Oooh, this looks nice

[SimonMW]

http://www.ikonoskop.com/dii/

S16 sized sensor, records RAW picture files, and preliminary price is 7k Euros. Takes S16 lenses.

[SimonMW]

In addition I have found that the camera comes with a f1.5 9mm lens, a battery, and one 80GB recording card. Basically everything you need for basic shooting for 7k Euros!

I'm liking the sound of this A LOT! I think I've fallen in love with a camera! The more I read about this thing the better it sounds. CineDNG format next year upgradeable via firmware. The camera even has an LCD on the side so that audio levels can be monitored. Yep, assuming there are no horror stories this will be my next camera.

So simple. Also because it isn't having to do any on board compression etc there is no need for cooling fans, battery power will be a cinch too and should last for a long time. If they incorporate an e-sata interface for fast footage dumping then it just starts to sound perfect.

[Alan Roberts]

With luck, I'll be testing it and lots of other newcomers in the next few weeks. I suspect that the sensor is 1920x1080, and almost certainly Bayer pattern, that would make it as good (or bad) as my A1). I'll know more in a few weeks.

[dominicwitherow]

WOW! That makes RED seem like a massive, lumpy, over-priced dinosaur! Looks like real film-making has just become truly affordable! :-)

[fuddam]

wow, mean machine :)

[Alan Roberts]

It's not in the same league as the RED, Ikonoscope is s16 size like the F23, RED is 35mm like the F35 and D20. That doesn't mean either is better than the other, just that they're different.

[SimonMW]

Even better is that it records uncompressed 48khz sound as well as timecode :) They really seem to have thought this one through, right down to having the LCD on the right side of the camera for directors and sound ops to see.

That makes RED seem like a massive, lumpy, over-priced dinosaur!

Yes. Actually this is really what Red should have been according to Jim Jannards original spec (small, light, inexpensive, and simple).

I could probably buy one of these with a couple of lenses and cards, and an EX1 for everyday shooting for less than the cost of even a 355 with a lens and batteries!

Alan, regarding the sensor size. Ikonoshop specify 10.6 mm x 6 mm. How does this compare with the actual size of so called 2/3" chips in normal video? I assume that normal 2/3" cameras CCD's are usually around 8mm x 6mm?

If this is a bayer patterned sensor, how would this affect the real captured resolution (usually stuff like bayer patterning goes over my head, but this time I'm interested!)?

[infocus]

Alan, regarding the sensor size. Ikonoshop specify 10.6 mm x 6 mm. How does this compare with the actual size of so called 2/3" chips in normal video? I assume that normal 2/3" cameras CCD's are usually around 8mm x 6mm?
My DSR500 manual quotes 9.6mmx5.4mm, so the S16 size are slightly larger.
If this is a bayer patterned sensor, how would this affect the real captured resolution (usually stuff like bayer patterning goes over my head, but this time I'm interested!)?
If it's a Bayer sensor of 1920x1080, the "real" resolution must be less than that, because of the way de-Bayering works, and the chrominance resolution will be worse than luminance.

A single unfiltered monochrome 1920x1080 sensor will give "real" 1920x1080 resolution. To get colour from the same basic sensor, you either have to have three of them and a splitter, or trade resolution for colour - which is exactly what the Bayer system does.

Note also that it won't be as sensitive as a 2/3" three chip system of the same technology.

[James lundy]

Try using that on a steadicam - even though you could use a cheaper rig, you'd still need to load up with quite a bit of weight for a smooth glide. :D

Anyway, it certainly seems like a nice camera - but for some reason the RED still seems more attractive to me.

[SimonMW]

Try using that on a steadicam - even though you could use a cheaper rig, you'd still need to load up with quite a bit of weight for a smooth glide

That's what weight cages are for.

[James lundy]

That's what weight cages are for.

Indeed, but it can still be a bit of a pain.

I much prefer using a bigger cam with my V25. Adding the extra weight to balance things out with my Z1 can 'sometimes' take a good deal more time. Of course this isn't always the case as I have the positions marked for balancing it out, but I imagine I would need to use 'even' more weight for this camera.

[Alan Roberts]

A single sensor with a Bayer pattern of photo-site* filters, of width W and height H (in pihoto-sites) has photo-site counts of:

R - W/2 by H/2
B - W/2 by H/2
G - W by H, but diagonally, not horizontally and vertically.

So, the R and B images have the "bandwidths" of 4:2:0 (the 2:0 part), while the G part is a lot harder to explain. G has full W resolution horizontally only if there is no vertical detail content in the image. Similarly, it has full H resolution vertically only if there is no horizontal detail content.

The best way to visualise this is to draw the frequency space as a 2-dimensional graph, horizontal horizontally, vertical vertically, with zero in the middle. The axes run from -W to W horizontally, -H to H vertically. The clean limits for R and B are defined by a rectangluar box from -W/2 to W/2 horizontally, -H/2 to H/2 vertically. The clean G limit is a diamond shape bounded by the points -W/0, 0/H, W/0, 0/-H.

Outside these clean limits zones, the sensor produces spatial aliases, detail moves the wrong way. Clearly, the limits for RB and G are not the same, only in the common area (half resolution) is the image clean. In the triangles out to -W, H, W, -H, there is coloured aliasing (because R and B produce aliases while G does not), and outside the G clean limit there is luma aliasing (because all of R G and B produce aliases).

It's very easy to see this using a coloured zone plate chart (which is why I produce and sell them). And you get a good idea of what bit looks like from my settings doc for the Sony A1 http://www.bbc.co.uk/rd/pubs/whp/whp034-add19.shtml.

So, for a sensor of W by H, clean resolution comes only from the central W/2 by H/2, 25% of the photo-sites. Not a good idea. There are better ways to get better results with only a slight increase in photo-site counts.

* in a 3-sensor camera, the photo-sites could be called pixels if their count numbers are the same as the video signal pixel and line numbers. A photo-site is a single charge well which intercepts light and produces a charge, becoming a single element of the image. Only when the photo-site accurately maps to an image pixel, should it be called a pixel.

[steve phillipps]

One of the major issues with the RED (for some users at least) is the 90 seconds or so start up time, wonder what this'll be?
Steve

[steve phillipps]

Apparently it's a Kodak Bayer CCD, video at http://www.viddler.com/explore/blacktar/videos/1/
Steve

[Alan Roberts]

But we still don't know the sensor photo-site dimensions, it matters :)

[steve phillipps]

According to a post on Reduser it's 5.5 microns square.
Steve

[SimonMW]

While I'm making sense of how the bayer pattern works (thanks Alan), I think there might be something cool about the framerate. In one video he was reluctant to say much, but I would imagine that it would be easy to be able to do speed ramping with this thing.

One of the major issues with the RED (for some users at least) is the 90 seconds or so start up time, wonder what this'll be?

I would imagine that it would be instant. Apparently the DII just has the framerate selector, an on/off switch, and record. No ISO rating selections, no shutter speed, nothing. It is as basic as possible.

Though I'm still figuring out if the lack of an electronic shutter speed will be a problem or not (ie will we get smeary motion as with other progressive cameras with no shutter?)

[Alan Roberts]

If tyhe photo-sites are placed on a 5.5 micron grid, then the sesnor must be exactly 1920x1080. And that means that it has exactly the same resolution and aliasing that my A1 has, which is truly nasty. That does not bode well at all.

The interview showed a side panel on which men us appear, I presume shuttering and such are controlled from there. But, as for ISO ratings, they are only a gain control, and since they are making much of the uncompressed RAW recording, all that should be done in post. the only thing you can't do in post is the shutter, that has to be done at the time of shooting.

[Daniel Browning]

But, as for ISO ratings, they are only a gain control, and since they are making much of the uncompressed RAW recording, all that should be done in post. the only thing you can't do in post is the shutter, that has to be done at the time of shooting.

Agreed. Unlike many CMOS sensors, CCD sensors don't have reduced read noise at high analog gain, so using it with RAW only throws away highlight headroom. (Some people think that it also reduces quantization error, but that would only be noticeable if noise was smaller than the quantization error, and it almost never is.)

[Alan Roberts]

Correct :)

[SimonMW]

Why do these new products always have a sting in the tail?