NEW Digital Cinema Camera - Operational and showing at NAB

[Paul Curtis]

Thanks everyone, it's all sounding too good to be true. I look forward to seeing more, especially raw full frames, the CML guys will produce some great results im sure.

What kind of shutter adjustments can be made?

The 10 bit log is great news, perhaps the best.

I assume the WDR mentioned in the faq is a 12bit -> 10bit log curve created at capture time, rolling off highlights nicely? The WDR doesn't do anything special to the sensor?

the full production implies that prior to that there might be limited availability?

cheers
paul

[Steve Nordhauser]

Full production means that it is after the date we have been given for production releases of sensors. If we can get sensors earlier, we can have some cameras earlier. There will be a few being used internally and for a couple of showcase projects but that is all we know for sure. I wish I could say - thousands next week but I can't.

[Kyle Granger]

> WDR mentioned in the faq is a 12bit -> 10bit log curve created at capture time, rolling off highlights nicely? The WDR doesn't do anything special to the sensor?

Absolutely correct.

As for shutter settings, yes they can be made, but I don't have accurate current information. You may want to repost this question on the Sillcon Imaging Forums.

http://www.siliconimaging.com/phpBB/index.php

cheers,

Kyle

[Jason Rodriguez]

BTW, I will be starting up a blog for a "Virtual NAB", so you can see the team in the Adobe booth, and any other goodies we think up :)

Thanks,

Jason
Silicon Imaging

[Jason Rodriguez]

Originally Posted by Kyle Granger

like Steve said.... ;-)

> We run the 1080p at 2x the rate in the camera head and drop every other frame. This cuts any rolling shutter artifacts in half while keeping the bandwidth

Plus additional vertical blanking, higher pixel clock, further reducing any rolling shutter Useful for 24p.

Guys, I've done lots of testing, and if you download the new clips on line, especially the girl and mom walking through the door, you will see there is no "rolling shutter" problems.

Thanks,

Jason
Silicon Imaging

[Forrest Schultz]

Since the 1080p is captured at double fps and then frops every other frame. Then when doing 72 fps at 720p does the camera also run at double speed and drop everother frame? or does running the fps that high get rid of any rolling shutter artifact anyway?

[Steve Nordhauser]

Forrest,
Your second question is correct. Running at 72fps solves the problem without dropping frames - that that point it is about speed.
-Steve

[Jason Rodriguez]

Okay, to clarify the rolling shutter situation versus "dropping" frames, what we're basically doing is just running the chip as fast as we can. Whether we're doing extra vertical blanking, or skipping frames, we run the chip at 75Mhz or greater, which means the readout of the chip happens really fast. Because the read-out happens so fast, there is no time for the bottom of the frame to integrate longer than the top of the frame, hence no visible rolling shutter skew. Now there's always motion blur, but any camera has that.

Rolling shutter skew occurs because the readout on other CMOS chips is really slow as the electronic shutter goes from top-to-bottom. So at the time that integration ends, the chip is read-out from top to bottom . . . if this takes too long, then the bottom of the frame has captured light long after the top of the frame has ended light-gathering. If the read-out happens really quickly (as it is in our case), then the top and the bottom of the frame are shuttered almost together, and there's no skew.

The speed of the rolling shutter is determined by the Mhz you're running the chip at. There are two ways to crank the chip Mhz up . . . either run at a fast frame-rate and then skip frames, or increase blanking so that the rolling shutter has to travel more "virtual" lines in a given time-frame before it gets back to the top of the frame. The more lines it has to scan through, the fast it runs, and the faster the clock freq.

So if you see a CMOS sensor that is 40Mhz or lower (like cell-phone chips, consumer cameras, etc.), that's going to mean trouble for rolling shutter, because it means that the fastest the shutter can move across the chip is at a 40Mhz rate. The Altasens on the other-hand is now capable of up to 90Mhz, which is super-fast, and means no visible rolling shutter skew.

Hope this all made sense.

Jason
Silicon Imaging

[Forrest Schultz]

Thank you Steve and Jason. thats pretty cool, i wonder what mhz my cameras sensor is maxed at.

[Jef Bryant]

As long as all this double speed/dropping frames business still allows true 1/48th a second exposure at 24fps on the non-dropped frames, it's okay.

It's not actually making the shutter speed shorter than that, is it?

[Wayne Morellini]

Should be the same.

[David Newman]

The are many reasons professional cinematographers need larger sensors with higher dynamic range, 10-bit, with greatly reduced compression.

[Ram Ganesh]

whats the latest scoop on this? its already 2 hrs into NAB :)

[Jeff Morrissette]

I want to see a picture of this awsome camera!!!

8)

[David Newman]

It is there and fully operational.