3 channel 36 bit 1280 X 720 low $ camera - Viper?

[Wayne Morellini]

Looking at the Canon HD product:

http://www.camcorderinfo.com/bbs/showthread.php?s=&threadid=98264

Same as I've seen here:
http://www.geocities.com/mammacow3/nab2005.htm

I don't think C likes us talking about unverified rumours here, so probably best to talk about i there.

Ron, excellent piont.

[Obin Olson]

movie cameras are using 6mp and up that is why bayer does not matter...my canon 10d is 6mp and it sure is plenty of resolution

[Steve Nordhauser]

Keep in mind that there are many different algorithms for Bayer filter de-mosaicing. Here is one article sifting though them:
http://www4.ncsu.edu:8030/~rramana/Research/demosaicking-JEI-02.pdf

I would think anyone in this industry should be watching the Bayer algorithm used. Since this is a post processing step, speed is not a big issue. I personally think that a good algorithm on a single chip camera might be close enough to be a good compromise versus a 3 chip camera (3x bus bandwidth, 3x recording speed, more complex capture). Is resolution better?

1280x720 is .92MB/frame
3 chip 1280x720 is 2.76MB/frame

1920x1080 is 2.1MB/frame

Of course if you convert the 3 chip to YUV in real time, the data size is reduced - but you are tossing data away. But any real-time processing is costly.

[David Newman]

Wayne,

It would best for discussions of Prospect HD to happen here. I much prefer this forum, after all I'm a moderator here (HDV Editing.)

CineForm is very much interested in very high resolution, frame rate and quality image gathering. Our current codec is can compress in real-time (software only) 1920x1080 4:2:2 at 10bit per channel in a visually lossless manner at 30p/60i (using a Dual Opteron.) Increasing compression performance and image depth is planned.

I can see the fit.

[Rob Scott]

<<<-- Originally posted by Wayne Morellini : What I'm suggesting would be a fraction the size of a laptop, more battery life, and cheaper (though more processing power, I don't know). -->>>

Problem is, you really need an array of drives to handle the bandwidth of this much data. If you're using affordable 7200 RPM drives, you'll need 3 or 4 in the array. If you're using VERY expensive 10,000 or 15,000 RPM drives, you might get away with 2 ... but at 3x or 4x the price.

Just as an example, the Kinetta (http://www.kinetta.com) camera's "magazine" uses a custom enclosure with an array of iPod drives -- 12, I think (!) -- to handle the bandwidth.

[Wayne Morellini]

Thanks David, I just thought to keep everything tight.

As you can tell there is much camera and computer hardware out there, and everybody has their personal likes and dislikes for editing aswell. I have suggested a general capture board on the hardware side (but nobodies biting on that idea) and on the software side plugins for capture and editing which you maybe able to help. What can your software offer in this respect, or will we have to buy a whole editing/capture package (I have not read the information on your site, I have been snowed under recently).

How much comrpession can you get with vissually near loosless, and how much true lossless?

The processing requirements you quoted, how much more improvement can you expect (I guess they are twin 2.8Ghz chips)? Are you using highly optimised C, or have you gone the final speed step to optimised Machine code, as these can give big gains?

I have not really told people too much about this, but there is a potential cheap technology out there that I want to research in the future that could reduce the pocessig requirement greatly.

As you can tell if people take to what we are expecting (and I might aswell go on the record as saying I will only take to a custom camera myself with specific price and visual performance advantages) then there will be a new market using a variety of datarates, resolutions and bit depths. Could this sort of flexibility be offered in your product for that market?

If people like your package what are the options and purchase options, and are there any attractive bulk purchase options that we could get into in private? Some people here would spend thousands, others would spend much much less.

I have allready located a Linux editing system that captures in HD, but it will be too much trouble for most, and a lot of power users probably would prefer something like your system.

Also there have been a few people wanting to find out more on your product here, maybe they can ask questions.

I would like to say I'm a bit of a novice to the issues in videography, but good on the technical side, so I might not understand all the issues.


Well I have to go to bed, so I hope I can pick up your reply tommorrow.

Thanks for turning up.

Wayne.

[Obin Olson]

I am talking with Norpix about a good quality bayer plugin for the norpix software..they are saying that if I buy the software they will get me a good quality plugin...I said if they show me a test image of the filter and it looks good enough I will buy ;) wait and see I guess...I got the 1300 camera today from Silicon Imaging...time to start testing i guess yaaaappeee!!

[Valeriu Campan]

Using a number of DSLR stills cameras in the last few years, I noticed that the RAW file format generated by them are approximative the same size in Mb as the size of the chip in megapixels. A 6 megapixels bayer pattern chip gives a ~<6 MB RAW file, that converted to RGB tiff will end up as a ~17MB.
A CMOS 1280x720 color bayer chip will give less than 1 MB/frame RAW file. If this data rate is recorded to disk as uncompressed, it needs ~24MB/sec transfer speed, which I can see easy to achieve. Of course, post processing is required to convert the RAW file to a format acceptable for HD with appropriate look-up tables. I think we need to look in creating an algorithm for converting the RAW file to an RGB color space and format, like a PlugIn in AfterEffects or a similar program.
Both Viper and Dalsa require a similar process and these cameras are NOT camcorders. There is a whole travelling circus of equipment and crew around them. The 4x2k Dalsa needs 1TB/15mins!!! That is not a joke and not cheap.

[Valeriu Campan]

Obin,
If you want to use your 35mm adaptor, I think you should mount the camera upside down and the images should appear in their correct position. I also saw at EdmundOptics some relay/macro lenses with C mount.

[Obin Olson]

big question, can I get a C-mount lens that has very few lens elements that's MACRO so I could use it to focus on a mini35 type adaptor for this rig I am building? if I could get mini35 style images outa this HD camera as far as I care this thing would be as good as film for many productions...4:4:4 with 10bit widescreen and any framerate you want!! so can this be done cheaply and with off the shelf lenses at a low f1.8-f2.x??

[Valeriu Campan]

Obin,
A bit pricey:
http://www.edmundoptics.com/onlineca...=2352&search=1

...A 35mm adaptor, though I would preffer lenses optimised for 2/3 or 16mm format:
http://www.edmundoptics.com/onlineca...=1459&search=1

...or, Rodenstock macros:
http://www.edmundoptics.com/onlineca...=1416&search=1

... and something cheaper, but I don't know about the macro performance:
http://www.edmundoptics.com/onlineca...=2411&search=1

Recently an old Kilffit macro for Bolex 16mm sold in Australia for about 600.00.

[Wayne Morellini]

Look at the XL1 adaptors, they use a reay lense to do away with the normal lense, you probably could do the same.

[Wayne Morellini]

<<<-- Originally posted by Wayne
Now I have another idea, something like a lenticular array could be designed to split the individual direction the image primaries are projected to be read by three chips or three areas of one ;) A bit of single direction image compression would do the trick. This is based on an idea I have had a long time ago (as well as a projection idea): a lenticular array laid over a single sensor could take all the light per pixel area, mix it and split the primaries to 3 ajioning pixels (acting like little prisms). What you get is a completely acurate (less any abreviations) colour, each primary is sent 100% to it's own pixel (no major filter loses), all on a single chip, at very cheap price. The other advantages is that you get near 100% pixel area coverage, not 70% max, like in cmos, so you reduce the fly screen and bayer motion induced luma/chroma artifacts, and increase the used light (if you design it to miss the interpixel spacing. The other benefits of these screens is that they could be used as projection screens for the adaptors. I think I gave up on the idea after the foveon came out. So would this reduce the costs? When done right (with a couple of other adjustments) you could deliver all the transmittable light from a MF lesne right down to 1/2in chip. I have other ideas I am wanting to work on commercially aswell.

thanks

Wayne. -->>>


I remember where I saw something that gave me tis lenticular idea. I seem to remember in the information on Sony's HAD/hyperhad?? that they used a small microlense screen over the sensor to concentrate light on the pixels pads. Now if we could use something like this with a single chip (does Bayer allow this) filter we could get competely accuate colour (less some fidelity). Actaully if they could be an array of splitting prisms then we could get almost 3chip like findelity and light gathering power from a single chip (and the mass produced price of such an array would be a very small fraction of a proper prism), 90% of te benefits for 10% of the cost. What do you think Steve I?

[Steve Nordhauser]

On the macro question, for film scanning you might want to try extension tubes or a bellows with a good lens. You are not concerned with DOF since the film plane if flat. I use a bellows and T mount lens all the time for experimenting with c mount cameras.

Valeriu is correct about image sizes. RAW data is one value per pixel - 8 bit is one byte, more than that occupies two bytes unless the capture device can pack data.

Data rates are a little different. There is horizontal and vertical blanking time so that 24MB/sec number is an *average* data rate. 24fps will take about a 30MHz pixel clock. Drives with 8MB buffers will take care of that at the storage, but you still have to watch the bus bandwidth.
Steve

[Adrian White]

Hi Steve, nice to see you've joined this forum. I'm sure you're contribution will be invaluble. Please email me with any further developments.

I came across yet another camera the other day.

SPECS: model IPC2M30HC
2 megapixel 8/10 bit HD camera
progressive scan 16:9
1920*1080 resolution
camera link interface
Programmable: resolution, framerate (will do 24p), electronic shuttter, long intergration, external trigger, pre exposure, strobe output, gain and offset!
(seems versatile!)
frame rate is programmable from 15-60 fps although it can only manage a max of 33fps at 1920*1080.

Uses a 1 inch progressive scan interline transfer ccd
can utilise C of F mount lenses

check it out at www.imperx.com

Please post you're opinions.