[opendtv] Update: Lens Performance on Small Imagers

  • From: dan.grimes@xxxxxxxx
  • To: opendtv@xxxxxxxxxxxxx
  • Date: Thu, 5 Aug 2010 15:55:49 -0700

I performed a few more tests on these lenses and would like to provide an
update to my previous post.  Clearly, the limiting factor was the 720P
format and not the lens, imager or processing.

I set the Sony EX-3 cameras to 1920x1080@30P and re-evaluated the lenses.

Sony VCL-614B2X (stock lens):
At 70mm, attained over 1000 lines per picture height (L/PH), where contrast
and aliasing were not considered.  More specifically, the best attainable
resolution (around f4) with 50% contrast was 700 L/PH and the contrast was
at about 10% for 1000 L/PH.  At the minimum aperture of f1.9, the 50% point
was around 600 (L/PH) and the contrast was 5% for 1000 L/PH).

When getting over f16, the contrast drops fairly rapidly on this lens.

Fujinon HSs18x5.5BRM:
 At 55mm, attained over 1000 L/PH, where contrast and aliasing were not
considered.  The best attainable resolution (f4) with 50% contrast was at
700 L/PH and the contrast was 20% at 1000 L/PH.  At f1.9 (minimum aperture
is f1.4) the resolution for 50% contrast was at approximately 620 L/PH and
the contrast was 20% at 1000 L/PH.  Similar results were seen for 15mm at
f2.8 with close to 700 L/PH at 50%  and 8% contrast at 1000 L/PH.

Fujinon HS18x5.5BERM:
Surprisingly, the extender version faired slightly better than the HSs
lens.   At f1.9, the resolution was easily 700 L/PH at 50% at 20% contrast
at 950 L/PH.  At optimum aperture (f5.6), slightly better results were
found.  In fact, even at 11mm (the widest I can go for the chart I have and
maintain focus) at f2.8, there were areas on the lens that hit 800 L/PH at
50% but the corners were much softer at 200 to 250 on one side and 300 max
on the other.

With the extender in, about 1-2/3 f-stops were lost.  At peak resolution
(f4), 500 L/PH was attained for 50% with 15% contrast at 950 L/PH.  When
full wide, extender in (5.5mmx2=11mm), the lens is softer only giving 450
L/PH at 50% in the center and varied, but good, results at the corners (400
to 550, depending on the corner).

Fujinon XS13x3.3BRM:
At 43mm, the lens was not quite as sharp as the others, showing 650 L/PH.
At 1.4 the resolution was 630 L/PH at 50% and about 15% at 950 L/PH.  At f4
the resolution was 720 L/PH for 50% and 20% at 1000 L/PH.

When going wide at 7mm (I could not go full wide angle on my chart and stay
within minimum focus distance) with f2.8, center resolution fell slightly
to 680 at 50% but maintained 20% at 1000 L/PH. The corners didn't do so
well with the best corner at 520 and the worst at 330 for 50% contrast.

Overall, I thought these lenses paired well with the Sony 1/2" imager,
including the included lens.  I wish I had taken the time to plot all of
this as a transfer function to show how some lenses have a smoother fall
off than others.  But overall, the lens that came with the camera is no
slouch relative to the more expensive offerings.

For the record (to validate or invalidate my findings):

I used two resolution charts, a DSC Labs Combi-DX1 CDM 24+4R V4 (which is
not a resolution chart but has trumpets) and an ISO12233 resolution chart
that I printed on a laser printer (cannot be considered scientific).  The
DSC chart only goes to 800 lines per picture height (denoted on the chart
as LPPH).  Measurements were taken for the horizontal at the center
portions of the lenses but visually the vertical contrasts were the same or
very similar while the outer areas did not fair as well.  As to be
expected, aliasing started happened around 950 L/PH.  Sometimes the ND
filter in the camera was employed to attain the desired aperture and the
ATM-18 adapter, which has glass inside, was used with the Fujinon lenses.
Detail and Knee were set to minimum for all measurements using the HD-SDI
output into a Harris/Videotek VTM4100 waveform monitor.

I would also like to make a couple of corrections to my previous post: 1)
the Sony VCL-614B2X did have some barrel aberrations at the minimum focal
length (5.8mm) but was corrected very quickly when zooming in, and 2) the
Fujinon XS exhibited barrel aberrations, not pincushion (I think the one
that had pincushion was on the PMW-700.)

Next test will be 2/3" lenses on this 1/2" camera with the ATM-21 adapter
and the same lenses on a couple of 2/3" imagers (Sony PMW-700 and Sony

Feedback and criticism is welcome.


----- Forwarded by Dan Grimes/UNLV on 08/05/2010 10:57 AM -----(edited)

In the past, there has been discussion on this forum on lens performance
and price relative to imager format.  There were also questions as to
whether the lenses that came with these inexpensive cameras could resolve
to match their imagers.

For those that are interested, I recently tested different lenses on a Sony
PMW-EX3 with a 1/2" imager while set to 1280x720p60.

The lenses were as follows:
Sony VCL-614B2X (a Fujinon 14x5.8mm lens that is manufactured for and came
with the camera, $8.3K for both)
Fujinon HS18x5.5BERM (approx. $16K)
Fujinon HSs18x5.5BRM (approx. $11K)
Fujinon XS13x3.3BRM (approx. $10K)

All lenses exhibited enough resolution for 1280x720, easily showing 600+
line pairs of horizontal resolution (full 720 Lines per Picture Height,
L/PH) at both minimum and maximum focal length.  However, the lens that
came with the camera seemed to be absent of a myriad of aberrations in the
other three lenses, each costing more than the camera and lens package:

-Both HS series lenses could not be focussed for both the center and
outside areas (spherical aberration) with the f-stop fully opened (f1.4).
Optimum and consistent resolution was found at f4 to f5.6.
-Both HS series lenses suffered from lateral chromatic aberrations on the
outsides, especially when not perfectly focussed.  Usually, one side would
have blue shifted to one side and the other red.
-Both HS series lenses would breath when focussed (focal length varied when
focussing, accenting chromatic misalignment).
-The HS18x5.5BERM exhibited all the typical problems with the multiplier
in, yet still maintained very high resolution up around 600 line pairs.
-The XS series lens exhibited all the HS issues.
-The XS series lens exhibited strong aberration at wider angles.

-The VCL lens did not have strong monochromatic aberrations at any focal
length (squares staid square) and no apparent chromatic aberration when
focussed or not.
-The VCL lens did have lower contrast and lower resolution when at smaller
apetures (f11 and higher).  Optimum resolution was at f2.8 to f4.

Conclusions?  All lenses were well matched for the 1/2" imager at 1280x720P
in general resolution and contrast.  Of course, I need to go back and
repeat the test in 1920x1080i to determine if they match the full
resolution of the imager.  However, the cheeper lens that came with the
camera outshined the more expensive lenses.  Clearly this lens was
optimized for this camera either optically and/or electronically.  I am
still checking to see if there is lens data that can be used for the
Fujinon lenses in the EX3 camera.

Now, there is still good reason to purchase and use the Fujinon lenses,
namely, the additional focal length range, available accessories and
filters, and the ability to use studio controls.  But I wouldn't encourage
purchasing the more expensive lenses for image quality.

I guess what I am really getting at is don't discount the included lens
because of its price.  And perhaps those cheap lenses can match the
resolution of the imagers.


Other related posts:

  • » [opendtv] Update: Lens Performance on Small Imagers - dan . grimes