DV Info Net

DV Info Net (https://www.dvinfo.net/forum/)
-   Apertus: Open Source Cinema Project (https://www.dvinfo.net/forum/apertus-open-source-cinema-project/)
-   -   Sensors (https://www.dvinfo.net/forum/apertus-open-source-cinema-project/236445-sensors.html)

Oscar Spierenburg May 31st, 2009 09:21 AM

Sensors
 
This version of the Elphel has a Aptina 5Mpix 1/2.5" CMOS sensor. It gives a very nice organic image and is able to do full HD at 24fps.
In lots of ways, this sensor is usable for our cinema project, but the main disadvantage is it's size and the rolling shutter effect.
The small size rules out a lot of lenses (incl most video lenses) to be usable and of course it gives a really wide DOF.
The rolling shutter almost certainly rules out handheld shooting, unless your very skilled or use some type of stabilization. I've done some nice tests with a dolly though, so we're not talking about a static camera.

Some time ago Matteo wrote:
Quote:

Economically, 2/3 sensor size seems quite right curently.

Next step would be aps-c (1.6 crop factor of 3 mm), then 35mm, but the
availability of those sensors and the prices won't be easy.

I prefer 2/3'' (above the 1/2.5") because they are almost made for video
instead of aps-c that are designed for photocamera because for example the d90
by nikon that have my uncle is a very good for photo but video is no good
evrithing is like a jellyfish ...rolling shutter is too evident
so for our project I think is better to take tested component for
video application
What 2/3" sensors are available with a global shutter?

Right now, for a hardware change like this, we need the help of Elphel. But if there is someone out there interested our project with knowledge about these things, please email me. Don't forget the Elphel is open hardware, so it should be a challenge to work on this!

Matteo Pozzi May 31st, 2009 09:44 AM

As you know only global shutter cmos can avoid rolling shutter artifacts in cmos sensor but they cost a lot (even RED do not use them) and, most important, they output 10 time the data of a same sized rolling sensor so with actual hw are impossible to handle

I think that to solve definetly the rolling shutter problem we can now switch to a ccd sesor made by kodak
is the same sensor used in a new digital camera: A-cam dII
A-cam dII - The camera loves you | A-cam dII | Products | Ikonoskop

it use a kodak KAI-02150
KODAK Image Sensor Solutions - KAI-02150 - Specifications

if we want to stay with better color (because cmos can produce better color than a ccd sensor) we could use
the altasens sensor used in the Silicon Imaging 2K camera and in the sumix discussed in this forum:
http://www.dvinfo.net/conf/alternati...1080-cmos.html

some more info about sensor that could be used on the elphel camera are here:
Sensors table - ElphelWiki

Oscar Spierenburg August 1st, 2009 09:16 AM

2/3" and an optical viewfinder
 
Hi everyone.
I think this should be an interesting thread. Imagine what we have in our hands when we have a 2/3" sensor on the camera.
One of the most exiting solutions it will offer is the optical viewfinder. It will immediately solve the real time preview issue and outdoor shooting problems (sun and LCDs)
I recently noticed the OVF from P+S technik for the SI-2k: P+S TECHNIK | Professional Cine Equipment Manufacture | SI-2K Digital Cinema Camera System - Viewing Options
This is an optional part, so no problem adapting it for Apertus. Another low budget option is an old Pan-Cinor (or similar?) Bolex Collector | Lenses | SOM Berthiot 1960s from a 16mm Bolex camera. These lenses have the beam splitter behind the iris, so you have a clear view even when stopping down the lens. I'm not sure how this would work on the P+S technik viewfinder.

So the main question remains: how to get a 2/3" sensor on our camera :-)

note: we should discuss this with the next Elphel model in mind, which is in a far stage of development.

Bob Hart August 1st, 2009 11:08 AM

Oscar.


Is the Pan Cinor zoom Super16 or 2/3" sensor area size through its entire zoom range?

The Angenieux 12mm x 120mm zoom with viewfinder for early non-reflex CP16 news film cameras is supposed to be able to cover Super16mm for generous parts of its zoom range. There were quite a few cameras made using this lens. Like many Pan Cinors, it fits up the early CP16 to a C-mount via an intermediate mount.

The Angenieux 10mm x 150mm covers Super16mm for parts of its zoom range but vignettes. An example can be seen here :-

YouTube - HM BARK ENDEAVOUR REPLICA LAUNCH 1993

It is also not as sharp as the 12mm x 120mm.

The side assembly which passes the image from the splitter in the lens barrel is a whisker too close to the SI2K camera head body for it to be used, even with the IMS ring removed.

With your camera, you may be able to design around it.

There is also an Angenieux 9.5mm to something zoom with the same side-finder but I think this may be too confined for Super16mm.

I guess you know about these but I thought I would mention them anyway.

Biel Bestue August 5th, 2009 07:32 AM

what do you guys think of the next invention at cinevate?

Cinevate - Home of the Brevis35 • View topic - Cinevate's B4 mount 2/3" relay lens...what?

i know this is for a 35mm adapter and has nothing to do with "sensors" but since you're talking about 2/3'' optics...

----------------

anyway, what i'm most interested at the moment is not about resolution (i'm ok at 720p) but what i'm mostly looking for is light sensitivity, and latitude, but light sensitivity is for me the most important thing, better light sensitivity means a better environment to be able to close the shutter, and that in the long run means: better "look" of the images, a lot more of control over how the camera captures images, now imagine what better sensitivity could do coupled with better latitude!

what do you think of this design for greater light sensitivity: http://en.wikipedia.org/wiki/File:RGBW_Bayer.svg this is the "panchormatic desing" for bayer filter at photosites

what are the chips that come with such "panchromatic" filter design? are only Kodak's? can it be used without breaking any pattent?

they cut in half the color resolution (4:2:2 at sensor) but they give a lot of light sensitivity, anyways since there is two designs for "panchromatic" cells (the other being this: http://en.wikipedia.org/wiki/File:Ko...W_patterns.svg ) the one that I posted up there at least has the same amout of pixels for every color, wich means better light sensitivity at no color cost, becaus there aren't double green pixels compare dto red and blue, yet still keeps the white pixel!

anyways, wouldn't this, coupled with pixel binning, get the color resolution back to 4:4:4?

i'v never seen discussion about those designs here at dvinfo! what do you guys think of panchormatic pixels? are they available?

Oscar Spierenburg August 9th, 2009 06:57 AM

Quote:

Originally Posted by Bob Hart (Post 1179571)
Is the Pan Cinor zoom Super16 or 2/3" sensor area size through its entire zoom range?
...
The Angenieux 10mm x 150mm covers Super16mm for parts of its zoom range but vignettes. An example can be seen here

YouTube - HM BARK ENDEAVOUR REPLICA LAUNCH 1993

Bob, I'm sure my Pan-Cinor is not for super16, but it was only 25 euro, I bought it just for testing purposes :-)
I also read somewhere on the forum that someone put an Angenieux on the SI-2K. Any test results yet?

Bob Hart August 9th, 2009 12:06 PM

I shoved a Pan Cinor Som-Berthiot 17mm-85mm zoom on front of a SI2K in 2048 x 1152 (2K) setting and this is what I got at f4 with a bit of gain in the low light I was in.

It seems that it does not vignette through the whole zoom range but there is a bit of corner/edge brightness falloff which increases towards the end of the zoom range and wider than f4.

My test was not valid as the collimation is off. I was not able to get the lens to seat down to flange due to the C-mount in the lens turning before the thread went fully home. Sharp focus on a target chart at approx 1M was 100ft on the focus ring

I would be reasonably confident you would get a 2/3" coverage with this lens but no guarantees as my test was very quick, and imprecise.

The side-finder clears the IMS ring on the SI2k but does not clear the body of the recorder unit so it is not usable on the full camera with the side-finder in place but would be usable on the Mini head with remote cable.

A frame grab is here :-

PAN CINOR SOM-BERTHOIT 17mm - 85mm ON SI2K By Bob Hart On ExposureRoom

Sorry - can't post grabs direct to dvinfo any more. Internet Explorer 8 is an unmitigated mess and gives me this cute message about having encountered a problem and must shut down - so much for upgrades and improved versions.

I have not tried the 12mm x120mm but the 17.5mm - 75mm "compact" which is being sold as new-old-stock (left over from a shortened production run of the CP-GSMO film camera) works on the SI2K.

The test is here :-

ANGENIEUX 17.5 - 70 "COMPACT" ZOOM LENS TEST ON SI2K CAMERA. on Vimeo


The vendor is not a dvinfo.net sponsor but is as far as I know the only source so I'll take a chance and post the name - pre-emptive apologies to Chris if I have sinned here :-

Vendor - Whitehouse Audiovisual - Contact. Ken Hale. You'll find the lens listed on that famous auction website under a search for Angenieux zoom.

Sebastian Pichelhofer August 15th, 2009 05:42 AM

New CCDs
 
There is a line of very interesting CCDs from Kodak coming soon. I do not have a lot technical detail yet but they will all feature (very large) 5.5x5.5µm pixels and will come in 1,2,4 and 8 Megapixels with the same pin interface. At 2 Megapixels (FullHD) this equals 2/3" optical format, so the 4 Megapixel one will have double the area of that. Framerates are 60fps at 2 Megapixels 30fps at 4 Megapixels and 15 fps at 8 Megapixels.

A CCD would finally eliminate all concerns about rolling shutter and will be especially at this pixel size very very light sensitive with low noise. But of course there are also the disadvantages of the much higher sensor cost, lower dynamic range (the KAI-02150 has 66db, the currently used Aptina one claims to have 70-74db depending on binning mode) and the increased power consumption which might even lead to requiring cooling.

Another big plus for Kodak is that they do not require their customers to sign any kind of non-disclosure-agreements and just publish all datasheets and measurements without access restrictions on their website. Most other image sensor manufacturers see their product documentation as trade secret and require every customer to sign a contract (NDA) with them that forbids them to publish anything sensor related. This is big issue for an open hardware project.

What do you think?

Guillaume Tisserant September 18th, 2009 05:12 AM

Is there new about choice of the new sensor?
Does the line of Kodak CDD is coming?
How much is it?
There is many parameters to consider :
Exact dimension of sensor (Interesting thing is we can use directly lens for 35mm in the camera.)
Dynamic.
Sensibility.
Definition.
Price.
framerate. (for me 25 pfs is enough, but perhaps somme people need more )
I think the non existence of non-disclosure-agreements is a very good point to.

I think it's important to do a table of sensor with all this informations, and comment about produces before do choice.
(Perhaps we can just complete this table Sensors table - ElphelWiki )
Where informations about sensors can be found on the net?

(Excuse by bad English, pehaps I have used bad word or bad formulation, I'm French)

Guillaume Tisserant October 8th, 2009 01:18 PM

I have another question :
Is it possible to make the sensor?
We can see many new camcorder maker...
Does all buy their sensor?
Or part of them make their sensors theirself?

Sebastian Pichelhofer October 9th, 2009 01:41 AM

Quote:

Originally Posted by Guillaume Tisserant (Post 1367972)
Is there new about choice of the new sensor?
Does the line of Kodak CDD is coming?
How much is it?
There is many parameters to consider :
Exact dimension of sensor (Interesting thing is we can use directly lens for 35mm in the camera.)
Dynamic.
Sensibility.
Definition.
Price.
framerate. (for me 25 pfs is enough, but perhaps somme people need more )
I think the non existence of non-disclosure-agreements is a very good point to.

I think it's important to do a table of sensor with all this informations, and comment about produces before do choice.
(Perhaps we can just complete this table Sensors table - ElphelWiki )
Where informations about sensors can be found on the net?

(Excuse by bad English, pehaps I have used bad word or bad formulation, I'm French)

Hello Guillaume,

Completing the Sensors table - ElphelWiki is definitely a great idea. Currently we do not have more information about the Kodak sensors than what they publish on their website.

Manufacturing your own image sensor is possible but will cost you roughly a few hundred thousand dollars, maybe up to a million depending on how many changes or improvement steps you want to incorporate into the final design.


All times are GMT -6. The time now is 12:24 AM.

DV Info Net -- Real Names, Real People, Real Info!
1998-2024 The Digital Video Information Network