Well I was just looking through some of the Slicon Imaging pages, and think that the their sensors (like 12bit 3.2Mp) defintely are worth looking at for the camera (but we have to find out real life perfoamcne characteristis first, the fooage I've seen doesn't do anything for me yet, but great for a consuemr grade camera, hint, hint).

BUT LOOK AT THIS:
(Visually Loosless compression warning again, that is simular to what Mpeg and MP3 claim(chuck away the bits your not going to miss, but you do, especially when you have to process it, previouse post):

http://www.siliconimaging.com/SPIHT.htm#Lossless.

I must say beautifal, I have been waiting for wavelet to get this good for a while, I definitely can even see the difference, but with compression like this we can afford to 8Mp at low comrpession ratios (less than 20:1, only a stabb in the dark here).

Had a look at some of the specifications of the cameras, 2 lux at F1, 66db Dynamic Range, 67.7db SNR 4* gain for the 3.2Mp camera (that's just the ones I can read for the moment), not the best figures (the F! lesne would cost a bit), I expect a bit of niose with gain. I haven't looked properlly, but I think the chips closer to HD have much better figures. The Smal hi-res sensor had 120db Dynamic range, and just fantastic for difficult lighting situations. Now if Foveon could do an X3 chip with the performance of the Smal sensor, and the speed of the Silicon Imaging sensor, we would be set. By the way, who manufactures the sensors for them. Well I'll have to look at the rest of them after I get some rest.

They also talk about a number of interfaces we could consider using (industry standard low cost camera link, usb2.0, ehternet, ad probaly Firewaire somewhere in there). They also calim 150MB/s sustained serial ATA?? (which I, and others here, would think was impossible, maybe there is a special mutihead/platter drive that can do it).

Now the question on what codec to use with editors, if it's free and good editor I like it. Now some of the lossless codecs list support for editors. I think if you look at the microsoft mediaplayer hd link in the last post you will get some info. I think the Black Magic might be one, and the alparysoft is another. The comrpession routine above might also be good, if it can be true looseless at 3+:1. We can choose the level of compression we use.

( http://www.blackmagic-design.com/site/techsupport.htm )
( http://www.alparysoft.com/prod/compression/index.php )
( http://www.microsoft.com/windows/windowsmedia/howto/articles/capturingforhd.aspx )

Now I must retire for a while, spending this many hours on stuff like this is not fun. So guys feel free to discuss these things and decide.

By the way, definetly a 4:4:4 mode for keying special effects (like sci-fi).

Thanks

Wayne.

Wayne, thank you for your time and support. I just got the camera yesterday, I'm going to install USB 2 PCI card now to check out the quality of monochrome output. I'll upload sample frames to my site with a reference here.

Hi everyone -

I think we are the future of image acquisition! The new modular camera will allow us the lenses of our choice, the chips of our choice (CCD, CMOS), raw capture, and eventual output to the codec of our choice.

All this with repair and upgrade similar to a PC.

This is a link to what that camera will look like: http://www.kinetta.com/home.php.

That HD SDI would probably be very expensive, as apart from the cost of the HDSDI board for the mini computer (which may not have a free PCI slot) all the equipment using HDSDI will be very very expensive.

Those interfaces suggested on the Silicon Imaging site are probably the way to go (I don't think that HDSDI can keep up, I think it might be limited to match the speed of compressed HD, and wont do raw images, there is a more expensive technology that does that).

I have had a look at that camcorderinfo site, and it looks like that is the suggestion of virtually one or two people (not a terribly great interest there). Can we list out links to the recommended pc hardware,capture, editing software and the costs, I supposed there is room for people to choose Boxx, but I need something cheaper and smaller, preferably 4:4:4 and lossless (not near lossless, the price difference will not be too much, except for the extra HD space).

Here is a minature handheld PC gaming system (with embedded Windows XP) example runing at around 500Mhz. Via can go as high as 1.2Ghz in a design like this (higher in the future), anybody fancy a HD handheld ;):

http://www.ministrymobile.com/MoMAweb-technology.html

By the way there is another thread that has started looking at a film camera adaption, a home made camera thread, and the Jaun thread, they have links to many designs and sensors.

Now somebody mentioned getting a camera and trialing it, how is that going?

Wayne, if you have better suggestions that give us compressionless HD video for cheaper, by all means do what you need to to make them a reality for us. Most of what I know of this techno-subject (creating a cam from scratch), I have learned from here in the last few days.

Look at the previouse posts and links, do research, discuss the options with everybody and decide. Sumix wants opinions and suggestions, otherwise they will do what they think might be best, and it might not turn out exactly how we like. We have laid down options (and there needs to be further research and discussion) for the calcualtions on processing power, costs, and hardware needs to be made, and options selected. There are likely upto 4 times more chioces than those mentioned, and some of them are going to be as much as 100% better. www.google.com is a good starting place, and picking up more technical terms along the way. All this stuff has to be done, preplanning helps, when you get that baby it will want the right setup ready to be attached to it, otherwise things can get painfull. Well it's 4:37am.

Does anybody know about this new Final Cut Pro HD? It claims 10 bit 4:2:2 uncompressed 1080p at 90-160 MBps via PCI transport . . . and from SOFTWARE alone (No additional hardware)?

I'm not sure I buy this. I will say that if it does what it claims, it would limit the 4:4:4 idea, but sure gives us enough for Hollywood level acquisition.

Somebody tell me what I'm missing here:

http://www.apple.com/finalcutpro/

you have to have pcix and a P4 2.8ghz atleast for full on HD capture..no handheld will do

It was abit of a joke ;) You should be able to capture HD on much less than a 2.8, there is 2Ghz version of the via technology to come out.

Here is the new "2GHZ and above" VIA processor, based on.o9micron IBM tech. At the moment it mentions the 1Ghz version. I have only scanned it, but they make it sound like a Athlon/P4 killer, of course well see how it goes, ohh yes, and I think I have just found my new MB with something like this (but when the 2 Ghz version).

http://www.via.com.tw/en/Digital%20Library/PR040518EPF.jsp

We must remember that while present computer tech might not be ideal, it might be by the end of the year, or this time next year.

Thank you for the link, Wayne

what is goin gto change computer tech by the end of the year?

Moores law, power doubles every 18 months. The VIA chip is a little behind, I don't have their release date map, but I am hoping the 2GHz version might be here by the end of the year ;), and hopefully definitely by a years time.

That's an embedded naked chip not for general purpose use. Or is that what you are wanting?

The core tech should also be available in their standard processor line, and integrated Eden itx form factor MB/s. So it should be available to all soon enough.

Hey Obin,

When you posted:

"you have to have pcix and a P4 2.8ghz atleast for full on HD capture..no handheld will do"


. . . was that a response to my Final Cut Pro idea? Please excuse that I'm not sure pcix means. P4 I assume is Pentium 4. If this was a response to the FCP idea, please be more specific. (I'm assuming that FCP would be okay for straight shooting with little fx work, just maybe some re-contrast for darker shaddows if needed and color timing.

Signed,

The Relative Layman

Yes, I am interested too, it also sounds like a reply to my handheld computer come palm corder jest.

PCI comes in a number of high speed versions aswell as the common 32bit 32Mhz version (132MB/s data rate). There is PCI 66Mhz (264Mb/s), PCI 64 bit 66Mhz (512MB/s), one other I can't remember (apart from the embedded formfactor versions, and the PC-Card hybrid version), at arround 1GB/s I think, and PCI Express, Intels new Serial bus that is coming to more and more motherboards soon (I forget the top end speed of latter vesion but more tha enough to handle Ultra HD (32+million pixel screens).

I have picked up Linux Format April issue (No:52), with a section on video progams for Linux, whch I will hopefully be reading over the next couple of days. It should be just ready to go off your shelves in the States. www.sourceforge.com is also a good place to search for these freeware editors. I have also relocated some of the cheap coprocessing tech that I mentioned (took another 20-40 hours).

If there is a good freeware editor for Linux, there maybe a free Windows version, and these people might possibly get interested in supporting custom formats for new alternative HD cameras, like we are hoping for. Free capture and edting software could save some of us $1K-2K.

The advantages of a true, completely, loosless codec, is that it is easiest to add editing support for, as all decompressed footage can be edited raw, and recomrpessed, as many times as you like without generational degregaion.

Now that wavelet editor that Silicon Imaging is supporting on it' s site, apparently supports true loosless compression (though I am not completely clear on this yet). They are trying to get it to be some sort of standard, and the lossy comrpession it supports is quiet good. So it appraently will do Lossless, and good lossy in one package, so we could get the best of both worlds.

http://www.siliconimaging.com/SPIHT.htm#Lossless.

About loosless, I also include key framed losssles in my support, as long as it allows us to edit reproduced raw proficently.

<<<-- Originally posted by Wayne Morellini :
If there is a good freeware editor for Linux, -->>>

There is a free/open-source editor called Cinelerra for Linux.

I believe it supports 10-bit video using an uncompressed QuickTime file format. It also supports multi-channel audio (surround sound).

http://www.heroinewarrior.com/index.php3

http://cvs.cinelerra.org/

<<<-- The advantages of a true, completely, loosless codec, -->>>

HuffYUV yields about 2:1 compression in my experience, but it's not wavelet-based. I believe it is limited to 4:2:2 8-bit; however, since the source code is available it may be possible to develop a 10-bit and/or 4:4:4 version.

http://www.avisynth.org/index.php?page=HuffYUV

http://neuron2.net/www.math.berkeley.edu/benrg/huffyuv.html

i think all we need is an app like after effects / combustion / digitalfusion for COLOR work and then we can just output to SD at 4:2:2 or whatever and do our editing on a SD based system like FCP or Premiere pro or AVID...no need in my mind to edit HD 4:4:4 at all UNLESS it is for FILM work...I don't work with film I work with VIDEO and all I want is a way out of dv dvcpro and compressed HD for color and greenscreen work..after that give me 4:2:2 SD allday long ;)

Yeah, there it is pn page 53 (it also mentions Broadcast2000). The others mentioned , from my scan are:

Cinlera is reccomended as having "..power functions to take your videos to the next level.".

Unfortunately they don't give links to them all:

Kino "to start" with

Avidernux "quick modifications"

Main actor "for ... ease-of-use":
www.mainconcept.com

It will take time to research them though, anybody want to do it, I'm researching a couple of othr issues at the moment?

Anout the SD, I think we are all intrested in film/HD work here. But I have a recommendation, I have heard of a hidden secret port on the JVC GY-500/5000. I don't remember wherever it was supposed to give uncomrpessed video off the head or just 4:2:2 or 4:4:4?? video, but it does deliver superior video. I suggest the Juan Pana DVX100 project though.

Thanks

Wayne.

<<<-- Originally posted by Wayne Morellini : Kino "to start" with -->>>

As I recall, Kino is DV-only. I haven't tried using it.

http://kino.schirmacher.de/

Here's some detail on the capabilities of Cinelerra:

http://www.lmahd.com/cinelerra.html

You might take it with a grain of salt, since it's from a company that bundles the software with high-performance Linux software.

Well, I have gone to the site and found links to most of the other editors (nice of them) and this Nab Tribute:

http://enews.primediabusiness.com/enews/videosystems/bob_turners_the_cut/current#cinelerra

Yes, HD NLE, HDTV 6 channel sound, and true 24p (plus render farm etc etc), AND IT'S SO SMALL, AROUND 11 mb. Maybe these people would like to make a seperate capture backend for the Sumix camera :) with some pro colour/lighting effects for Windows and Linux (have to read this thing). But alas, no present Windows version :(. Maybe we can put together 20 of those VIA nano-itx 1GHZ MB, for a portable render farm with simular power consumption as our present systems ;) .

From the site:
------------------------------
"There are two types of moviegoers: producers who create new content, going back over their content at future points for further refinement, and consumers who want to acquire the content and watch it. Cinelerra is not intended for consumers. Cinelerra has many features for uncompressed content, high resolution processing, and compositing, with very few shortcuts. Producers need these features because of the need to retouch many generations of footage with alterations to the format, which makes Cinelerra very complex. There are many more standard tools for consumers like MainActor, Kino, or Moxy, which you should consider before using Cinelerra."
--------------------------

http://www.virtualdub.org/

Mainconcept has a DVCpro codec, and the MainConcept MPEG Pro plugin for Premier PRO will allow editing of HD10 footage. Present version of maainactor does not mention HD on it's page either.

<<<-- Originally posted by Wayne http://www.virtualdub.org/ -->>>

VirtualDub is an excellent video/audio processing utility, good for things like filtering, deinterlacing, smoothing, re-compressing, etc.

I use it to smooth/deinterlace video from my cheap DV camcorder to remove interlacing artifacts. There are some excellent plugins, including several "smart" deinterlacers that do a good job of detecting motion -- not just blurring two fields together (or removing one field entirely).

It can also remove 3:2 pulldown and convert between frame rates.

It is not a full NLE, however, and does not have color-correction features. It's also limited to 8-bit video (per channel).

----------------

You also might check out Jahshaka ...
http://www.jahshaka.com/

It's still under heavy development, but appears to be similar to After Effects. I don't know if it will support 10-bit video and/or color correction.

Supposedly, future plans include NLE features.

Understood, it was just an interesting editing (processing) technology.

<<<-- Originally posted by Wayne Morellini : Understood, it was just an interesting editing (processing) technology. -->>>

Yeah, VirtualDub is a terrific piece of software! I can't recommend it highly enough. There are also a couple of alternate versions (such as VirtualDubMod) that support things like Ogg and Matroska "container" formats and MPEG encoding.

Apologies if I came across as a know-it-all. I've been learning about this stuff for a few years and haven't been involved in many forums. I'm going to shut up now. :-)

No your fine, have a look at all the stuff I've put in here, unless sombody else does it, I may lend up doing it.

Keep researching and talking.

Thanks

Wayne.

As a side note, since someone mentioned editing and end product in SD, I say I'm tired of SD. There's only so far I can go with it as a filmmaker. It's one thing if I wanted to do TV or straight to rental or what not, but I DON'T. The only reason I do any of that is for money. I want to make major theatrical releases man, high concept story-telling, the biggest of the big. So for my 2 cents, I say let's try to achieve that here and then downscale if we feel like it. If I want SD, I can walk one room over into my office.

By the way, one of my concerns in a system for capturing is dependability and user-friendliness. Wouldn't it be risky to put a system together that the programing-unsavvy might have trouble with? I've spent YEARS working on a system that crashes ALL THE TIME and don't intend to get caught up in that again.

By the way, again, if someone could check out the possiblities of that Final Cut Pro HD on a standard Mac G5 and let me know what they think, that would be awesome.

Thanks

Yes, this is why I am putting such time into prompting people to think about working out a system, because one thing that will stop this from going large volume is te ease of use and reliability (apart from wanting to see the product succeed). Most people will go to a simular priced HDV then put up with hard to workout crashing super HDD HD. That leaves enthusiast and inde film makers, still profitable but 10 times less so (I believe). Most people want a plug and play (or install and play in this case) no brainer, reliable system, that is why I am hoping for good free Windows software aswell as Linux, because most people know how to handle the Windows beast allready.

If a flexible capture and/or editing system is worked out now, then it can be used for the Sumix or any other future project.

One note Steve, is that if Sumix could supply a seperate cheap, capture to disk board with camera controls, Lanc control port, and LCD support (with user configurable capture for any alternative input/camera module imaging format), in that price then all we would have to do is buy the drives, and connect them to the main computer for editing. It could be a bit much for them but well worth it for us if they feel inclined to attempt it. The right board could even be resold to people to put any module they like on it. That Russian guy has some of this in his freeware design.

It would be good to have completely lossless compression mode (that will allow 1080) and/or twin usb 2.0 (three for SHD). If a capture board i]s used then all we need is a straight "camera link" to the capture board (this would allow SHD imaging in future). While I think we will do fine with 720p we should aim for 1080i/p upgradability in the future (SHD would even be better).

If you could pass on these suggestios it would be good. Well, how is it going on the hardware front Steve?

Thanks

Wayne.

Is 1080 considered SHD? I thought it was regular. If it's possible to do a 1080 24p chip, I think it would be better, as the difference on the screen (given multiple edits and color correction and the like) would be quite noticable. Of course, I won't balk at 720p, but I figured I'd mention my preferences.

Man, I think connecting straight to the board would be awesome!!!

By the way, I know that everyone seems to be trying to get this whole system up and running for a few grand, but truth be told, that may not be realistic if we are talking a reliable user-friendly system on a short time line. I'd be willing to pay 10 grand if that provided ample storage and the camera with it. This would most likely clear the editing system as well, but I'd go up to 15 grand for the whole thing. I know that sounds like a lot to some, but we might be considering that if it takes us 2 years or more to finish the thing (even a year for that matter), that the dollars we saved really weren't saved at all, they were spent via time wasted. The way I see it is, I'll sell a few things here, a few things there to upgrade. That'll knock down the price down some for me.

The rest of the cash will be some extra time at work and eating out less for a while, or it will be a relatively small bank loan.

The way I see it is this. Most people pay big money to drive a lusher car, but won't pay money to do what makes them happy. What makes me happy is having a real career in filmmaking. If we can swing it without it breaking us, I think we should.

Laurence,

I agree, paying 15 K for such a system well worth it. But the fact is - a lot of people on this part of the forum can't afford paying that much. The good thing though, if Sumix ever gives us the camera, it will be a start - they might come up with a finished product (PC based, USB 2, 3 monochrome CMOSs cam) later or people on this forum will polish the solution.

The reliability of the final camera will be achieved this way or another, it's just time :(

Wayne,
I wish I had any news on the point, but Sumix went silent after I described what the camera should be. There was no confirmation, nothing. They could be developing the cam now or they might have desided the trouble is financially inattractive. I am sending them a letter immediately. Sorry for such an attitude, I was very buisy these two weeks. Among other things, I purchased a USB 2 1280 X 1024 webcam that does 11 frames at full resolution, hoping to reverse engineer it and get 24 frames from a subsampled mode.
The thing is, I turned out to be overestimating my hacking abilities :(
Besides, CMOS imager on the camera is bayer RGB, not monochrome as I hoped buying it.
Although I didn't waste my time ( I found software for synchronizing the 3 monochrome streams, premultiplying by corresponding colors and outputting the final .avi or frame streams - http://www.montivision.com), I am a bit dissapointed: now there's no choise, but to spend more on expensive industrial monochrome cameras if Sumix pulls out.

Only just joined this forum, based in the UK. I have the same dream of a low cost HD system as many others. I'm not nearly as technically savvy as you guys as I'm a writer first and foremost. Jesus, some of you guys sound like engineers! I've been following the new developments very closely, including kinetta, sony POV, JVC etc. over in the UK we're limited to the sony hdw750p, the sdx900, imx, cinealta. All are either too expensive or do not provide enough quality.

Last week I rang a guy called Steve Nordhausen who works for silicon imaging. Within 8-12 weeks time a new camera will arrive! Here is the spec so far, forgive me if they are not detailed enough.

SI-1920 HD camera. 1920*1080 3.2 megapixel at 24fps poss 23.976 as well. Single cmos chip. Will stream to computer (PC) poss USB2 connection altough serial ata configuration was hinted at.
Frame grabber software will be required. Apparently 10bit and 12bit solutions are available. Images will be uncompressed. Frame grabber will be required along with another piece of software which I'm still a bit hazy on. Ready for the best bit? Camera will be approx $4000!
Frame grabber $1500. Camera will apparently be compatible with 16mm bolex or sneider lenses along with others. Steve mentioned that a press release would be availble close to release which may be July!

I would appreciate hearing what you guys think about this. Personally I don't think it's vapourware as they already have a 6.6megapixel camera that does the same but at different frame rates. Look forward to hearing from you guys.

Adrian, thanks for posting, it was very vise to call the guys instead of dumb ass mailing :)
A few questions though - why would you need a frame grabber if you have USB 2 option? Is it that the data rate exceeds USB 2 top limit or is it just an alternative to USB?
Another thing is low light, what will be the dynamic range of the cam? Have they told you which CMOS is planned to be used?
I'm sorry for this flow of interrogativeness:), but the stufff you wrote about could be one piece of hardware!

thanks for response. I shall forward these queries onto steve. Watch this space.

Steve, no your fine, don't get over anxiouse, but if they have decided to bail out (it might have to go before a couple off montly board meetings first) they should have told you.
Laurence,

I think we are looking for realistically priced systems the price of HDV equipment ($5K or less) but a couple of levels up (lossless to disk, exchangable 35mm lense, really good sensor). At this price people can say, do I buy HDV or do I buy this. Even if people are grumbling at having to pay the HDV price, they can quickly see they are getting better value for the same price with the Sumix and pay for that anyway. Loosless to disk is also cheaper to make than highly comrpessed to tape mechanism (then you get the problem with buying extra HDD).

Steve, in Linux there is a group that does a video aplication layer used by USB drivers for webcams, ask them about this (they probably can direct you to the Linux driver writers), maybe they know of customisable Windows driver too:

www.thedirks.org/v4l

The problem with webcams, is that they probably use very cheap and nasty sensor chips, unless it is by chance something like the Smal sensor. It could also be a bit off putting to Sumix too, but I don't seriusly think there is much chance that any of these cheap wcams will have pro performance.

Adrian, unfortunately we are not engineers, though I am a computer Scientist that had Engineering freinds at Uni (one was one of the best young engineers in the country, and the top silicon engineer in the States graduated from the uni nearly 20 years before). If we were engineers we could do the lot ourselves.

I hate to say this, because the moderators might not like it. But look on the Silicon Engineering site in the news, there was a contract with Sharp for non HDV equipment.

The new chip, unless they use compresson they wont be able to get 1080i through a USB2.0 link. I have thought of the Serial ATA idea, and it could allow some by passing of the controller boards buss, and higher through put, but I have not suggested it because it is not standard in this application, and a better solution is needed, maybe the desktop version of PCI-Express serial bus. I'm not certain but I think there might be a usb like version to come. There is also a consumer AV cable standard that does around 5 Gbit/s per second (can't remember the name).

$4K+, yes there is a market, above the Sumix at that price, as you are talking of 2.5K+ more, and beyond the non pros threshhold of pain.

I have just received a Javascript notification that the limit of messages are around 1000 characters and I'm allready double that. I think one of my messages in another thread has dissapeared too. This is unrealistically small for such technical issues (10K would be better). If a lot of work was being done here then there would be many 1000+ messages floating back and forwards from time to time. Believe me I have aeen this in tecnical discussions (well there is another few hundred characters).

Steve, you mentioned an engineer, a while ago, called Vladimire, or something like that, that name sounds familiar. It could have been a video group, or the Misc (Minium Instruction Set Computer) mailing list I know it from. If he was in Misc, say hi from fellow member.

Wayne, thank you for typing that much :) I would sure use your advise and look for guys who deal with webcams but I came to realize that the thing I bought a few weeks ago won't be suitable for serious filmmaking. The hardware bayer pattern on the CMOS sensor ruins all hopes.
As for saying hi to the Russian guy, I would definitely do so if he replies to repeated inquiry I sent yesterday :)
You sure did a right thing making it off with a diploma in computer science ; ) It might be off topic, but what are you doing for living with that big diploma?
It's just I've been into this "cinema-SFX-help_yourself" utopia for the last 2 years, but having no proper tools for getting written scripts shot, I am gradually shifting towards electronics. The area is vast and amazing. Just curious how (if) you apply your knowledge towards sustainable living :)

Steve,

If you have scripts, I'd like to read them. We're always looking if they are of the standard length. No zombie movies though.

I am an software engineer, and I have done some embedded systems work. What I would love to see is an "open" hardware platform that people like me could write and adapt open source software for.

For example, the camera itself would have ...

- A standard CPU/memory/flash configuration
- A standard I/O port such as USB2, Firewire800 or Gigabit Ethernet.
- Support for a few selected imaging chips

If you could upload software into flash (Bayer algorithms, codecs, etc.) then the camera would be incredibly flexible.

We could define a standard protocol (or better, use one that already exists) for the video stream coming down the wire, then provide open-source tools to convert it into the necessary formats for the NLE -- AVI, Matroska, QuickTime, etc.

It's obviously necessary to work with any NLE out there -- Final Cut, Premiere, Avid, etc. in addition to supporting a free alternative (such as Cinelerra).

I for one would *love* to work on a project like this (though my time is already way overcomitted!). If it meant I could get an inexpensive high-quality "cine" camera, I would be really tempted to dig into the software.

[personal preferences]
1. I would like to see a Gigabit Ethernet port on the camera. This is quickly becoming a commodity and is even faster than Firewire 800. (Though I do not have specs on sustained throughput.)

2. To simplify the design I think the camera should not have audio capability -- external sync audio is very practical for filmmaking. This would also free up I/O bandwidth so the video would require less compression.
[/personal preferences]

I was following these threads with fascination in the last few weeks. I will briefly "chip in" in this thread.
I am a cinematographer and I have worked with all formats from 35mm to miniDV. I can see that a homemade HD cam is not far from being a reality. Even the so called Hi end products that are to be released (Viper, Dalsa) are not far as concept from what I've seen here lately. I welcome Rob's observations for a prototype and would like to add few things:
- the camera can be built around a motherboard with Gigabit Ethernet/FireWire
- a minimum 2/3" chip: to keep as close as possible to the depth of field and angle of view of bigger sisters/brothers we want to emulate. Larger chips also have bigger photosites and lower noise. The single CMOS (Bayer) color chip option is attracting some difficulties and postprocessing requirements like unsharp mask... and other things that I would like to have them done at that stage anyway. I can see it as bonus though: the raw files are the "negative" that we all want to have. Both Viper and Dalsa are using the same aproach. From my experience a 1.2 - 2 megapixel RAW file would generate an excellent image for the current projection environement. It will also offer an acceptable data rate to be handled by an inexpensive frame grabber card or even current logic boards and SATA+RAID drives. Don't forget that bigger files and a large data vault will make the post-production costs prohibitive for the indie filmmaker.
I have to go for now, I have an early start for the next day.

PS A link to an interesting chip. Check it out:
http://www.photonlines.com/English/Eframe_intro.htm

<<<-- Originally posted by Valeriu Campan : - the camera can be built around a motherboard with Gigabit Ethernet/FireWire
- a minimum 2/3" chip: to keep as close as possible to the depth of field -->>>

Here's an example of a embedded motherboard that might be suitable:
http://www.technoland.com/tl_embsbc845.htm

This board has built in Gigabit Ethernet and a CPU with enough power (hopefully) to handle the job. Also, it has built-in video, so it may be feasible to have full-resolution preview to a CRT or LCD monitor.

Here's the chip (the 3530 I think) used in the Kinetta camera:
http://www.altasens.com/products.html

The 3560 version (see above) has support for 1920x1080 progressive at up to 60fps. It can also be downsampled to 1280x720p at up to 120 fps (!).

I did some quick calculations, and it appears that Gigabit Ethernet can handle 1280x720 up to 36fps with no compression required. 1920x1080 would require compression of some sort (or multiple Ethernet connections) even at 24fps. Handling frame rates of over 24fps would be nice because I think overcranking is going to be a desirable feature.

(I am not a hardware designer, so I'm not sure if this combination would work -- how the motherboard would communicate with the chip, etc.)

Ohh I am such a complete cretin, I was over at the other thread and realised that if your using a normal one chipper than your raw image data rate falls to one third, because you actually have one monochrome pixel (behind a bayer filter etc) behind each scree pixel. So 2million pixels at 24fps, is actuall 48MB/s, going to 10 bits and we are still in the realm of possibility of fitting on USB2.0.

Anyway we don't want consumer grade single chip solutions do we ;)

Thanks for the compliments guys. I actual have a degree with distinction. But I have been too sick to make a descent living, I have had Ross River Virus, Glandular fever leading to CF, am becoming hyper sensitive to chemicals and a range of other things (this mostly started before I got the degree). So generally I'm not really upto stuff and am really out of the loop as far as software programming goes (I learn't the year before they started teaching Windows programming). Over in the States you have been more lucky, if you can do it you can get a job, over here you have to wave that little peice of paper mostly and then have the track record. But I have a creative mind and whatever it creates I tend not to forget, so I can workout some nifty stuff, like my OS design (which I often can't work on either). I also used to spend every Friday morning going through the electronics engineering industry periodicals at the Uni, and still dream of one day doing my own gadgets (requires a fair bit of study using my one real weak piont, maths). So if I was earning lots of dough, I'd be having a $20-$50K camera, and paying some engineer to design a cheap one for me to sell or something, but allass, it isn't so. The only camera stuff I may think of doing is a film lense adapter with some special features.

Ron, I hear you and your idea, and that is my reason for presenting so much information, that this camera could be received and setup on a preworked out syastem over days instead of months (and after that withn minutes). That we could have a backend system that could be upgraded and reused for any future camera. There is allready a Linux movement, and if you go to the Toas links above, there is an open content platform they have started, and I recently remember an open hardware platform, I think there is one for Power PC too, and I think MS might be running one of thm. But as far as the platform, we have the open work done on the Russian camera, and we have the cheap (compared to embedded boards) nano-itx pc standard by Via. You have sourceforge and in there you will find a wealth of open source video projects for Linux and Windows, and people that might want to help, or even join a project like you are interested in. We are probably almost at a stage of being certain that there is a variety of camera modules out there that such a project could support (but all one chippers).Unfortuately I am too stuffed to do most of it, but I have come up with what information I could for others to work with.

I would support Gigabit erthenet for 720p or 1080 4:2:0, but for 4:4:4 1080 it is still a couple of hundred mb/s short (pluss over head), and unfortunately it doesn't look like anybody is going to do lossless comrpession at the camera head either.

Valeriu, yes the size of the vault is the biggest worry I have aswell, but I am hopng somebody knows of a cheap backup medium used n computer servers that canbe used to store footage. We could also use an adaptive comrpession routine and compress nearlossless with bcakup footage that don't really need lossless. What do you think?

Valeriu

I had a look at those photonlines cameras, quiet an impressive range.

I was wondering, being a digital cinetographer, if you would like to give you opion on the performance of some of these chips sometimes? As they have a wealth of stats, some I don't even know, and the response curves, to the primary colours, on some of them looks a bit worrying (they are not very even/matching responses).

I notice they also have a PCI-X camerlink card too.

Thanks

Wayne.

<<<-- Originally posted by Wayne Morellini : Ohh I am such a complete cretin, I was over at the other thread and realised that if your using a normal one chipper than your raw image data rate falls to one third, because you actually have one monochrome pixel (behind a bayer filter etc) behind each scree pixel. So 2million pixels at 24fps, is actuall 48MB/s, going to 10 bits and we are still in the realm of possibility of fitting on USB2.0.-->>>

That hadn't occurred to me either. Assuming we're sending pre-Bayer images down the pipe, then we have only a 10-bit depth per pixel to deal with?

Actually, some chips appear to send 12 bits per pixel, which would (hopefully) mean you lose less information. I guess you'd then truncate to 10 bits after the Bayer filter, right?

1080p @ 10 bits @ 24 fps = 59 MB/sec
1080p @ 12 bits @ 24 fps = 71 MB/sec

Either one is too much data for USB 2.0 without some compression. Firewire 800 or Gigabit Ethernet could handle it easily, in fact Gigabit Ethernet could handle it up to 36 fps.

<<<-- Originally posted by Wayne Morellini : ... the cheap (compared to embedded boards) nano-itx pc standard by Via. -->>>

Thanks for the heads-up on the Nano-ITX board. I looked it up and it looks interesting ... but only has 10/100 Ethernet. It does appear to have USB 2.0, but something faster would have to be an add-on board.

It also appears to have a 1-GHz CPU. I wonder if that would be fast enough to shove the raw data down the pipe and provide a real-time preview via the RGB port? Would be interesting to find out.

Hey Steve,

I've got a couple of scripts I wrote. You say you are looking for scripts. Are you involved in a producing company?

Wayne,

You say 1080p 4:4:4 might be difficult and that we might have to settle for 4:2:0. I think the happy medium of 4:2:2 should be fine to start with, don't you? Perhaps that should be the goal.

By the way to all,

I'm not sure I should offer this, because I could be getting myself in over my head with certain social inticracies of my own personal life, but it's always a possibility.

As it turns out, my girlfriend's father is a recently retired ELECTIRCAL ENGINEER. Wayne just mentioned if he was an elecrical engineer or knew one, that some of these construction problems might be solved. After several weeks of looking at these threads on DVcommunity, I'm seriously considering talking to him about helping us construct a working model.

Before I get ahead of myself, I'm not exactly sure what type of projects he was used to working on at his work place, but he's supposed to be a real crackerjack. He only retired last year so I'm sure if he was capable of helping us, all he'd need to be brought up to speed technically is some communication with you guys.

Of course, I should mention that most people don't retire in their mid 50's so they can spend an inordinate amount of time with their daughters boyfriend creating digital film cameras, but my girlfriend is quite certain he would love working on such a project for both my benefit and also for the challenge of it. What I could bring to the table would be mostly gruntwork and necessary research, as I'm more of an artist than a technician. But hell, filmmaking is all about gruntwork now isn't it.

Perhaps there's already enough knowledge on the board so this is not necessary, but the words "If I was an electrical engineer" spurred me into realizing that it might be the missing link.

Maybe? Maybe? Input please.

As for logic/motherboard option, all current models AppleMacs have on board Firewire800+400, USB2, Gigabit ethernet, RAID option as part of the OS... and running a flavour of UNIX, I think they would be also Linux friendly.

No, Laurence, I'm not with a production company, I wrote that I have scipts but lack the equipment :(( Thanks for the offer, though :)