Sony G90 - Manual (Page 80)

film sources but doesn’t match the quality of inverse-
telecine deinterlacing.
Performance
My primary source for evaluating the IEV was a Sony DVP-
S7000 DVD player, using its composite, S-Video, and compo-
nent outputs to test the various Turboscan inputs. I also used
a Pioneer CLD-97 laserdisc player with its composite and S-
Video outputs. As a progressive video reference, I used a
3Dfusion PC video card with an Mpact-2 processor decoding
from a DVD-ROM drive to produce RGB output. The 3Dfusion
creates progressive video without deinterlacing artifacts
when playing film source DVDs, equivalent to the result of
Faroudja’s inverse-telecine processing. It is not nearly as
effective at deinterlacing DVD video sources, nor does it
accept any sources other than DVD. (See Issue 24 for a review
of the 3Dfusion and a discussion of its deinterlacing process.)
For display I used an Electrohome ECP-4100 front projec-
tor on a 6-foot wide Stewart Studiotek 130 screen (16.9 aspect
ratio). For a brief time, I was able to compare the Turboscan
to a Faroudja VP-250 line doubler on a Runco IDP-980 Ultra
front projector. (A line doubler requires a TV with progressive
video inputs and higher than normal scan-rate capabilities,
but these inputs and capabilities are standard with any graph-
ics-grade front projector.)
With real-world material, it was evident that the Tur-
boscan was not doing 3-2 pulldown removal, as the Faroudja
and 3Dfusion do (each in different ways). The Turboscan
sometimes displays line twitter artifacts on moving objects,
and while not nearly so objectionable as the artifacts seen
with non-line doubled sources, they make the Turboscan dis-
play slightly inferior to that seen from the Faroudja or 3Dfu-
sion. For example, in the opening scene of The Fifth Element
(chapter 3, absolute time 03:20-23), one can see fine line twit-
ter in the tent ropes and moving “jaggies” on camel backs at
the center of the screen as the camera pans left to follow the
running boy. In chapter 9 (absolute time 28:23-42), twitter arti-
facts appear as the camera zooms in on the curved top left
edge of the chamber that LeeLoo is kicking to get out of.
The Turboscan has a “motion filter” that appears in the menu
only if you have “Input Setup” turned on. The filter
slightly softens the image, but IEV recommends
that this filter be turned on for video. However,
there are examples where the image looks much
worse with the filter on. The horizontal pan of the
bridge during the video montage segment on Vi d e o
E s s e n t i a l s
is one. The vertical cables of the bridge break up
badly as they move horizontally with the camera pan. In other
cases the filter reduces, but doesn’t eliminate, some residual arti-
facts, such as jaggies in the hair of the pink T-shirted girl with the
b i c y c l e .
I used the Video Essentials DVD and the Dolby Labs DVD
(DVD-TEST1) to explore other aspects of the Turboscan per-
formance. The tests on these discs are rigorous and repeat-
able, even though they are much more demanding than what
you will usually notice on real-world video images.
With its sharpness control set at “0,” the Turboscan had a
significant high-frequency roll off above 4.2 MHz in all input
modes, but at a sharpness setting of +6, it became reasonably
flat to 5 MHz without outlining artifacts. This sharpness con-
trol isn’t the equal of Faroudja’s adaptive detail-enhancement
circuit, but it is useful for reducing the nastiness of some
DVDs with overdone edge enhancement; it is also handy for
slightly boosting the sharpness of soft video sources, espe-
cially many laserdiscs. You can set this control individually
for each of the three input sources.
The Turboscan showed minor artifacts on color-bar pat-
terns that were not present with either the Faroudja or the
3DFusion. For a component source, there was a faint light
band along the cyan side of the vertical yellow/cyan transition
and along the magenta side of the vertical green/magenta
transition. S-Video and composite sources did not display the
same artifact. Instead they had a black smudge along the two
transition regions characteristic of the lower chroma band-
width of these video formats. Some luma-chroma delay was
evident on the red vertical stripes against a white background
(title 18, chapter 4 of VE-DVD). There was a thin black line
along the left edge of the white/red transition, just inside the
red bar. On S-Video and composite sources, this artifact was
replaced by a smeared black region on both inside edges of
Faroudja VP251 IEV DVDO
Turboscan 1500 iScan Plus
Line Doubling Yes Yes Yes
Line Quadrupling No No No
Line Scaling (Other) No No No
Inverse Telecine (Film Mode Deinterlacing) Yes No Yes
Video Cal - Color,Tint,Brightness,Contrast All All None
Detail Enhancement Adaptive Detail Sharpness Only None
Noise Reduction None None None
Composite Video Inputs 1 1 1
S-Video (Y/C) Inputs 1 1 2
Component Video Inputs 1 (YPbPr/RGB) 1 (YPbPr/RGB) None
Component Video Pass-thru Yes Yes None
Input Loop-thru All Inputs Composite/S-Video None
Stored User Settings 4/Input 1/Input None
Other Features RGB Pass-through Input RGB Gain/Offset Film-Mode LED
Price $7,500 $2,495 $700
Key Features: Video Upconverters