New DIY HD Cinema Camera Project

[Take Vos]

Originally Posted by Igor Babic

http://www.digit-life.com/articles2/sas-and-sata/

I dont quite understand this but as I can see from the bottom of this article, you just hook up SAS disk to your existing SATA controller and your inn.
Has anybody try that?

That only works in one direction. You can attach both SAS and SATA devices on a SAS bus. Bot not SAS disks on a SATA bus. See the wikipedia article.

[Igor Babic]

http://www.cs-electronics.com/sas-adapters.htm
Yes, I can see now.

[Take Vos]

MacMini, has a intel AHCI SATA II controller. I am not sure if it is possible to run two disks on this port using SATA2->eSATA cable connected to a eSATA disk array.

The MacMini also has a AirPort Extreme that is connected to a 1x PCI Express. Probably through a "PCI Express Mini Card" connector.

I couldn't find a Mini Card to PCI Express extender, but it may allow connection with a capture card or firewire 800.

[Serge Victorovich]

Guys, look at Wearable Computer System for HD Capture

[Jose A. Garcia]

Looks very good. I left him a post in his thread to see if he can come and take a look at our project.

By the way, I've got a problem.

The third week of testing is almost over. I asked Micron for this extra week because I was hoping to have both the Micron and the Omnivision sensors to test them together, but the Omnivision board hasn't come yet.

If I don't return the Micron board this week, I have to keep it. That's about $1300 taxes included that I've already paid but of course I get my money back if I return it.

Should I keep it? Should I buy the Elphel 353? That's about a half of the price, the same sensor and it already has less data rate that's our main problem now.

[David Delaney]

Does this device help?
http://www.newegg.com/Product/Produc...82E16812232002

[Igor Babic]

Jose has USB camera with USB port on steroids. So this is probably no go. LS-371 has 2xSATA1 and IDE ports onboard and they probably have enough speed to do what we want. Problem are single 2.5" SATA disks that are portable but have not enough speed to take 2K uncompressed bayer video (and like all sata disks when they pass 50% capacity they are even slower, alot). 3.5" at 10000 RPM are probably OK but those are not for carrying around and hot swaping with out much extra care and alot bigger case. SAS disks has also 2.5" version and hot swap option but this needs extra controler, miniPCI to PCI conversion board (and this is over 350$ extra excluding SAS disk, and you dont know that this is working becuse SAS controlers are mostly PCI-X type ) SAS disks are also much more expensive.
Jose:
Are you shure that you can use 10000 RPM disks?
Why you dont use 1920 size frames if this is OK? (every time someone start project like this he wants max size that is available, and everytime tehnology is almost there but noth enough. Wayne is asking for 720p recorder for couple of years now and we are talking about 2K but no one has succes in DIY 1920 yet, exept you.) I think if you wanna do it now do it with what you got that is close enough and relyable enough. You can always upgrade later. There is alot other stuff unknown in front of you and this will be great expirience. Everybody doing their products like this. You are now already better then HV20, and this is quite good enough for start.
Elphel is Linux. This is for me major "dislike" and you have to wait until is finished. I think also that if you wanna try Omnivision record same things with Micron that you will record with Omnivision. You will lose only your shipping money if you return Micron.

[David Delaney]

Jose, why don't you return the board and get someone else here to get it back the same way you did? If I understand, the demo boards are to be used for a few weeks, right? Ask someone else on the project to get the same demo board? Just an idea...

[Seth Kersey]

Jose,

So far, I think you have captured incredible images with your Micron.

I think the datarate/HDD problem is solvable. As Take has suggested, software RAID may be enough, and would not require additional hardware... or, as Igor suggests, simply using the 10000 rpm disk you can currently record with and upgrading in the future. I think that is how PJ did it with RED, they had only the bodies and used off the shelf drives (though lost a full day of aerial footage because of helicopter vibration).

As for whether you should keep this dev board, was there any further word on the $89 dev board Steven mentioned a while back? $1300 is a lot of money if there is an actual $500 solution out there... but I do not know the specifics about the board Steven was referring to.

[Steven Mingam]

Why use software RAID that will hog your CPU when you could be using this CPU to losslessly compress the stream 2:1 and use one single 2"5 disk ? Don't compensate lack of software with insanely expensive and complex hardware solution. (Even more when the software part is quite simple. Everything is out there, you just need to put the bricks together)

For the board i was talking about (atmel ngw100), it's the same, there is everything to interface a cmos sensor (and usb 2, 2 ethernet 100, sd-card slot) but you will need to program everything (it's running linux). Well at least a lot of things (but there is already guy who did it, interfacing a micron MT9P031 with IDE on the board. If you buy the more expensive devkit (500$) you even get a QVGA LCD :))

[Agustin Vrljicak]

Well, right now you have to chose either to bet on your project, or to go for the safe choice. If you are looking for a practical solution, I think Elphel would be best. Linux? Does that REALLY bothers?
Or you can bet on your project, and risk a lot of money, and probably get where you want to get (or not).
But (real) 1920*1080 is not really too far from 2K (128 pixels horizontal). You will probably get a very similar image.
Whatever you do I wish you best.

[Igor Babic]

Steven, can you post link to ngw100 camera project please.
I have here a guy who has done some programing for Atmel camera that has to be used for film scaning. Maybe he can help us. This board looks intresting...
Sorry Jose this is again miles away from your project, but this same guy can help us with your stuff too. I will try to cantact him ASAP.

[Jose A. Garcia]

Ok, I'm going to return the board. I can always get it back whenever I want. I'd like to know more about that dev board Steven's talking about.

I want to test the Omnivision sensor and would really want to know of other possible solutions out there. There has to be another cmos manufacturer offering demo boards. You guys are great finding stuff, we could spend a week or so trying to find a different cmos+board (not already built cameras). I'll pay to get one for testing.

Also I think the perfect solution would be to low down data rate before recording. That way we won't have to push hardware limits. I'm sure I want 2.39:1 2k and I know it's possible. 1920x1080 is almost 2k. Steven is right. We've got a lack of good software. We need to find someone who can code under windows and wants to help us. Or even better, someone who can program a FPGA to lossless compress the stream before it gets to the computer. We get that, we have our 2k camera without ANY problem.

[Jose A. Garcia]

Take a look at this... Lots of info about hardware encoders and other stuff.

http://videobits.org/general_vendors.html

[Seth Kersey]

Here is a link to an open source FPGA resource, they have many "cores" available including video compression cores (second link).

http://www.opencores.org/

http://www.opencores.org/projects.cg...stems/overview