Sony's new 4K video sensor

http://www.sony.net/Products/SC-HP/c...imx144cqj.html

- Diagonal 9.33 mm (Type 1/1.7) approx. 12.40M-effective pixels (4072H x
3046V)
- Pixel size: 1.85 µm unit pixel
- Supports 12.40M-pixel imaging at approx. 35 frame/s
- Back-illuminated CMOS image sensor featuring high sensitivity, high
dynamic range and low noise
- Provides 4K video mode (4096H x 2160V, 60 frame/s)
--

Alfred Molon
------------------------------
Olympus E-series DSLRs and micro 4/3 forum at
http://tech.groups.yahoo.com/group/MyOlympus/
http://myolympus.org/ photo sharing site

On 2012.10.14 08:54 , Alfred Molon wrote:
http://www.sony.net/Products/SC-HP/c...imx144cqj.html

- Diagonal 9.33 mm (Type 1/1.7) approx. 12.40M-effective pixels (4072H x
3046V)
- Pixel size: 1.85 µm unit pixel
- Supports 12.40M-pixel imaging at approx. 35 frame/s
- Back-illuminated CMOS image sensor featuring high sensitivity, high
dynamic range and low noise
- Provides 4K video mode (4096H x 2160V, 60 frame/s)

For shallow DOF shooting it will present difficulties. The diagonal is
some 4.6X less than FF. A "normal" view would be at about an 11 mm FL.
(note the aspect ratios are not 3:2 so the above numbers will vary a
bit - but the basic idea is there.

In article , Huuter says...
For shallow DOF shooting it will present difficulties. The diagonal is
some 4.6X less than FF. A "normal" view would be at about an 11 mm FL.
(note the aspect ratios are not 3:2 so the above numbers will vary a
bit - but the basic idea is there.

Yep, but the 60 frames/s at 4K output is impressive.

By the way, isn't there a way to make an F1 or F0.5 lens for a small
sensor like this one, to get low DOF?
--

Alfred Molon
------------------------------
Olympus E-series DSLRs and micro 4/3 forum at
http://tech.groups.yahoo.com/group/MyOlympus/
http://myolympus.org/ photo sharing site

On 14/10/2012 20:41, Huuter wrote:
On 2012.10.14 08:54 , Alfred Molon wrote:
http://www.sony.net/Products/SC-HP/c...imx144cqj.html

- Diagonal 9.33 mm (Type 1/1.7) approx. 12.40M-effective pixels (4072H x
3046V)
- Pixel size: 1.85 µm unit pixel
- Supports 12.40M-pixel imaging at approx. 35 frame/s
- Back-illuminated CMOS image sensor featuring high sensitivity, high
dynamic range and low noise
- Provides 4K video mode (4096H x 2160V, 60 frame/s)

For shallow DOF shooting it will present difficulties. The diagonal is
some 4.6X less than FF. A "normal" view would be at about an 11 mm FL.
(note the aspect ratios are not 3:2 so the above numbers will vary a
bit - but the basic idea is there.

It's small even for a camcorder sensor, which seems to be the market
it's been designed for. According to the webpage linked, it's supposed
to replace current 1/2.3" sensors. I don't even know which camcorders
use a 1/2.3" chip, most XDCAM ones I'm familiar with are 3-chip cameras,
either 1/3" or 1/2", with a trichroic beam splitter rather than a Bayer
mask.


--
audentes fortuna iuvat

On 2012.10.14 17:53 , Alfred Molon wrote:
In article , Huuter says...
For shallow DOF shooting it will present difficulties. The diagonal is
some 4.6X less than FF. A "normal" view would be at about an 11 mm FL.
(note the aspect ratios are not 3:2 so the above numbers will vary a
bit - but the basic idea is there.

Yep, but the 60 frames/s at 4K output is impressive.

By the way, isn't there a way to make an F1 or F0.5 lens for a small
sensor like this one, to get low DOF?

I seem to recall that there's a limit around f/0.7 or f/0.5 where the
lens will almost be in contact with the sensor. (That may have been
limited to a particulart FL &| frame size however). Then (quibble)
there is the case of the sensor which has a glass cover over the sensor
proper (a mm or 2?). At these scales that counts.

The other side of that of course is the cost, not to mention things like
shutters. I don't know if any/many video cameras use actual shutters
(like cine film cameras do). Red One (for example) does not have a
shutter. It also had (has still) a rolling frame issue when the camera
is panned or when there is fast action.

On 2012.10.14 17:53 , Joe Kotroczo wrote:
On 14/10/2012 20:41, Huuter wrote:
On 2012.10.14 08:54 , Alfred Molon wrote:
http://www.sony.net/Products/SC-HP/c...imx144cqj.html

- Diagonal 9.33 mm (Type 1/1.7) approx. 12.40M-effective pixels (4072H x
3046V)
- Pixel size: 1.85 µm unit pixel
- Supports 12.40M-pixel imaging at approx. 35 frame/s
- Back-illuminated CMOS image sensor featuring high sensitivity, high
dynamic range and low noise
- Provides 4K video mode (4096H x 2160V, 60 frame/s)

For shallow DOF shooting it will present difficulties. The diagonal is
some 4.6X less than FF. A "normal" view would be at about an 11 mm FL.
(note the aspect ratios are not 3:2 so the above numbers will vary a
bit - but the basic idea is there.

It's small even for a camcorder sensor, which seems to be the market
it's been designed for. According to the webpage linked, it's supposed
to replace current 1/2.3" sensors. I don't even know which camcorders
use a 1/2.3" chip, most XDCAM ones I'm familiar with are 3-chip cameras,
either 1/3" or 1/2", with a trichroic beam splitter rather than a Bayer
mask.

With this sensor's back illumination structure (lower noise - good at
this size), the advantages of trichroic seem to be all but gone (esp.
when considering optics cost).

The "movie" people (which drives 4k) seem to be going to RGB sensors
while ENG seems to be the holdout for trichroic. Some ENG cameras are
equipped with fabulou$ optics - there may be accompanying reluctance to
dump trichroic - they also have no needs above 2k (or practically 1280
if it's strictly "NG").

As a 4k sensor, the one at hand is not aimed at ENG, obviously.

Sony must anticipate a market for rather small beyond HD cameras for the
amateur action film maker... or something like that.

On 15/10/2012 11:34 a.m., Huuter wrote:
On 2012.10.14 17:53 , Joe Kotroczo wrote:
On 14/10/2012 20:41, Huuter wrote:
On 2012.10.14 08:54 , Alfred Molon wrote:
http://www.sony.net/Products/SC-HP/c...imx144cqj.html

- Diagonal 9.33 mm (Type 1/1.7) approx. 12.40M-effective pixels
(4072H x
3046V)
- Pixel size: 1.85 µm unit pixel
- Supports 12.40M-pixel imaging at approx. 35 frame/s
- Back-illuminated CMOS image sensor featuring high sensitivity, high
dynamic range and low noise
- Provides 4K video mode (4096H x 2160V, 60 frame/s)

For shallow DOF shooting it will present difficulties. The diagonal is
some 4.6X less than FF. A "normal" view would be at about an 11 mm FL.
(note the aspect ratios are not 3:2 so the above numbers will vary a
bit - but the basic idea is there.

It's small even for a camcorder sensor, which seems to be the market
it's been designed for. According to the webpage linked, it's supposed
to replace current 1/2.3" sensors. I don't even know which camcorders
use a 1/2.3" chip, most XDCAM ones I'm familiar with are 3-chip cameras,
either 1/3" or 1/2", with a trichroic beam splitter rather than a Bayer
mask.

With this sensor's back illumination structure (lower noise - good at
this size), the advantages of trichroic seem to be all but gone (esp.
when considering optics cost).

The "movie" people (which drives 4k) seem to be going to RGB sensors
while ENG seems to be the holdout for trichroic. Some ENG cameras are
equipped with fabulou$ optics - there may be accompanying reluctance to
dump trichroic - they also have no needs above 2k (or practically 1280
if it's strictly "NG").

As a 4k sensor, the one at hand is not aimed at ENG, obviously.

Sony must anticipate a market for rather small beyond HD cameras for the
amateur action film maker... or something like that.

The "4.6x less than FF" statement with regard to (shallow) DOF is a bit
of a red-herring. It's about 2.8x less than (typical) 35mm movie format.
They still use crews to operate cine cameras, digital or film, with
"focus pullers" being an important job.
For ENG where you're not going to have a crew to operate your camera, or
"home cinema", the deep DOF from smaller sensors is an advantage - not a
disadvantage.
ENG will have an eventual need for UHDTV/4k - if...
Sport - definitely.
Home movies shot in UHDTV format will suck (technically) just as badly
as they did in HD and SD and 8mm.
I don't think this Sony sensor is for the home movie market.

Now ITU have released "Rec 2020" for UHDTV in 3840x2160 (as well as the
"next step" - 7680 x 4320), and the first Rec 2020 compliant TV sets are
arriving (if you have a spare $20k for a TV - with no UHD content to
display on it), at least one thing may be being put to rest - whether
the "next step" from HDTV was to be this format, or "4K" movie format.
Distribution/broadcasting is the problem.
More on-topic for this forum, a Rec 2020 compliant UHDTV, which includes
a new ultra-wide gamut spec and for 10/12 bit colour depth should be
pretty freaking awesome for still image display and editing.


If UHDTV "takes off", then theatre owners, some expressing reluctance at
investing in 4k / 3D projection equipment to upgrade the 2k gear they
recently bought, might get a little more nervous. If UHDTV format
movies are available to the home, and I had a 100" UHDTV set, then I
doubt I'd want to step foot in a cinema ever again. If a movie producer
made most of their $$ from cinema release, then they're not going to
want to see their movies released on (UHD) disk in a hurry. The studios
aren't going to re-master all the movies made over the past decade and
more, almost all intermediate mastered in 2k, just so they can sell a
few disks to the faithful, so you may as well stick to the BD version.
It makes me skeptical that it's all going to take off any time soon.
Even bluray disk uptake is poor, despite massive advantage over SD/DVD,
and low cost of equipment.

Alfred Molon wrote:
In article , Huuter says...
For shallow DOF shooting it will present difficulties. The diagonal is
some 4.6X less than FF. A "normal" view would be at about an 11 mm FL.
(note the aspect ratios are not 3:2 so the above numbers will vary a
bit - but the basic idea is there.

Yep, but the 60 frames/s at 4K output is impressive.

By the way, isn't there a way to make an F1 or F0.5 lens for a small
sensor like this one, to get low DOF?

Sure. Glass-Air --- f/0.5 is the theoretical maximum (for
conventinal lenses. You can build f/0.35, but that's one
block of glass (with a mirror) and the image is formed
*inside* the glass:
http://www.dg77.net/photo/tech/fast_...idt_solide.jpg

In fact American Optical had a f/0.38 lens, and f/0.67 seems to
be the fastest non-mirror lens. The f/0.7 "back side of the moon"
lens is probably the best known of the ultra fast lenses.

Consider the prices for a 50mm f/1.8 , f/1.4, f/1.2, f/1.0
(all EF lenses from Canon)[1] you can see where the prices for
a f/0.7 would be --- it'd be many times cheaper to use a normal
large sensor and a f/1.2 lens and get *much* less DOF than using
a f/0.7 lens and yon tiny sensor.

Oh, note that the fastest common zooms are f/2.0. Everything
faster is not a zoom, AFAICT. So your very fast lens would mean
one single focal length --- try to sell *that* to someone who
just bought a very high end (i.e. 4k) cam corder.

-Wolfgang

[1] 12k Yen (the II version), 49k Yen, 185k Yen, 359K Yen,
according to the camera museum, you need to add 23 years(!) of
inflation to the f/1.0 ... which doubles the price. Then add
in that the yen is much stronger today ... another factor
of 1.5, unless I'm mistaken.

Basically each half-stop over that range quadruples the ---
towards 1.0 it goes to almost 6x per half-stop. Based on that,
f/0.7 would be 360 times as expensive as a fast f/1.4 lens.