Kodak 1-inch CCD Cameras shootout at DVinfo.net

Anmol Mishra · March 28th, 2008, 08:54 AM

This is a comparison of 1-inch MVCs using the 1-inch Kodak Kai sensor.
The cameras are
Pike F-210c
Imperx LYNX IPX-2M30H-GC
Prosilica GE-1900C

Imperx and Prosilica also have CL and firewire models. The firewire based models are more expensive but there is the advantage of a single cable for power and data.

The GigE versions seem to be more reliable for data capture. As far as I know, Paul from dvinfo and Cesar Rubio from davidrubio3d are the only 2 people looking at these cameras for HD production..

My background is in embedded systems (software) and I am currently doing a VHDL course at Sydney Uni.

Both Paul and Cesar pointed out interesting stuff about these cameras.

Anmol Mishra · March 28th, 2008, 06:07 PM

Here is Paul's view

>>>>>>>
I don't know the imperx camera so cannot comment on that.

Prosilica does not do 12bit and it does not support onboard LUT tables
either which means you cannot really squeeze the range into 8 bit log.

But the prosilica does correct for dark noise and automatically adjust the
two taps to balance. Noisewise it's pretty good. Out of the box it has a
good image.

The basler does not do dark noise correction but does LUTS and 12bit. Im
hoping to test one of these in the next month but im in the middle of a
production right now so don't have much time.

The Pikes do not do dark noise either, and from Takes work on dvinfo the
pikes look awful (unless he has a bad camera head) but he's done some great
work getting the best out of it.

There doesn't seem to be a clear winner here and I really don't think most
people understand that there's quite a bit of work to do to get these things
working in a variety of situations. If there's no dark noise correction then
software has to be written to do this and sample black frames need to be
captured as the cameras warm up (noise changes related to temperature).

Streampix is okay for testing but in no way can I see it being used in the
field.

Davids stuff is all locked off from what I've seen. When he moves the camera
you need to see how the sensors perform with fine detail. The cineform RAW
workflow isn't ideal either plus the debayering used so far with it is not
very good (but they have other debayer options so hopefully that's a minor
point). As well as streampix you'd need to buy cineform too as streampix
requires the cineform RAW encoder which isn't supplied with. I used Prospect
HD anyway so am quite familiar with the workflow.

Also each sensor head will have a different IR cut off filter (if they have
one at all) and this can affect the image quite a bit.

All CCDs bloom and the KAI sensor is quite easy to bloom (vertical
streaking). Im undecided whether this is a show stopper for me, and whether
the sample I'd been using is good or bad. It's a difficult thing to judge.

You have to ask yourself what is it you're trying to achieve?

>>>>>>>>>>>>>

March 28th, 2008, 08:54 AM	#1
Anmol Mishra Major Player Join Date: Jun 2007 Location: Sydney Australia Posts: 401	Kodak 1-inch CCD Cameras shootout This is a comparison of 1-inch MVCs using the 1-inch Kodak Kai sensor. The cameras are Pike F-210c Imperx LYNX IPX-2M30H-GC Prosilica GE-1900C Imperx and Prosilica also have CL and firewire models. The firewire based models are more expensive but there is the advantage of a single cable for power and data. The GigE versions seem to be more reliable for data capture. As far as I know, Paul from dvinfo and Cesar Rubio from davidrubio3d are the only 2 people looking at these cameras for HD production.. My background is in embedded systems (software) and I am currently doing a VHDL course at Sydney Uni. Both Paul and Cesar pointed out interesting stuff about these cameras.

March 28th, 2008, 06:07 PM	#2
Anmol Mishra Major Player Join Date: Jun 2007 Location: Sydney Australia Posts: 401	Paul's findings Here is Paul's view >>>>>>> I don't know the imperx camera so cannot comment on that. Prosilica does not do 12bit and it does not support onboard LUT tables either which means you cannot really squeeze the range into 8 bit log. But the prosilica does correct for dark noise and automatically adjust the two taps to balance. Noisewise it's pretty good. Out of the box it has a good image. The basler does not do dark noise correction but does LUTS and 12bit. Im hoping to test one of these in the next month but im in the middle of a production right now so don't have much time. The Pikes do not do dark noise either, and from Takes work on dvinfo the pikes look awful (unless he has a bad camera head) but he's done some great work getting the best out of it. There doesn't seem to be a clear winner here and I really don't think most people understand that there's quite a bit of work to do to get these things working in a variety of situations. If there's no dark noise correction then software has to be written to do this and sample black frames need to be captured as the cameras warm up (noise changes related to temperature). Streampix is okay for testing but in no way can I see it being used in the field. Davids stuff is all locked off from what I've seen. When he moves the camera you need to see how the sensors perform with fine detail. The cineform RAW workflow isn't ideal either plus the debayering used so far with it is not very good (but they have other debayer options so hopefully that's a minor point). As well as streampix you'd need to buy cineform too as streampix requires the cineform RAW encoder which isn't supplied with. I used Prospect HD anyway so am quite familiar with the workflow. Also each sensor head will have a different IR cut off filter (if they have one at all) and this can affect the image quite a bit. All CCDs bloom and the KAI sensor is quite easy to bloom (vertical streaking). Im undecided whether this is a show stopper for me, and whether the sample I'd been using is good or bad. It's a difficult thing to judge. You have to ask yourself what is it you're trying to achieve? >>>>>>>>>>>>>