Sumix 2/3" 1920x1080 CMOS

Jason Rodriguez · January 10th, 2008, 02:30 AM

The SI supports c-mount . . . it actually has a universal mount system, i.e., it can take a variety of mounts from C to PL to F, etc. without having to recalibrate the mount when you exchange them.

Paul Curtis · January 10th, 2008, 04:13 AM

Jason,

Are you aware of anyone actually using c mount in production though? I'd imagine most would be using PL...

cheers
paul

John Papadopoulos · January 10th, 2008, 07:32 AM

Paul,

In c-mount, never buy anything expensive without seeing a sample and a photo of the lens with dimensions. Manufacturers rarely document the optical performance or the distortion and things like bokeh have to be seen.

The aps sensor is much larger than the 1" sensor so it needs a lot more glass and diameter to get a good result even at a smaller aperture. it's also a zoom.

If you check our samples on the other thread, all come from an 80 euro 1" prime (non megapixel) on a 2/3" sensor. There are much better lenses in machine vision. You can use them, it's just not very practical.

In most sensors and with most lenses f1 will not be very sharp. The practical limit is about f1.2. Cine lenses are not sharp wide open also.

The focal lenght doesn't change with sensor size because it's a true focal lenght. You will never see equivalent in 35mm terms in these lenses. To calculate the crop factor (the equivalent focal lenght factor) compared to 35mm still film just calculate 36/(horizontal size of sensor in mm). For super16mm it's about 12mm/(horizontal size) and so on. If you sensor is 6mm wide, it will give you half (6mm/12mm) the coverage compared to the same lens on super16mm or 1/6 the coverage compared to 35full frame still photography (36mm/6mm). For 50mm equivalent in full frame 35mm, you will need a 50/6=about 8mm with a 6mm wide sensor.

EDIT: The focus on the c-mount 2/3" and 1" primes I have seen is about 180 degrees turn and the iris is 90 degrees. They come with 8 aperture blades or more. They look much better than the typical 4 blade prosumer camera and sometimes the 6blade apertures on the professional zoom lenses.

Paul Curtis · January 10th, 2008, 08:34 AM

John,

Some good advice, thanks. And good news on the focus range, 180 isn't so bad. As you say it's been quite difficult to find image samples of various lenses.

I am aware that the focal length doesn't change just the field of view, im just using the wrong terminology here. About the widest i've found for 2/3rds is around is around 4.8mm (pentax) which would have the same fov as 24mm on 35. But nothing below that so far, at least on cmount.

It's an interesting idea, my only concern being that the resolution, once debayered, is obviously 'less' than 2k. Now is the colour depth and fidelity worth it compared to a 3-cmos EX1 with recorder off the HD-SDI output.

That is something i can't really tell yet. From you're project i guess the colour fidelity is paramount and i can see where you're coming from, but how does the CCD perform vs this Altasens CMOS (I believe you're using CCD?). Time will tell but there has been some stunning work from the SI cameras. How much of that is down to SIs tweaking and customising i don't know.

cheers
paul

John Papadopoulos · January 10th, 2008, 08:53 AM

If a sensor is 2/3" it's usually 9.6mm wide. This means the crop factor is 36mm/9.6=3.75. The 4.8 will be a 18mm equivalent in full frame or something like a 11mm on a crop factor Canon SLR. For 24mm equivalent you just need a 6.4mm.

The normal in 2/3" is a little more than 12mm.

Debayer gives you 80%. But do you know of any cameras that give more? A 1080p is usually 500 to 900 lines. You should worry more about pixel quality and lens quality and contrast, not resolution in lines. Those things are immediately obvious in all scenes, not only in test charts and some wide angle stopped down shots.

The samples of our project are all with UNCORRECTED sensors:) We haven't implemented colorimetry yet.

CMOS has some disadvantages, but the most important is camera motion. Limit it and it will do the job. We are only using ccd. CCD has smear but in good sensors it's not limiting as in small cameras. It will not appear in night shots from normal street lights etc.

John Papadopoulos · January 10th, 2008, 09:20 AM

Btw, we are not against CMOS. I actually contacted Farhad today on adding this camera to our line if it covers the requirements. It's a popular chip and perhaps we can make it look really good. But we want to differ in that we describe what's good or bad for each sensor technology to the customer so that he will make an informed decision. It's smear and extra cost vs motion problems and lower cost. A CCD camera head costs 5000 euro for a good resolution. An equivalent CMOS camera is much more affordable. This allows a cheaper retail price for our product but the customer has to know the differences.

Noah Yuan-Vogel · January 10th, 2008, 10:57 AM

paul, i picked up a schneider c-mount 17mm f0.95 lens like the one you mentioned. got it on ebay for under $20, and it is quite sharp even wide open but its a very small lens and it has just been gathering dust for a while. i recall there being something weird about the aperture though. i think it was like 3-blade or something, creating triangular bokeh. i can double check if thats what it was though, i havent used it in a while.

Noah Yuan-Vogel · January 10th, 2008, 10:59 AM

make that $1+$10 shipping. a pretty incredible deal. guess there arent that many people looking for c-mount lenses on ebay (yet).

Paul Curtis · January 10th, 2008, 11:44 AM

Noah, Sadly ebay in the UK doesn't have quite the same range! There's a cool looking Super Teagea 1.98mm for £1000! I'd love to see some imagery from that lens, perhaps stopped down it will be very nice?

John, as for the CMOS motion problems im well aware. But having said that none of the SI footage exhibits anything that i would consider an issue. And some of the examples (the one for film out from elite studios) are very shaky.

cheers
paul

Michael Maier · February 2nd, 2008, 06:48 PM

So what are the short comings of this camera? What is the catch?

Juan M. M. Fiebelkorn · February 3rd, 2008, 04:12 PM

Sumix camera in its actual incarnation isn't quite usable outside of industrial applications.It is not flexible enough.
Maybe they change that in a not so near future, but I fear it is not a tech issue but "a way of doing things".

That was the biggest advantage of SI.The understood that they needed to change things a little from what they were doing and... VOILA, they got it.

Michael Maier · February 3rd, 2008, 04:22 PM

Quote:

Originally Posted by Juan M. M. Fiebelkorn

Sumix camera in its actual incarnation isn't quite usable outside of industrial applications.It is not flexible enough.
Maybe they change that in a not so near future, but I fear it is not a tech issue but "a way of doing things".

That was the biggest advantage of SI.The understood that they needed to change things a little from what they were doing and... VOILA, they got it.

Yeah but for a price much, much, much, (did I say much?), much higher. So I don't think it even compares. My feelings are that unless I had the specific need for a 2/3" camera, I would rather buy a RED which is actually cheaper (body vs body).

Farhad Towfiq · February 4th, 2008, 12:13 AM

Juan,

The catch is that SMX-12A2C does not have standard manual control. Everything must be controlled from PC side using a custom software (yet.) This camera is designed to give the total control of the sensor to the user. The only limitation is that at 1080p 60 fps GigaE bandwidth does not allow all 12 bits. You must choose a table look up to transfer 8 bits to PC at 60fps. So far compression is totally lossless. We are working on a compression with some acceptable loss that can allow 12 bits at 60 fps also. Camera has rich API with memory that provides capability for skipping frames (this is a software feature being released soon.) One of our customers needed LDVS output of the sensor so they can use their own interface they decided to buy our camera and use our API to control the sensor.
This camera is as flexible as it comes, except for little GigaE bandwidth limitation. The price is low only because we decided to set it so. Our business model is to let the resale value to be above our price so our risk of inventory and production will be low.

Paul Curtis · February 4th, 2008, 03:01 AM

Farhad,

Do you have any more images or video from the head yet? And what sort of timelines are you running on with shipment and software updates?

many thanks
paul

Farhad Towfiq · February 4th, 2008, 05:27 AM

Paul,

We prefer that our customers making share images and video witch each other. It is also possible that we place one camera to be controlled remotely, so people can take turn in changing adjustments and experimenting. I discuss it with our support people.
We are shipping cameras now, but there is a 1-5 weeks lead time.

January 10th, 2008, 07:32 AM	#108
John Papadopoulos Major Player Join Date: Nov 2007 Location: Athens Greece Posts: 336	Paul, In c-mount, never buy anything expensive without seeing a sample and a photo of the lens with dimensions. Manufacturers rarely document the optical performance or the distortion and things like bokeh have to be seen. The aps sensor is much larger than the 1" sensor so it needs a lot more glass and diameter to get a good result even at a smaller aperture. it's also a zoom. If you check our samples on the other thread, all come from an 80 euro 1" prime (non megapixel) on a 2/3" sensor. There are much better lenses in machine vision. You can use them, it's just not very practical. In most sensors and with most lenses f1 will not be very sharp. The practical limit is about f1.2. Cine lenses are not sharp wide open also. The focal lenght doesn't change with sensor size because it's a true focal lenght. You will never see equivalent in 35mm terms in these lenses. To calculate the crop factor (the equivalent focal lenght factor) compared to 35mm still film just calculate 36/(horizontal size of sensor in mm). For super16mm it's about 12mm/(horizontal size) and so on. If you sensor is 6mm wide, it will give you half (6mm/12mm) the coverage compared to the same lens on super16mm or 1/6 the coverage compared to 35full frame still photography (36mm/6mm). For 50mm equivalent in full frame 35mm, you will need a 50/6=about 8mm with a 6mm wide sensor. EDIT: The focus on the c-mount 2/3" and 1" primes I have seen is about 180 degrees turn and the iris is 90 degrees. They come with 8 aperture blades or more. They look much better than the typical 4 blade prosumer camera and sometimes the 6blade apertures on the professional zoom lenses. Last edited by John Papadopoulos; January 10th, 2008 at 08:08 AM.

January 10th, 2008, 02:30 AM	#106
Jason Rodriguez Trustee Join Date: Mar 2003 Location: Virginia Beach, VA Posts: 1,095	The SI supports c-mount . . . it actually has a universal mount system, i.e., it can take a variety of mounts from C to PL to F, etc. without having to recalibrate the mount when you exchange them.

January 10th, 2008, 04:13 AM	#107
Paul Curtis Major Player Join Date: Mar 2005 Location: canterbury Posts: 411	Jason, Are you aware of anyone actually using c mount in production though? I'd imagine most would be using PL... cheers paul

January 10th, 2008, 08:34 AM	#109
Paul Curtis Major Player Join Date: Mar 2005 Location: canterbury Posts: 411	John, Some good advice, thanks. And good news on the focus range, 180 isn't so bad. As you say it's been quite difficult to find image samples of various lenses. I am aware that the focal length doesn't change just the field of view, im just using the wrong terminology here. About the widest i've found for 2/3rds is around is around 4.8mm (pentax) which would have the same fov as 24mm on 35. But nothing below that so far, at least on cmount. It's an interesting idea, my only concern being that the resolution, once debayered, is obviously 'less' than 2k. Now is the colour depth and fidelity worth it compared to a 3-cmos EX1 with recorder off the HD-SDI output. That is something i can't really tell yet. From you're project i guess the colour fidelity is paramount and i can see where you're coming from, but how does the CCD perform vs this Altasens CMOS (I believe you're using CCD?). Time will tell but there has been some stunning work from the SI cameras. How much of that is down to SIs tweaking and customising i don't know. cheers paul

January 10th, 2008, 08:53 AM	#110
John Papadopoulos Major Player Join Date: Nov 2007 Location: Athens Greece Posts: 336	If a sensor is 2/3" it's usually 9.6mm wide. This means the crop factor is 36mm/9.6=3.75. The 4.8 will be a 18mm equivalent in full frame or something like a 11mm on a crop factor Canon SLR. For 24mm equivalent you just need a 6.4mm. The normal in 2/3" is a little more than 12mm. Debayer gives you 80%. But do you know of any cameras that give more? A 1080p is usually 500 to 900 lines. You should worry more about pixel quality and lens quality and contrast, not resolution in lines. Those things are immediately obvious in all scenes, not only in test charts and some wide angle stopped down shots. The samples of our project are all with UNCORRECTED sensors:) We haven't implemented colorimetry yet. CMOS has some disadvantages, but the most important is camera motion. Limit it and it will do the job. We are only using ccd. CCD has smear but in good sensors it's not limiting as in small cameras. It will not appear in night shots from normal street lights etc.

January 10th, 2008, 09:20 AM	#111
John Papadopoulos Major Player Join Date: Nov 2007 Location: Athens Greece Posts: 336	Btw, we are not against CMOS. I actually contacted Farhad today on adding this camera to our line if it covers the requirements. It's a popular chip and perhaps we can make it look really good. But we want to differ in that we describe what's good or bad for each sensor technology to the customer so that he will make an informed decision. It's smear and extra cost vs motion problems and lower cost. A CCD camera head costs 5000 euro for a good resolution. An equivalent CMOS camera is much more affordable. This allows a cheaper retail price for our product but the customer has to know the differences.

January 10th, 2008, 10:57 AM	#112
Noah Yuan-Vogel Major Player Join Date: Dec 2004 Location: New York City Posts: 613	paul, i picked up a schneider c-mount 17mm f0.95 lens like the one you mentioned. got it on ebay for under $20, and it is quite sharp even wide open but its a very small lens and it has just been gathering dust for a while. i recall there being something weird about the aperture though. i think it was like 3-blade or something, creating triangular bokeh. i can double check if thats what it was though, i havent used it in a while.

January 10th, 2008, 10:59 AM	#113
Noah Yuan-Vogel Major Player Join Date: Dec 2004 Location: New York City Posts: 613	make that $1+$10 shipping. a pretty incredible deal. guess there arent that many people looking for c-mount lenses on ebay (yet).

January 10th, 2008, 11:44 AM	#114
Paul Curtis Major Player Join Date: Mar 2005 Location: canterbury Posts: 411	Noah, Sadly ebay in the UK doesn't have quite the same range! There's a cool looking Super Teagea 1.98mm for £1000! I'd love to see some imagery from that lens, perhaps stopped down it will be very nice? John, as for the CMOS motion problems im well aware. But having said that none of the SI footage exhibits anything that i would consider an issue. And some of the examples (the one for film out from elite studios) are very shaky. cheers paul

February 2nd, 2008, 06:48 PM	#115
Michael Maier Trustee Join Date: Apr 2005 Posts: 1,269	So what are the short comings of this camera? What is the catch?

February 3rd, 2008, 04:12 PM	#116
Juan M. M. Fiebelkorn Major Player Join Date: Jun 2004 Location: Buenos Aires , Argentina Posts: 444	Sumix camera in its actual incarnation isn't quite usable outside of industrial applications.It is not flexible enough. Maybe they change that in a not so near future, but I fear it is not a tech issue but "a way of doing things". That was the biggest advantage of SI.The understood that they needed to change things a little from what they were doing and... VOILA, they got it.

February 4th, 2008, 12:13 AM	#118
Farhad Towfiq Regular Crew Join Date: Sep 2007 Location: California Posts: 63	Juan, The catch is that SMX-12A2C does not have standard manual control. Everything must be controlled from PC side using a custom software (yet.) This camera is designed to give the total control of the sensor to the user. The only limitation is that at 1080p 60 fps GigaE bandwidth does not allow all 12 bits. You must choose a table look up to transfer 8 bits to PC at 60fps. So far compression is totally lossless. We are working on a compression with some acceptable loss that can allow 12 bits at 60 fps also. Camera has rich API with memory that provides capability for skipping frames (this is a software feature being released soon.) One of our customers needed LDVS output of the sensor so they can use their own interface they decided to buy our camera and use our API to control the sensor. This camera is as flexible as it comes, except for little GigaE bandwidth limitation. The price is low only because we decided to set it so. Our business model is to let the resale value to be above our price so our risk of inventory and production will be low.

February 4th, 2008, 03:01 AM	#119
Paul Curtis Major Player Join Date: Mar 2005 Location: canterbury Posts: 411	Farhad, Do you have any more images or video from the head yet? And what sort of timelines are you running on with shipment and software updates? many thanks paul

February 4th, 2008, 05:27 AM	#120
Farhad Towfiq Regular Crew Join Date: Sep 2007 Location: California Posts: 63	Paul, We prefer that our customers making share images and video witch each other. It is also possible that we place one camera to be controlled remotely, so people can take turn in changing adjustments and experimenting. I discuss it with our support people. We are shipping cameras now, but there is a 1-5 weeks lead time.