 has also slightly
less saturated colors.

 - talking about color: neither of the two cameras is very saturated,
giving a rather "dull" appearance to the pictures (compared e.g. to TV
programs). AWB is set a little too blue, the "outdoor" manual setting may
improve that, more on the PC-1 than on the TRV900.

 - a big disapointment was the TRV900 image stabilization. It is markedly
less smooth than on the PC-1 (and the PC-1 is MUCH smaller...). Electronic
image stabilization is not that bad, after all!

 - in plain daylight, the images are indeed comparable (the PC-1 being
slightly softer, as noted). Interestingly the PC-1 adds some slight
artifacts, but the artefacts from the interlacing seem slightly more
noticeable on the TRV900. The artifacts in the PC-1 (or any single-chip DV
camcorder I tried) are a bit like watching the scene through a sheet of
structured glass. I suppose those artifacts come from the fixed structure
of the color filters used in the single CCD.

 - when there is less light, the difference is far more obvious: even in a
brightly lit room, the PC-1 gives very noisy pictures with muddy colors
while the TRV900 is as good as in daylight. You can improve the PC-1
results a lot by tricking it to use a lower shutter speed, but it stays a
clear loser (and you have to know the trick, because the PC-1 has far less
controls than the TRV900). Note that from a quick comparison in a shop, the
new PC-3 is far better than the PC-1 in that respect (read on). An
important difference is also the appearance of the noise: if you lower the
light enough, the TRV900 starts to add noise as well, but its appearance is
far "nicer" than the one of the PC-1 (PC-1 noise is more colored and
"grains" appear bigger). Low light is indeed a reason to go 3CCDs!

3. Sensitivity: On the PAL TRV 900 used at 0dB, normal scan, I get the same
exposure readings than on my SLR if I use a sensitivity of about 250-320
ASA (one or the other, depends on the scene). This is using 1/1000 and 1/50
shutter speed. So it appears that the PAL model is more sensitive than the
NTSC one. Surprising.

[I'm surprised it's that much different, but note that different SLR
cameras can have different lightmeter calibrations also. -jpb]


I also had the occasion lately to test the TRV900 side-to-side with the
Sony PC-2 (essentially a PC-3 without the memory stick) and a Panasonic
NV-DX110 (3 CCD) in a shop (just brought a cassette with me, the camcorders
were on display...), again all 3 PAL models. The Panasonic is similar to
the model called AG-EZ30 in North America, see
e.g. http://www.geocities.com/ResearchTriangle/3749/panafaq.html . While
this test cannot be as extensive as the other one (because I have a limited
choice of indoor scenes to film, mainly the inside of the shop!), the
results are interesting. Basically:

 - the PC-2 has very reduced low light noise compared with the PC-1. In the
shop, I got the same exposure readings than in my "brightly lit interior",
BUT the PC-2 still exhibits pictures about as noiseless than the
TRV900. Interestingly, it is not that the CCD is much more sensitive, but
that even with +12dB gain, one gets almost no noise.

 -the rest of the comparison stays valid, just read the part above about
testing in broad daylight: the PC-2 suffers from the same artifacts than
the PC-1 and one notice the interlacing a tad more on the TRV900. But keep
in mind that this is a walkman-sized machine.

 -the Panasonic is also very interesting. I first tested it without image
stabilization, which is known to degrade the picture on that machine and if
you mount the camera on a tripod it gives slightly better images than the
TRV900: same low light noise, but -comparing the images side by side- one
realises that the TRV900 uses some kind of edge enhancement to boost
apparent resolution, which the Panasonic does not. The Panasonic AWB also
gives more pleasing (warmer) colors, at least under the fluorescent light
used in that shop (but took 5 seconds to do so). With image stabilization
on, the Panasonic gives rather fuzzy images (worse than the PC-2, but not
as bad as VHS as some people say), so it is true that image stabilization
is almost unusable on that machine.

However, the Panasonic has one very interesting feature: it can do
progressive scan at full frame rate (25 frames/s compared to 12.5 fot the
PAL TRV900). I also tested the results in that setting and all I can say is
that this mode works. Note that some people reported a "strobing effect" in
progressive scan. This effect is of course very marked on the TRV900 (which
lowers the frame rate), but also exist (much subdued) on the
Panasonic. Interestingly this effect depends on the output device: it was
more marked on the (LCD) monitors used in the shop than on my (also LCD)
video projector. In all cases, the artifacts from the interlacing are much
more discrete (as was to be espected), but do not completely disappear
(probably because the output device is still interlaced).

Prospective underground cinematographers should know that a progressive
scan PAL camcorder is the best choice for transfering to film, see:
http://www.dvfilm.com/faq.htm

I also have a comment on the GV-D900. That machine normally only displays
one of the fields in pause mode, simply doubling the lines to extrapolate
the other (my PC-1 does the same, but *interpolates* the missing lines). I
had the pleasant surprise to find out that this does not appear to be the
case is the tape has been recorded in progressive scan (presumably there is
a flag in the data that tells so). From the portion recorded with the
Panasonic, I get far better pictures in pause mode.

The conclusion of all this is that -on picture quality only- the Panasonic
wins if you use a tripod (an is much cheaper too: if you can live without
firewire in, the older model DX100 can still be add for something like 50%
less than the best price TRV900, which makes it the same price than single
chippers like the PC2), and may be a choice for cash-starved would be
cinematographers. It is surprising that it is so much less known than
Canon's or Sony's 3-chips offerings. The TRV900 still rules in practical
use because of the better stabilizer (and may have better control than the
Panasonic, at least it has a much bigger flip-out screen). But seeing the
low noise figures of the PC-2 / PC-3, one may consider to wait till we get
*those* CCDs in a 3-chips camera.