Can you convert it for the film production?

http://vfm.dalsa.com/catalog/Pantera_SA-2M30_Product_Sheet_00128-09.pdf

The Dalsa seems to be a nice camera. The problem is the price. I think it's almost 10k.

That is ridiculous. JVC will have three 2/3" 1080p CMOS camera for $20K, with tape, veiwfinder, audio recording, everything. Dalsa should cost no more than 1/6 JVC camera cost, or about 3,300 USD.

That might be ridiculous but it is true. After everything is said and done, that one dalsa camera is out of reach of most people. I contacted them a long time ago and a few others did and it was close to 10k USD. Unfortunate because it looks really sweat.

Keep in mind that that jvc cameras records to hdv which is just mpeg2 at 19.2 mbits compressed on the fly. I don't want to start a debate, but I don't think mpeg2 transport streams are not what most people in this section of the forum are looking for.

There is also new JVC 720p HDV camera, uncompressed output for 5,500 USD with lens. Three CCD's. If that one could be made record to hard drive or small PC, it would be winner. It has viewfinder, etc. Uncompressed output is 60p. It records 30p and recorded image is not so good, but uncompressed is superb, per NAB reports. It would be great filmmaking camera.

Yeah but it's a 1/3" .

But it close to single 2/3" in area hit by light, it has deeper DOF, when needed you use 35 mm adapter for shallower DOF.

One 2/3" may not better than three 1/3" CCD's.

Radek

You guys trying to develop system from 4,000 USD camera, without lens, viewfinder, sound, easy picture settings, etc., when you could develop similar system for HD100. I would much better buy HD100, then use your system when I would go film out, than use offbrand industrial camera. If you develop something for JVC HD100 or Sony FX/Z1, I'm buying, and probably many more would buy too. Your work is pioneering and may be reason for which Sony, JVC Panasonic adding uncompressed output. Thank you. But now is time to stop and reevaluate what market wants. Does it wants industrial camera that hard to use and good for film out only, or it wants more universal, versatile, neater solution. There will 1/2" and 2/3" 3-chip cameras in future from major manufacturers that you could adopt to same system. System Juan built for Panasonic may be way to go and in case of new HDV cameras you need not hack anything. 3 CCD's require more data be recorded, but maybe some light compression could be applied.

Maybe need pole to see what would members here want better, 4,000 USD industrial camera or new JVC, etc. to use with uncompressed recording.

I vote for FX1E/Z1 in 1080-50i mode, to deinterlace later.

I vote for FX1E/Z1 in 1080-50i mode, to deinterlace later.

Radek,
A 720p will look much nicer than 1080i rescaled & deinterlaced.

I already have FX1E and I read somewhere that like 35,000 Z1's were already sold. HD100 may be better, but if someone would make such system for FX/Z1, he could make fortune. When Juan started his system, there was DVX, nothing else. Now we have budget HD.

Radek

How are you going to capture analog component HD. You Can't directly! No capture device exsists that takes in component into something and saves it digitally. W-VHS is the best you got and the only thing that can do it natively, but that is just an analog vhs recorder with smaller heads.

The only way to capture analog component HD is to convert it to HD-SDI which isn't cheap.

It would be about 3k to get a component to hd-sdi converter, and then you need a computer that could record it or a digital deck and your back to some compression, but if you can afford the deck then you could afford something better than a hdv.

Their is also another technical limitation for the reel-stream mod. Bandwidth. Even if you hack the sensor on these prosumer hd cameras you don't have enough bandwidth over 1 cable. So 3 usb then, but that doesn't work because the controller can't handle the bandwidth. So 3 controller and 3 usb, but this doesn't work because usb is a pci bus (this might not be true, because a lot of usb are embedded in the south bridge but nobody knows the real bandwidth)and the whole pci can't handle that much bandwidth. The next step will be insane, 3 computers, but now you have the problem of trying to sync them. But even if you got away with one computer because of multiple embedded usb controllers, you need a computer with either hugh horsepower to compress on the fly, or pci-x to handle the storage. Its harder to overcome.

I'd suggest you check out my thread on DIY development plateform, which has turned into me developing a hd-sdi camera. I also have plans for an fpga hard drive deck (NO filesystem) that will capture hd-sdi and component and will have hd-sdi out and component out. But the camera head is being developed first. This is not a pipedream, i alread have half the fpga for the camera programmed.

Keith,

Thanks for sharing. I don't really understand all stuff you're talking about. But good luck. If someone comes out with better and less expensive solution filmmaking, count me in as customer. I just think that big companies are well aware about what is going here. They could and will overrun you any time you get into their market. If nothing, least you guys will responsible for pushing big giants to give reasonably priced filmmaking tools. I think that it's beginning to happen already.

I just don't understand the politics. You and Juan can make reasonably priced filmmaking camera in garage while giants charge 100,000 USD for compressed 1080p.

Radek

I don't know how many people know what i talk about half the time. Right now i had to figure out how an asyncronous FIFO works because of all the different clock speeds in my camera head. I doubt many people know what a fifo is, oh well.

You are so right on the big companies. At any minute they could go and build something like what i'm designing and have it designed, built, and tooled up for production in a couple of months. But broadcast gear has a set price, and within the industry it is a taboo to break that price point to quickly. Give the masses HD to fast and they destroyed 10 years of profit.

If someone came out with a hard drive recorder like the one i have in the pipelines and sold it for less than 10k I think it would be chaos in the industry, and some really large companies would be really angry. I like that though because that is what i want to do. I'm only building the camera head because I need something cost effective for the deck idea now (and learning about the data flow and how to overcome similar problems in the design such as different clock speeds and stuff), but IMO the deck will be the big thing for this design because even if people don't want the head, they can get an ikegami HDL-40 or a sony HDX300 and for 20-25k would have a camera that you destroy all hdcam and dvcpro hd cameras at the same price.

RIGHT!! :).
I had the same idea some time ago.I suggested it to a camera manufacturer but they didn't pay attention to it.
I have the same problems trying to understand asynchronous communications as you.
I just know there is some kind of packet system, handshakings and the like, anyway really weird stuuf for my head right now.
Could you point me to some info about it?

commercial policy:
Big companys will never sell products for low prices if they can sell it with high profit.

FIFO:
First in, first out. Its a waiting queue, or a buffer. Its like a elastic band conveyor. Non continuous datas gos in and continuous datas go out. Each interface, like network, HDD, USB or whatever need FIFOs. Without it, you will lost datas.

I knew what it meant, I was just trying to emphasize that once your down to the electronic level most people get lost when i try and explain stuff to them. I had a feeling you would know what it meant anyway. But i'm really glad some people know what i'm talking about, horray!

Juan:
Since i'm using an fpga I'm playing ignorant and either going to use an external asynchronous FIFO and write programs around sending clock when needed and when data in/out is ready. Or just download the app note code from xilinx and modify it to fit the depth and width i need. In fpga's your code gets translatted to gray code or something, which to me is completely beyond me or anyone i know. Texas insturments make FIFO's and I've been going over their data sheets to help me understand and I think I almost have solved my current problems in the camera head design. Hopefully. But i want to see if i can change the ccd timings so that i can avoid having to use ram as a buffer, block ram isn't big enough and i don't want to wire external ram.

Yes, I know what you mean.
I know that situation with VHDL.What happens when you are placing the logical structures by drawing the gates through any of the available software packages out there? Is it still grey code?

Honestly I don't completely understand what goes on inside an asyncronous fifo and I have even less of a clue about what this gray code is.

What i do understand is that it should end up like a syncronous fifo, so that if your reading to fast your going to get into a situation where your fifo is empty and that will have to be handled, or you will get data gaps or something. I've haven't had a chance to test it though. It only gets translated to gray code, then it is output into binary. I think the gray code has to do with transistioning from one clock speed to the other.

I don't know if that is what your asking about. I'm just a beginner with fpga stuff.

This thread is completely off topic, but now back onto topic of the dalsa camera. I just clicked on the specs to the pantera SA-2M30 and something caught my eye. The 7.4 um x 7.4 um pixels. Then i looked at some of the other stuff and the pixel response stuff.

I know dalsa makes ccds, but what is really odd is that the handful of specs listed match the kodak KAI-2093 that i'm designing around. Even the absolute QE is pretty much a match. It could be just nothing, but the specs are so similar. Not much to go by except pixel sizes, avaiable resolutions (kodak has 2 2mp sensors, both the same res), type of sensor and QE and optical format size. Because of this I think that it might just be a cameralink camera based on the KAI-2093, but i could be wrong. Just a thought

Radek,
A 720p will look much nicer than 1080i rescaled & deinterlaced.

All a matter of opinion.

take note that if you were discarding one field the final image would be (HDV) 1440x540 which gives a little less pixels than 1280x720.
I guess best situation would be using some expensive motion compensated deinterlacer (e.g. Faroudja).In this case we could say that final image would be a little bit better than 1280x720.
Anyway Sony's camera is 3 sensors of 960x1080i, so things are difficult to determine.

But it close to single 2/3" in area hit by light, it has deeper DOF, when needed you use 35 mm adapter for shallower DOF.

One 2/3" may not better than three 1/3" CCD's.

Radek

Yes, thats right, but it is a bit more complex than that. On signal colour chip the light each pixel can absorb (because of the color filter discarding the wrong colours) is nearly the same. The colour that doesn't match are discarded. But in the end the light for the 1/4 matching pixels between single and three chips, is not too far apart (maybe clipping in the shadows because of the filter). They just calculate the missing pixels in Bayer. The problem that should crop up is thermal noise, as that wasted light has to be dissipated as something, most likely heat, this will probably reduce your range and introduce noise into the bottom end of the light range.

I don't know how many people know what i talk about half the time. Right now i had to figure out how an asyncronous FIFO works because of all the different clock speeds in my camera head. I doubt many people know what a fifo is, oh well.

Keep going Keith we need more intelligent people around to balance things out.

As you guys summarise. Companies stratify the features of their products to manipulate people into paying a lot for the best features. If they did everything near right, then their profits might drop to near a tenth (still a lot for us). Radek, there are a lot of people running around here with no awareness of the commercial realities of what they are doing, or enough of the technical side (believe it or not). It requires a very big investment in time and research to gather the required knowledge and information first. Mike is getting taught this at university, so we are lucky to have him. But even he will be learning a lot of specialised knowledge outside that the university course will cover. Sure things can be done cheap and quick but by the knowledgeable. "Fools rush in where angles fear to tread" is the saying and I am not biting off more than I can chew, but waiting. Seen it happen before, where somebody lands up broke, disgruntled, happy with nobody but their inner circle, and nobody really happy with them. Just be happy with what you know people can do. So be patient and something should turn up one day. Actually the Sumix camera is probably going to be he closet to what you want, soon, very very soon, hopefully cheap, I have been in contact with them.