Ilmari Reitmaa

I recall DALSA presenting a tech paper at the IEEE AIS 1999 conference, describing a functional 4k x 2k 8.4um CCD sensor prototype with 16 x 37.125MHz outputs and a system board to go with the sensor; this being six years ago, I find it unlikely that they wouldn't have been able to work out a system to accommodate this particular sensor or a variant. Are you suggesting there are prohibitive restrictions other than the sensor itself, or could you please clarify? I probably have the paper archived, I'll try to look it up.

If anyone is going to attend the NAB Metadata Summit today (see News & Events thread), I would greatly appreciate if the key points of the discussion could be summarized here.

Genesis is 16:9 (or 1.78:1). D-20 in Video Mode is 16:9, in Film Mode 4:3 (more precisely 1.37:1). Letterboxing can naturally be applied to change the aspect ratio, as well as anamorphic lenses; note however that anamorphic lenses are somewhat problematic on Origin and Genesis because the sensors are already widescreen to begin with. Have a look at the tech specs for more info.

1 & 2) For D-20, the technical specs are available online at http://www.arri.com/prod/cam/d_20/tech_spec.htm There are different modes of operation for D-20, in "Film Mode" the camera is more or less 3k 12-bit raw Bayer-data, in "Video Mode" the capture is HD-resolution x 1.5 (to make up for information lost in Bayer-interpolation) and the output 1920x1080, either 4:2:2 YCbCr or 4:4:4 RGB. For Genesis, see http://www.panavision.com/product_detail.p...de=c0,c202,c203 Genesis is single-sensor 5760x2160 (non-square pixels) which is then downsampled to 1920x1080 10-bit logarithmic 4:4:4 Dual HD-SDI. Incidentally, strictly speaking 4:4:4 is not applicable to Bayer-pattern images as such, since full resolution color information isn't available before interpolation. Furthermore 4:4:4 gives an impression that the resolution for each is color channel is the same, which is not true for Bayer-pattern images. 3) I might add at least viewfinders (optical on Origin and D-20; electronic on Genesis) and data recorders (external on Origin and D-20, need cabling; dockable Sony SRW-1 on Genesis, extra cables not necessary). Perhaps available aspect ratios, too (D-20 is, in my opinion, the most sensible construction in this particular sense). 4) There's been a lot of speculation what "true RGB not Bayer-pattern" actually means. Probably it means just some other sort of colour filter pattern, possibly RGBRGB instead of GRGB; search earlier Genesis-threads on this forum for further discussion. D-20 uses a Bayer-pattern CMOS sensor.

Depending on where you're located, you can try these: http://www.pro8mm.com/index.html http://www.andecfilm.de/html/start_english.htm They've got package deals for transfers as well.

Thank you Mr. Schweickart, will keep that in mind; won't be coming over to LA just yet but I'll let you know once I have my schedules straightened out.

Actually just noticed myself that DALSA is also renting a Viper package, for $1700/day. http://www.dalsa.com/dc/dcc/products/produ...ails.asp?cat=HD

DALSA finally seems to have updated their digital cinema www-pages, e.g. equipment rental rates are available, confirming earlier $3000/day for Origin w/ data recorder, among other things. http://www.dalsa.com/dc/index.asp

Preparing a business trip to LA... Could anyone please point me to LA rental houses that rent the Viper? I was able to google up Plus8 and The Camera House; anything else?

Quite true, I did mean to refer just to the camera internals. However, I feel the above is maybe more of an issue of properly designing new gear to accommodate existing high-quality optics. It doesn't necessarily have to make up a new investment.

I really doubt this will continue to be so. The gear itself doesn't follow Moore's law as it isn't just ingrated circuits but the digital portion of the internals definitely do, and for SD and HD gear this cost will be nearing practically zero somewhere in the near future; it has got to reflect in the prices. So far the required digital hardware has been more in the high-end zone which probably amounts for the high prices. But I agree a dominant market position will help you choosing your prices. This is getting a little off-topic, sorry for that.

Hate to correct you there, Phil, but that's actually incorrect in the general case; the prism block colour separation isn't based on subtractive filtering but on reflective coatings, refractive indices and whatnot, thus preserving, theoretically, all of the incident light. There are certainly variations and combinations, though. Would work with a single chip configuration, though, i.e. replacing a Bayer-pattern (or which-ever-pattern) chip with a monochrome one. That, actually, might be feasible with the Genesis (might), or better yet, Kinetta, yielding about three times more incident light on the CCD.

This or that, I wouldn't think one should be particularly worried about it all. I mean, constant re-education and learning new stuff as going along is vital in very much any profession. The jobs, whatever they will be called, will probably be there, it's just a matter of filling the positions with the most competent people; people currently working in the film business are in an excellent position to re-educate themselves to be the most competent, should the need arise. In any case, streamlining the production process and cutting human resources costs, should this come to pass, is unlikely to be a generally bad thing in the long run.

For what it's worth, CCD spectral sensitivity does not go much beyond 1200nm (which is well into the infrared zone though), usually being most sensitive in the 600-800nm area. Human body heat radiation peaks at 9300-9400nm, so one shouldn't expect to be able to pick it up much, even with IR-filter removed. The Genesis uses, I would think, a Sony HAD CCD, so probably either Sony or Panavision will be able, if willing, to provide spectral sensitivity curves for purposes of comparison with, say, film stock sensitivity curves. (Still, probably wouldn't try to explain it quite like this to a four-year old...)

Allow me to sidetrack here a little... I must say I see it a practical impossibility that the price of any digital equipment could be artificially kept unaffordable. Anytime soon is naturally a relative term though, but considering the rate the price of digital technology keeps coming down, business competition will bring the prices down in this case also. Roughly, both hard drive storage capacity and integrated circuit component density keeps doubling every year and a half, data bus bandwith takes a little longer. A pessimistic prediction is that this development will continue for the next ten years. Today, you can get a 3.5" 300GB hard drive for $200 that would accommodate roughly 12min 30s of 4k 16-bit Bayer-data; in five years, the same $200 buys you 2.4TB or 1h 40min; in some ten years you get 19.2TB or 13h 20min. The cost of the in-camera circuitry will also drop accordingly. Not implying that there aren't any technical issues in putting the components together to make up a camera, but the price development of the underlying digital hardware (which today, along with R&D, makes up a major portion of the cost) is pretty drastic. Pardon me for getting technical :).

I wonder if anone happens to know of productions (either on the big screen or on DVD) that have used the 5229? (Akeelah and the Bee, obviously...)

P2P broadcasting (and P2P research in general) is on the increase, currently the emphasis seems to be on mostly promotional material... (I think Jackson's King Kong production diaries are available as BitTorrent downloads). Distribution-wise, definitely something to keep an eye on, lots of undiscovered/unexploited opportunities.

"Edo is a high-performance and high-quality image compositing application designed for maximum flexibility. Edo lets you combine and manipulate video files, 3D sequences and still images in a non-destructive way using a powerful node graph visual interface. The use of modern video card graphics processors (GPUs) enables true realtime performance, and support for industry standard file formats helps integrate Edo into many kinds of workflow." Features include tree-model compositing, 32-bit floating point per color channel and OpenEXR-support, GPU and Altivec-optimized CPU rendering, video input over Firewire, and a bunch of other features. The release is free of charge. http://www.anioni.com/edo/

"The optical prefilter reduces the aliasing created by the periodic arrangement of CCD photosites at 8.4um pitch. It is a layered birefringent crystal that splits each light ray into four rays, each offset 8.4um vertically and horizontally on the focal plane. " [image Resolution of the One-CCD Palomar Motion Picture Camera, Charles Smith et al., 2003] In effect, the low-pass filter disperses the light that would be falling on one pixel over adjacent pixels, thus creating an overlap and performing spatial frequency low-pass filtering. So actually not only does it blur the lines between the pixels, which usually cannot be resolved by lenses anyway, it convolutes adjacent pixels. So, yes, you do lose detail, however asking for any more detail would be a tradeoff. Don't know about the UV-filtering, though.

A while ago I remember seeing a thread discussing the actual resolution of current digital cinematography cameras but can't seem to spot that discussion now, so you'll have to pardon me for not replying to that particular thread. Anyway, there is a short technical paper on the resolution of the CCD of DALSA Origin, with modulation transfer function graphs and stuff. Check it out: http://www.dalsa.com/dc/documents/Image_Re...PTE_37_2003.pdf

Even at the risk of nitpicking, here goes... :P I wish to bring to your attention two constantly repeated terminological errors. 1. The unit of measure for thermodynamic temperature is kelvin, abbreviated K, not degrees Kelvin. Historically, temperature was actually measured in degrees Kelvin, but degree was officially dropped already in 1967. See http://www1.bipm.org/en/si/base_units/kelvin.html for more information. Notice also that this renders terms like Mired (micro reciprocal degrees) rather dubious; I've sometimes seen the abbreviation MK^-1 used instead. 2. The abbreviation of prefix kilo is k, not K (as K is the abbreviation of kelvin) Actually this is slightly more complicated and officially depends on the context (whether kilo is taken to be 1000 or 1024); however K is always incorrect, and while not necessarily official, the de facto abbreviation is k in all contexts (see http://physics.nist.gov/cuu/Units/binary.html for more information). Therefore, expressions like 4K and 5000KW are incorrect. While terminological correctness is secondary to getting your point across, any use of incorrect terminology does tend to undermine the credibility of the context where it occurs, at least in the eyes of an outsider (here a non-cinematographer). As these mistakes frequently occur even on the pages of such esteemed publications as the American Cinematographer magazine and the ASC Film Manual (including the 8th edition), I would like to see them corrected someday. Regards, Ilmari Reitmaa