Sumix 2/3" 1920x1080 CMOS

Jay Burlage · March 18th, 2008, 07:10 AM

Yes much like a HDR photo workflow... Basically, change the exposure time frame by frame and save the brackets. Then in post I'll blend them back together into a radiance sequence.

So there is not that level of shutter control eh? I assume your controlling the shutter via software... I was hoping that a script could control the sensor frame by frame and dump the individual frames to a drive. Shucks. I'll keep looking!

Till then I'll still be interested it see how this camera works for you!

Daniel Lipats · March 18th, 2008, 09:03 AM

Quote:

Originally Posted by Jay Burlage

So there is not that level of shutter control eh? I assume your controlling the shutter via software... I was hoping that a script could control the sensor frame by frame and dump the individual frames to a drive. Shucks. I'll keep looking!

You can control exposure and sensor frequency through the software. I think its possible to do what you ask but I don't know what kind of framerate you will be able to get.

I posted the software that came with my camera in another thread. Here is the link if you want to play with it:
http://www.dreamstonestudios.com/per...2/development/

Jay Burlage · March 18th, 2008, 09:25 AM

Ah! All may not be lost... thanks for the input Daniel!

Yes the $2500 question is what sort of frame rate could be accieved. Like I said if I could get anywhere in the ballpark of 5-FPS I'd be so exited. With my current setup I get around 1-FPS so it takes 5 seconds to gather 5 brackets (albeit at 4K but I'd be very happy with a faster 1080)

If I could get 10-FPS I'd be all over this!

Farhad Towfiq · March 18th, 2008, 10:37 AM

Jay,

Very good idea. We will put it in the firmware. You would be able to choose double frequencies. Every other frame will be captured with every other frequency therefore having a different exposure.

We are also going for a 3D camera (several years project) with an array of 2x2 or 3x3 sensors (1/3" sensors) each with its own lens, filter, and now perhaps with a different frequency. The big advantage of a 3D camera is that you can control the depth of focus. In addition you have other information like optical flow that you can turn into brightness or color.

Jay Burlage · March 18th, 2008, 11:09 AM

Wow, great. I'm very inspired! So this would only allow 2 brackets correct? or could 3, 4, or 5 be stacked? Actually 4 would probably do quite well at 1.5eV spacings... I'm forgetting what the dynamic range of the sensor was estimated at.

I'm definately no programmer but after a quick look at the API my crazy idea looks promising.

Strap this 180d lens on it and "live" action sky dome light probes could be had... That's my target!

Farhad Towfiq · March 18th, 2008, 01:53 PM

camera supports 7 frequencies, approximately 10, 15, 20, 25, 36, 68, 74
How easily we can fit complex configurations in our FPGA is the question.

Serge Victorovich · March 18th, 2008, 02:24 PM

Quote:

Originally Posted by Farhad Towfiq

We are also going for a 3D camera (several years project) with an array of 2x2 or 3x3 sensors (1/3" sensors) each with its own lens, filter, and now perhaps with a different frequency. The big advantage of a 3D camera is that you can control the depth of focus. In addition you have other information like optical flow that you can turn into brightness or color.

Farhad, i can't find info about 3D cameras with 1/3" sensors on sumix.com...
These also GiGe cameras ? Working with Norpix & Cineform?
SMX-12A2C cameras have same ability for synchronisation master-slave as SI2K ? http://dvinfo.net/conf/showthread.php?t=116954

Thanks.

Jay Burlage · March 18th, 2008, 02:36 PM

Quote:

Originally Posted by Farhad Towfiq

(several years project)

There is your answer...

Gottfried Hofmann · March 20th, 2008, 11:37 AM

Quote:

Originally Posted by Jay Burlage

Yes the $2500 question is what sort of frame rate could be accieved. Like I said if I could get anywhere in the ballpark of 5-FPS I'd be so exited. With my current setup I get around 1-FPS so it takes 5 seconds to gather 5 brackets (albeit at 4K but I'd be very happy with a faster 1080)

Can you give some information on your current setup? Because that is exactly what I am looking for!

Jose A. Garcia · March 20th, 2008, 03:25 PM

Another question... Would it be possible to add some kind of viewfinder to this cam? I mean, are there electronic viewfinders with VGA in?

Daniel Lipats · March 20th, 2008, 05:19 PM

Quote:

Originally Posted by Jose A. Garcia

Another question... Would it be possible to add some kind of viewfinder to this cam? I mean, are there electronic viewfinders with VGA in?

A good viewfinder option for this type of camera is an optical one built into the lens.

Like this one:

http://cgi.ebay.com/ANGENENIEUX-9-5-...3286.c57.l1288

You can sometimes find one on ebay.

I'm using a 7" 1024x768 touch screen lcd. It works great as an interface with the computer, except its difficult to see outdoors. I have not yet looked at any anti-glare products. A hood is not a good option since its a touch screen.

If anti-glare products fail to give a good result I guess a black cloth will be the last option.

Jose A. Garcia · March 20th, 2008, 06:25 PM

I just bought a 7" touchscreen lcd too, but I've been asked if I could find a viewfinder, since one of the camera operators is very used to shooting that way.

I've seen OEM 2.5" lcds too. Maybe adding an optical viewfinder on top of it...

Farhad Towfiq · March 21st, 2008, 06:27 PM

Serge,

Jay is correct. First we implement multi sensor camera for increasing pixel count and field of view for our interferometers and let the expenses be covered this way. 3D cinema camera introduction will be possible only after USB 3.0 or 10 GiGE is practical. The reason I mentioned it here was not to tease but hopefully get feedback on lens requirements. This camera will have custom cluster lens, multiple lenses, one for each Black and white sensor. Lenses can be designed very low cost and super high quality when bandwidth of light is limited by filters. Every lens will have a narrow color filter, up to nine filters for extreme color restoration. From my experience with ultrasound lens design, you must give your lens a sweet range and certain differentiating capabilities. If you go for an overall performance then you may lose the specialness of the lens. So, my question is this: For cinema what are those sweet spots in lens characteristics?
Perhaps we must start a new thread for this discussion.

By the way, SMX-12A2C cameras can work as master slaves using their external triggering feature.

Jose A. Garcia · March 21st, 2008, 08:11 PM

By the way... Wouldn't it be possible to use 4 2/3" cmos sensors to create things like 4K resolution cameras with the equivalent of a 4/3" sensor? That way we would have more resolution and a DOF closer to 35mm.

But anyway, I guess there would be bandwidth problems and some parts of the scene wouldn't be covered due to the borders of the sensors...

Farhad Towfiq · March 21st, 2008, 10:02 PM

Depth of focus will depend on spacing between sensors and not on size of each sensor. A 3D camera assigns depth value to each point on the image. You can achieve multiple depth of focus by blurring pixels which are outside of the selected depth zones. You can even change digital magnifications for different zones. Sky is the limit when you have 3D information and optical flow in high resolution video.
Sensors can be chosen for sensitivity, speed, and cost and not for the size. Lenses will be low cost, small and high quality. This is the beauty of this approach That in such camera every lens is specialized, every sensor is specialized (sensitive to single color) and because of this specialization total cost will be low.
For today's camera a single lens must perform from red to blue color, this is about one octave or 100% bandwidth. Not even the most expensive lenses are perfect. They have many elements and many surfaces creating multiple reflections and loss of contrast.

March 18th, 2008, 10:37 AM	#169
Farhad Towfiq Regular Crew Join Date: Sep 2007 Location: California Posts: 63	Variable Exposure Jay, Very good idea. We will put it in the firmware. You would be able to choose double frequencies. Every other frame will be captured with every other frequency therefore having a different exposure. We are also going for a 3D camera (several years project) with an array of 2x2 or 3x3 sensors (1/3" sensors) each with its own lens, filter, and now perhaps with a different frequency. The big advantage of a 3D camera is that you can control the depth of focus. In addition you have other information like optical flow that you can turn into brightness or color.

March 21st, 2008, 06:27 PM	#178
Farhad Towfiq Regular Crew Join Date: Sep 2007 Location: California Posts: 63	3D Camera Serge, Jay is correct. First we implement multi sensor camera for increasing pixel count and field of view for our interferometers and let the expenses be covered this way. 3D cinema camera introduction will be possible only after USB 3.0 or 10 GiGE is practical. The reason I mentioned it here was not to tease but hopefully get feedback on lens requirements. This camera will have custom cluster lens, multiple lenses, one for each Black and white sensor. Lenses can be designed very low cost and super high quality when bandwidth of light is limited by filters. Every lens will have a narrow color filter, up to nine filters for extreme color restoration. From my experience with ultrasound lens design, you must give your lens a sweet range and certain differentiating capabilities. If you go for an overall performance then you may lose the specialness of the lens. So, my question is this: For cinema what are those sweet spots in lens characteristics? Perhaps we must start a new thread for this discussion. By the way, SMX-12A2C cameras can work as master slaves using their external triggering feature.

March 18th, 2008, 07:10 AM	#166
Jay Burlage Regular Crew Join Date: Sep 2007 Location: Ann Arbor, MI Posts: 39	Yes much like a HDR photo workflow... Basically, change the exposure time frame by frame and save the brackets. Then in post I'll blend them back together into a radiance sequence. So there is not that level of shutter control eh? I assume your controlling the shutter via software... I was hoping that a script could control the sensor frame by frame and dump the individual frames to a drive. Shucks. I'll keep looking! Till then I'll still be interested it see how this camera works for you!

March 18th, 2008, 09:25 AM	#168
Jay Burlage Regular Crew Join Date: Sep 2007 Location: Ann Arbor, MI Posts: 39	Ah! All may not be lost... thanks for the input Daniel! Yes the $2500 question is what sort of frame rate could be accieved. Like I said if I could get anywhere in the ballpark of 5-FPS I'd be so exited. With my current setup I get around 1-FPS so it takes 5 seconds to gather 5 brackets (albeit at 4K but I'd be very happy with a faster 1080) If I could get 10-FPS I'd be all over this!

March 18th, 2008, 11:09 AM	#170
Jay Burlage Regular Crew Join Date: Sep 2007 Location: Ann Arbor, MI Posts: 39	Wow, great. I'm very inspired! So this would only allow 2 brackets correct? or could 3, 4, or 5 be stacked? Actually 4 would probably do quite well at 1.5eV spacings... I'm forgetting what the dynamic range of the sensor was estimated at. I'm definately no programmer but after a quick look at the API my crazy idea looks promising. Strap this 180d lens on it and "live" action sky dome light probes could be had... That's my target!

March 18th, 2008, 01:53 PM	#171
Farhad Towfiq Regular Crew Join Date: Sep 2007 Location: California Posts: 63	camera supports 7 frequencies, approximately 10, 15, 20, 25, 36, 68, 74 How easily we can fit complex configurations in our FPGA is the question.

March 20th, 2008, 03:25 PM	#175
Jose A. Garcia Major Player Join Date: May 2007 Location: Sevilla (Spain) Posts: 439	Another question... Would it be possible to add some kind of viewfinder to this cam? I mean, are there electronic viewfinders with VGA in?

March 20th, 2008, 06:25 PM	#177
Jose A. Garcia Major Player Join Date: May 2007 Location: Sevilla (Spain) Posts: 439	I just bought a 7" touchscreen lcd too, but I've been asked if I could find a viewfinder, since one of the camera operators is very used to shooting that way. I've seen OEM 2.5" lcds too. Maybe adding an optical viewfinder on top of it...

March 21st, 2008, 08:11 PM	#179
Jose A. Garcia Major Player Join Date: May 2007 Location: Sevilla (Spain) Posts: 439	By the way... Wouldn't it be possible to use 4 2/3" cmos sensors to create things like 4K resolution cameras with the equivalent of a 4/3" sensor? That way we would have more resolution and a DOF closer to 35mm. But anyway, I guess there would be bandwidth problems and some parts of the scene wouldn't be covered due to the borders of the sensors...

March 21st, 2008, 10:02 PM	#180
Farhad Towfiq Regular Crew Join Date: Sep 2007 Location: California Posts: 63	Depth of focus will depend on spacing between sensors and not on size of each sensor. A 3D camera assigns depth value to each point on the image. You can achieve multiple depth of focus by blurring pixels which are outside of the selected depth zones. You can even change digital magnifications for different zones. Sky is the limit when you have 3D information and optical flow in high resolution video. Sensors can be chosen for sensitivity, speed, and cost and not for the size. Lenses will be low cost, small and high quality. This is the beauty of this approach That in such camera every lens is specialized, every sensor is specialized (sensitive to single color) and because of this specialization total cost will be low. For today's camera a single lens must perform from red to blue color, this is about one octave or 100% bandwidth. Not even the most expensive lenses are perfect. They have many elements and many surfaces creating multiple reflections and loss of contrast.