Ikonoskop A-cam dII at IBC2008

[Bill Koehler]

Originally Posted by Noah Kadner

Kinda leery of that whole thing- what's the tech support/third party NLE support likely to be like?

-Noah

Adobe Premier Pro.

It's Adobe that's pushing CinemaDNG, so this should be fairly obvious.
With a camera being released designed around DNG, it should also seem obvious that the CinemaDNG spec. should be really, really close to being finalized.

[Bill Koehler]

Originally Posted by Kurth Bousman

.....actually I believe I read November we will have the details of the new scarlet

Given all the noise there's been about Scarlett before, during, and after the latest changes, I'll start believing it when it is at the same level of reality as the a-cam dII.

[Brian Drysdale]

I don't believe that Ikonoskop consider their product as anything other than a niche product for the professional market. Unless the new Scarlet becomes a much smaller, somewhat different creature, the Ikonoskop will have its own unique selling points which will be interest.

There are car manufacturers who produce small numbers of vehicles each year, but they still have their place along side the mass production manufacturers like Ford and quite a few produce better cars than Ford - if much more expensive. All have their place in the market and are just as valid as each other.

[Chris Hurd]

(Thank you Brian)

Can we keep this discussion on topic please? We already have a RED forum. Thanks in advance,

[Paul Curtis]

Originally Posted by Chris Hurd

Slightly off-topic... this is the point where I insert my rant about the uselessness of measuring image sensors with imperial units based on an equivalency system held over from the old tube camera days. The industry needs to start expressing chip sizes with the metric system by stating their *actual* dimensions in millimeters, instead of their equivalent tube size, where 2/3rds is not really 2/3rds.

Ikonoskop has a chance to show some leadership here...

Amen to that. I've been away for a little while so missed most of this but based on the sensor size then i expect it to have slightly less DOF than the 2/3rds range. Based on my research for a custom camera project i did, the 2/3rds size is around 8.8mm across and this is 10.6mm which is also the same for 16mm (not S16 though which is 12.52mm). f1.6 on this would have the same DOF as f1.1 on 2/3rds which would be the same as f4.5 on 35mm. (These figures are all from some spreadsheet i made a while back and i don't have the original spec sources handy - please let me know if someone thinks im wrong)

I hope Kodak have improved vertical smearing because that was the biggest issue with the previous generation in a real life situation. Other than that i found the results from the previous sensors really nice.

This is exactly what i've been waiting for too. The only caveat about this i have is that it is a single sensor bayer and you cannot change the analogue gains for RGB separately which means that colour temperature balancing isn't as good as 3 chip solutions (i.e. you cannot analogue gain just the blue because there is only one sensor). The fact that you can get the full 12bit range helps here although a lot of that could be noise.

As simon says the workflow is ideal, and all you want is to capture the raw. I marvel at how little i use the controls on my EX1. Turn all the electronic crap off and it produces lovely images :)

Are there any images or footage online yet?

cheers
paul

[Kurth Bousman]

It seems that the Scarlet comparison is an obvious one to make -

ProVideo Coalition.com: FRESHDV by Matthew Jeppsen & Kendal Miller

and also the comparison between the two companies and their capabilities seems relevant.

[John Sandel]

Originally Posted by Paul Curtis

The only caveat about this i have is that it is a single sensor bayer and you cannot change the analogue gains for RGB separately which means that colour temperature balancing isn't as good as 3 chip solutions (i.e. you cannot analogue gain just the blue because there is only one sensor). The fact that you can get the full 12bit range helps here although a lot of that could be noise.

Paul, I understand most of this. By "single sensor bayer" do you mean "a single sensor [as opposed to 3 separate sensors] which uses a Bayer filter"—?

(As: Bayer filter - Wikipedia, the free encyclopedia)

If I have you right, can you describe the basic differences, on this issue, between a sensor like the Ikonoskop's and, say one of the new video-enabled dSLRs that have been burning up the various fora (Nikon D90, Canon 5D2)?

>NOTE: Hi, Chris! I'm not trying to change the topic. My main interest in this thread is the A-Cam & its place in the current ecosystem of cameras.<

I.e., if the analogue gain on X chip's color-channel can't be changed separately, and the result is unwanted noise, is that a problem on the larger chips in the dSLRs? Or are they so big they can hide/control the noise somehow?

If the dSLR's larger chips suffer from such noise as you point out, would it be fair to say—though you may've seen none of these cameras in person—that the Ikonoskop A-Cam's sensor would be intermediate, in the amount of noise, between a vidcam with a smaller sensor and such a dSLR?

Am I clear?

[Paul Curtis]

John,

Yes, a single sensor.

The main difference between the sensor in the ikonoskop and the dSLRs is that the ikonoskop is CCD based, dSLRs are mainly CMOS. The CCD gives a global shutter (so no skewing and wobbles) but the tradeoffs are usually lower resolution and frame rate. But for 1080p (as opposed to dSLRs multi megapixel resolutions) this is fine. On a personal level i've found better colour reproduction from CCDs, i've found that most CMOS go through a lot of colour correction to get a decent image. Of course technology marches on and these differences will eventually change.

For colour balancing (or white balancing) you want to be able to increase gain of the various RGB channels. CCDs can have analogue gain increased on the chip but only for the whole chip. So if you wanted to increase blue to get a better response in tungsten then on a single chip you can't without increase RG as well. (I believe it is technically possible on a CMOS to increase gain on a per pixel basis but i've not first hand seen this being done).

However the sensor in question outputs 12 bit raw and so there is some latitude in there for white balancing after capture, depending on how noisy the sensor is. I would say that you're not going to see 12 bit of clean colour range by a long shot though.

I like this camera because it's simple and does what is needed. The sensor is a known design but has had many iterations of improvement. It has big pixels (good for low light) and good colour response. It's exactly what you need out of a digital cinematography camera.

60fps isn't so bad, things to look for would be differences across the image because to get 60fps off 1080p the sensor will have multiple taps that a read simulataneously. I think 4 in this case, which means each quarter of the image is running through separate onboard hardware on the sensor which if not carefully calibrated can result in slightly different images (especially as temperature changes for example). So you can look at the cost of the sensor but you cannot easily factor in the effort ikonoscope have done to get a working camera (with firmware). It really isn't as simple as get a sensor and rig some laptop and capture...

It will be interesting what ikonoscope have done with regard to noise reduction, calibration and bad pixel mapping. DNG supports bad pixel mapping and even black calibration data. They might have taken RAW to it's ultimate conclusion and actually given the sensor data warts and all in which case there would need to be clean up in post. Im not sure what tools are available to actually do this at the moment...

i hope this helps or is of interest
paul

[John Sandel]

Originally Posted by Paul Curtis

The main difference between the sensor in the ikonoskop and the dSLRs is …
[etc]
i hope this helps or is of interest

Both, thanks. I agree that Ikonoskop's A-Cam looks interesting—a welcome divergence from the usual consumer-influenced video camera design. The simplicity appeals greatly.

I'm concerned about vertical smear on the sensor but don't have enough experience with the subtleties of chip technology to worry any more than that. As far as I know, Ikonoskop haven't posted any footage or frames.

[Simon Wyndham]

If anything it appears that the only real worry with this camera will be possible high amounts of aliasing.

[John Sandel]

Originally Posted by Simon Wyndham

If anything it appears that the only real worry with this camera will be possible high amounts of aliasing.

Would that be because it's a CCD? Or does this particular chip have that repute?

[Paul Curtis]

Originally Posted by Simon Wyndham

If anything it appears that the only real worry with this camera will be possible high amounts of aliasing.

Have they mentioned whether they are using a OLPF on it? I assume they are of some kind but yes aliasing is a concern.

As you're getting the RAW bayer data off it then problematic scenes could always be debayered in different ways.

All CCDs have vertical smear to some degree because they are read out vertically and the charge can leak, the generation before this sensor had it quite badly. I would hope they had made some advances here.

cheers
paul

[John Sandel]

Originally Posted by Paul Curtis

Have they mentioned whether they are using a OLPF on it?

What's OLPF?

Originally Posted by Paul Curtis

As you're getting the RAW bayer data off it then problematic scenes could always be debayered in different ways.

Paul, do you mean "debayered [before the information leaves the chip]"—or "debayered [in post]"?

[Chris Hurd]

Originally Posted by Paul Curtis

Have they mentioned whether they are using a OLPF on it?

It's my understanding that everything has an OLPF, and not all filters are created equal... also not all processing is equal... so I'm not sure how a determination of aliasing, moire, noise etc. can be made without actually having seen the image...?

[Paul Curtis]

Originally Posted by Chris Hurd

It's my understanding that everything has an OLPF, and not all filters are created equal... also not all processing is equal... so I'm not sure how a determination of aliasing, moire, noise etc. can be made without actually having seen the image...?

You can get industrial machine vision cameras with and without filters (this is where most of these sensors are easily available to play with). The sensors work either way. And as you say you can have many types of filters - ones that combine IR suppression as well as OLPF (optical low pass filter).

We'll only be able to see when we can play with the cameras but i would assume that they've selected something appropriate.

cheers
paul