Home made camera designs?

[Wayne Morellini]

Rob

I agree, but the problem is that we need to know what camera specs before we can do too much with what we are going to use and which path to go. Perople also have not decided what path. The other problem is that there are so few people involved on these threads, and reality so few things to do, that further splitting them up might make them a bit sparse.

Each camera design has it's own thread, except for Silicon Imagaing that is roaming around, and has adopted this thread, so I think we are part way there, all we need is a Home made Camera Developement thread for the Backend camera system (meaning the capture computer and editing sections vs the front end camera and optics). People can work in teams among themselves and email back and forwards (like the Robs) and report into the thread about what is happening and to ask discuss issues with the rest of us. That could even be done on this thread.

So far we have been gathering suggestions and data (plus making arrangemanets).

Next we need to decide what path (format, interfaces, software and parts) we want to go with. I have made a number of suggestions in the three threads. Knowing the specs of the initial cameras and interfaces coming in a few months time,would really help.

The we can draw up plans, and devlopement tasks (that the people involved then discuss among themselves how to schedule). And then we start, and hopefully mostly finish by the time the cameras arrive.

Meanwhile we trial trying to get the existing Silicon Imaging system up and runnign as a test case.

But frankly untril all the good optios are researched and laid on the table to be agreed upon, there is little piont in starting. As there are two Robs who want to do the software, and little agreement from elsewhere, the software will probably be whatever they would like to decide for us.

I must say I have yet to read the rest of the posts in the thread yet.before making this comment.

Thanks

Wayne.

[Wayne Morellini]

Rob Scott

Problem with HD lcd screens is that you have to wait for cheap ones. I heard of a 4 inch 800*600 PDA LCD last year. But the best chance is to go to the Micro Displays that can be used for headmounted (one of the things I was going to do), veiwfinders, panels (like Kopkins does for teh old JVC Mini DV cams, and some stills camera), and can do projection (one unit could be adapted to all). This was another seperate area of research I was as looking into previously. You can get HD sizes easily in them, my favorite in microdispklay corp that has a no loss colour filtering system in development (uses grattings) with most circuits on the silicon display chip. Though they are cheap to manufacture the end products they are in really jack up the price.

Beware of OEL's, tehre biggest problem is that the primaries don't even fade over time, so affter 1 to two years you are going to have an underiable strong tint of colour, as they loose there white balance, and overall life is down. They also would use much more than the Liquid Crystal on Silicon (LCOS) chip microdisplays I just mentioned, also look at Sharp they have a couple of really low powered screen techs (and also the one I mentioned above, should perhaps be in the news archives of www.brighthand.com or www.pencomputing.com).


Rob, about the filtering Bayer to Tiff 16. Maybe it would be good to have an option to save in Bayer and do the filtering to desired format after? This would rediuce processing and HDD load.

Thanks

Wayne.

[Wayne Morellini]

What has happend to Steve I?

Somebody said that they were going to talk to their Girlfreinds father about the electroncis, what happened?


Steve N

I though cameralink was rated upto 3.2Gb's or something like that, but that the PCI interface is what holds it back, (hence my suggestions for AGP, PCI-E, or Via ref baord witrh cmaera link bypassing PCI bus)?

I thought this camera was 24MB/s wouldn't USB2.0 stream upto 50MB/s?

Obin, we look foward to your Pix's

Posted this to the wrong thread earlier:

Hi

Some itneresting news:

http://www.camcorderinfo.com/bbs/showthread.php?s=&postid=75642#post75642

quote:---
Concept HD
Guys,

Go to the HD Forum of Cinematography.com. You will find many arguments there by professional DP's against the use of industrial HD cameras in HD productions. I would urge everyone to wait for our camera announcements. Our cameras will not have these shortcomings.
----

Haven't had time to read the threads mentioned (haven't even read this thread yet) but well and truely worth looking at. I still say that any machine vision company should be able to pull through the goods, if they want.
Thanks

Wayne.

[Wayne Morellini]

About delaying the Mini-ITX MB and doing developement for a normal PC first, I agree. I would like to know what the fastest performance we can expect RAW unconverted on the normal PC before deciding on a minipc baord.

But we could probably findout the Spec of Mini-ITX boards for the next 4-6 months if the developers explained and asked VIA (they might even consider modifying a board to suit our needs as well (like mulitprocessor, gigabit ethernet (is 10Giga ethernet on main boards yet) that would also suite blade servers)? A big ask but you never know, they have many reference platforms for nitch markets, and are competing with the big I.

Steve, if all the Cameralink interface member manufacturers asked VIA they probably would consider making a variation of a server type multiprocessing (2 *1 Ghz to 4*2 Ghz) mini-itx board with cameralink, dual+ GigaE (or 10 Gb Ethernet) onboard for your market. You could then sell customers the whole PC, that could fit in a small (even a rackmounted) silent faneless box, which offers a lot of ergonomic workplace advantages.

Thanks

Wayne.

[Wayne Morellini]

I'm just wondering about Lossless Mpeg, does it use interframe compression, it doesn't seem to?

Interesting small computers:

www.xenarc.com/product/mp-fl8.html

Thanks

wayne.

[Laurence Maher]

Wayne,

I think that's excellent. Nothing will ever be started until we have an agreed upon goal. I think we should all be extremely blatent about what we are looking for from a finished system, and then find the point of agreeance for the entire thing from optics to editing, so we have a set-in-stone battle plan. Until then, we will have little more than cool ideas flying around the thread.

So I guess I'll be the first to say what I'm looking for (keep in mind I'm a technical layman, so all I can do is list basics, and some may sound more lavish than is capable):


A. THE CAMERA

In the camera, I'd like to see . . . .

1. 24fps up to 60fps in full 1920 X 1080p resolution.

2. 3 chips hopefully at 2/3 inches or bigger (at least biggest possible 1 chip), 10 or 12 bit, with native 16:9 aspect ratio.

3. minimum 4:2:2 separtion of colors, hopefully 4:4:4.

4. 66Mhz for 11 stops of latitude? (May not be realistic)

5. (I'm sure we'll need much faster).

6. 35mm still slr lens mount of some kind

7. Some kind of 2ond output tap for a mini-screen for shooting reference.

8. If possible, the camera should have some form of MATRIX ADJUSTMENT (probably in the software).



B. THE RECORDING INTERFACE

1. A camera to computer/hard drive INTERFACE with a data rate fast enough to deliver the above camera specs . . . without worry of glitches due to overload, movement, or general instability, (Ya, I know, vauge . . . I know of SDI, PCI, Camera Link, Firewire 400, Firewire 800, Gig Ethernet, Dual USB). Cable should be durrable and able to carry a signal up to a minumum of 100 feet without signal degradation. I would prefer a single cable so it doesn't get tangled, etc. Connections should be durable if possible. Whatever it is should be plug and go.

2. The COMPUTER AND HARD DRIVES need to be the most user-friendly they can possibly be. Plug and go. It would be wonderful if the hard drives used for recording were easily accessable and interchangable, (and of course inexpensive . . . lolol). This allows long shooting days. The codec used for recording onto the drives needs to create files directly readible by all common platforms . . . Mac, PC, Linux) Personally, I'd prefer the recording computer itself to actually be a Mac for stability purposes, but as long as the recording system is stable and delivers a 100% compatible video file, I'll go with it. In case it matters, I will most likely be editing my projects on FCP HD.

3. SOFTWARE CONTROLS FOR THE CAMERA that will adjust frame rate, matrix settings, exposure, gain, gamma (hopefully with several film-like gamma pre-sets), data rate, resolution, and white balance (hpefully with pre-sets for tungsten and sunlight). Other filters of some kind would be great, but I hardly expect this.

4. WAVEFORM MONITOR/VECTORSCOPE monitoring of the camera signal will be imperative for big screen productions, so we need to be able to run a software application that can do this (adobe premiere, FCP, etc.) on the capturing computer during filming, or we will need a separate output from the capturing computer or camera so we can feed it into a laptop for signal monitoring. (Perhaps use the same out as the HD monitor, but that may not be a good idea . . . not sure, where do you guys normally intercept the signal for monitoring?).

5. A COMPRESSION CODEC should as I said deliver a compatible captured file type. I really don't want to deal with conversion from one file type to another. It's inconvenient, unreliable, and might cause signal degredation of some kind. The data transfer rate should be scaleable, so one can choose to conserve drive space if needed. Hopefully the high end would allow 10 bit, 4:4:4 uncompressed (for near uncompressed/highly transparent compression) recording.


OUTPUT TO MONITOR:

1. Ability to output from recording system to some kind of 1080p HD resolution monitor (monitor being anything that is relatively mobile and gives a REAL REPRESENTATION of the recorded image . . . ya, i know, vague). If necessary, we might have to use this signal to also check waveform monitor and vectorscope.


POWER SUPPLY:


1. Power supply could hopefully be something like a typical car battery, and hopefully will power the entirety of the system.


NOTES:


The reason I suggest 1080p as opposed to 720p is because you can get 720p out of an 1080p image, but you can't get a 1080p image out of a 720p image. Therefore, the 1080p image capability will satisfy all crowds (accept those who want even more, which I have not seen mentioned on these threads). 1080p will allow definite ability to 35mm film blowup when considering possible conversions that might have to be made for effects or color timing or whatever. Whenever you colortime in post, you will lose an average of 20% resolution. As most indie filmmakers do colortime in post, it is better for them to start with the best possible signal. Also, HD projection is not standard practice in theaters. So filmmakers wanting to blow up to 35 will most likely make a conversion to film if they are successful, and there will be yet another massive loss in that process. By the time an HD project does get to a 35mm print, there will most likely be a great amount of lost information, and the best possible with a 1080p image may be a projection result. If we go with 1920 X 1080p then everyone here in the end will get what they hoped for, a great immage no matter what the distribution format.


CLOSING:


Ya, I know, this was a mouthful. I'm sure some of it may sound technically inept. Please feel free to mention what makes sense and what doesn't here. And realize this is my "dream setup". I'm aware that all of us, including me, will have to make compromises for the good of the group. That said . . .


LET's DO IT!!!!!!!!!!!!!!!!!!!!!!!!!

[Laurence Maher]

By the way, Wayne,

Yeah, it was me who was supposed to ask about my girlfriend's dad . . . kind of touchy with my girlfriend last couple of weeks, so didn't get the courage yet . . . things better now.

You know how that is, no matter how much you want to separate business from your social life, the two always seem to try and trip each other up. Anyway, I'll ask soon. Hopefully I'll talk to him by this weekend.

[Rob Scott]

<<<-- Originally posted by Wayne Morellini :
I would like to know what the fastest performance we can expect RAW unconverted on the normal PC before deciding on a minipc board -->>>

Keep in mind these are all calculations and have not been tested in the real world ...

I expect to get 30 fps writing out raw files (1280x720 10 bits) with a single desktop drive. I'm hoping to get 60fps with a 2-drive RAID 0 array.

1920x1080 10-bit 60 fps would require at least a 4-drive array. The same array should handle 30 fps at 12-bit. 60 fps as 12 bits requires 8 drives ... and so forth. I guess this is why the Kinetta camera has 12 drives in it.

<<<-- But we could probably findout the Spec of Mini-ITX boards -->>>

Good idea. (BTW, 10 GbE is barely available at all, much less on the motherboards.)

[Rob Scott]

<<<-- Originally posted by Laurence Maher :
Ya, I know, this was a mouthful. -->>>

Excellent ideas here, Laurence.

I think my "design" matches up in many respects to your list of requirements. There are many "dream" requirements that I've left out -- for now -- just because I'd like to see if we can put together a system that works with the minimum possible feature set first. See http://www.obscuracam.com/wiki/wiki/...RobScottDesign

Since the firmware should be field-upgradable, we can add dreamier stuff later. :-)

You'll notice I'm currently planning to use the SI-1300 camera, mostly because the bandwidth is managable with 2 drives and because it's relatively affordable. I would love to go for a 1920x1080 system, but I'm not sure my personal budget is going to support it. Having said that, we're should design the firmware so it could handle whatever imaging chip was put into it.

[Obin Olson]

good choice Rob, you will like the 1300 it has very good images and what is lacking you can ad in post because of it's 4:4:4 and 10bit...it is softer then 3 chip HD but that is to be expected...some may even like it MORE then 3 chip because it's like having a soft filter on the front of your lens ;) I don't mind because it's still 4:4:4 and 10bit it' sure beats the hell outa the IBIS5 chip that is also in the same ballpark pricewise as the 1300

[Wayne Morellini]

Been over to the cinematography HD forum, and checked out the threads on our machine vision camera projects (haven't finished them yet), and they seem to know more about some of the things on the Sumix camera than we do ;) with some useful comments, have a look here (there are more threads):

http://www.cinematography.com/forum2...machine+vision

http://www.cinematography.com/forum2...showtopic=1262

From past reading, an agressive mob so watch your step.

Thanks

Wayne.

[Wayne Morellini]

<<<-- Originally posted by Laurence Maher : By the way, Wayne,

Yeah, it was me who was supposed to ask about my girlfriend's dad . . . kind of touchy with my girlfriend last couple of weeks, so didn't get the courage yet . . . things better now.

You know how that is, no matter how much you want to separate business from your social life, the two always seem to try and trip each other up. Anyway, I'll ask soon. Hopefully I'll talk to him by this weekend. -->>>

Take all the time you want, I know how touchy these things are, don't rush into it, actually, lay bac,k and forget all about it for a while, love is worth a thousand cameras.

Actually I'm taking the rest of the week off from the boards, I may see you on the weekend. You have all my desires listed in these threads, and they are practical specs to deliver the best perforamance. I advise everybody to keep these things in mind when picking their specs. Like, I don't care if the 16:9 image comes from 4:3 chip (which I can use for standard TV or even IMAX, but no 8mpixel chip), if the 4:3 chip is cheaper better go for it, the difference should be very small for 16:9. We should not accept a single Bayer that is less than 1080, as it softens the image and loses RAW resolution and colour infromation (2/3), inmatter of fact a Mega Pixel bayer image that resolves accurately down to 1080 or 720p would be very good. Otherwise it is 3 chip all the way from 720p to 1080+ for me. Whatever we can do (not resolution or picture quality wise) to get a cheaper digital PC interface is good with me (now that multiple USB2 streaming is out), if anybody finds $100 cameralink interfaces on google etc let us know.

Actually we should be discussing this int he Viper thread, would somebody like to quote this there.

The rest of your comments I hope to look at when I come back.

Thanks

Wayne.

[Rob Scott]

<<<-- Originally posted by Wayne Morellini :
From past reading, an agressive mob so watch your step.
-->>>

Yes, they are, aren't they?

But then, again, they seem to be mostly professional DPs. With time schedules more important than money considerations, of course they don't want to fool with a do-it-yourself camera. Of course they are going to pooh-pooh the idea.

But I think there is large gap -- much too large -- between the $40k+ equipment meant for professionals and the $2k-$5k barely-adequate cameras like the DVX100 and GR-HD1.

That's where we fit in.

[Wayne Morellini]

Yes Scott, thats right, underneath Conceptual HD.

[Les Dit]

I agree that there is a huge step between the $40K HD cameras and the $4K HDV cams like the JVC I have.
So now that we are starting to speak of costs and feature wish lists, what are people willing to spend on a camera and supporting computer capture system?
Laurence, you forgot item 9 on your wish list, the price point!

-Les