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

[Rob Scott]

Steve Ipp wrote:
Rob, I forgot whether it was you whom I sent hardware configuration info for a PC based HD NLE

Yup, I got it. Thanks!

[Steve Ipp]

Just curious, did you get this 20000 US $ soft?

[Rob Scott]

Steve Ipp : Just curious, did you get this 20000 US $ soft?

No, I don't have $20K to drop on anything. If I did, I'd probably buy a Kinetta camera and not bother with building my own! :-)

[Steve Ipp]

Wayne, did you get the Cyborg?

[Laurence Maher]

Wow guys,

Things seem to really becoming reality. So exciting. Currently, here's where I am. I went over and checked out Mac OSX with FCP HD, and it's beautiful. This is the kind of stability with incredible and user friendly features I was hoping to get in a NLE. My production quality and times will be expotentially increased. Plus, if things go wrong Applecare will take care of me for 3 years, which I'm gonna need in times of trouble (not tech savvy). So as much as I don't want to spend around 10k for the whole editing setup with 2 HD monitors and a terabyte of space, I think that for my knowledge and needs, the price is worth it. So what I'm looking at now is how to intigrate these current cameras with such a system. Hopefully they will also have Mac Camera Link Cards? If not, I assume I could throw together a cheap pc for capture and then transfer the files. Any way you look at it, I'm getting one of these cameras to come. They sound mouth-wateringly awesome. Input anyone?

Thanks.

[Wayne Morellini]

<<<-- Originally posted by Steve Ipp : Wayne, did you get the Cyborg? -->>>

Sorry, I didn't I have too many other things at the moment, and I can't afford it.

I am interesed in maybe making casings for these units commercially. I had designs in mind when I was looking into doing my own camera last year. I am currently thinking of an air tight, water tight, demountable (lego like unit) with some nifty styling, and ergonimic like features, in a shoulder mount version. A handheld version with one drive (so only single chip 720 at first) might be a possibility.

So does anybody know of a firm that could fund, make and sell these sorts of cases cheap and offer some design help. If I do it right then any commercial camcorder could also use it (firewire controls).

There is at least one plastic extrusion company locally here as well.


<<<-- Originally posted by Steve Ipp : Thank you for advice, Wayne.
Just add "camera" in front of the "@" :) -->>>

Still the name of the person, email me if it is conf?

<<<-- Originally posted by Steve Nordhauser : OK, this has come up a bunch of times and I've hinted at it. We are getting an Altasens design running. First proto is taking images. Please don't ask too much more but I will release information here when it is appropriate. Figure 6-8 weeks to first shipments. -->>>

As long as somebody offers a good quality 3 chip in the near future?

From another post I constructed:

Hi

I've just read the Bayer pattern doc at Stanford, and just realised I have been laboring under a missconception about Bayer. I thought the bayer was a overlaping complimentary scheeme (that would allow upto double the light sensitivity, as two primaries are sampled for every position) and more correct colour and resolution than a primary scheeme, a bearable compromise at 1080). Does such a scheeme exist, and is it's performance really upto scratch?

The algorithms on the additive primary bayer pattern are mainly res upscaling and smoothing, with a real Raw resolution of 640*360 (double in the green, even though it is cheaper, is this much better than upscalling 3chip RAW DV?) estimation from other primiaries is never going to be completely accurate, as a estimation of primary colour instensity change is going to be hard to predict even if you try to use the intermediatory of another primary, but of course the colours are brillant. The problem then comes into effectively upscaling an bayer upscaled image. So I think we need 4 sensor pixels for evey true pixel (but extra pixels reduce performance), well just my opinion and one I think would affect my purchase.

Thanks

Wayne.

[Steve Nordhauser]

Wayne on Bayer:
Intrinsic t the Bayer concept is that the image is 50% green pixels. Since green is in the middle and the filters aren't perfect, you get some response from part of the blue and green areas. Also, the eye is most sensitive to green. This means that for a scene with the normal array of colors, you get the majority of spatial information (edges, high frequency) from the majory of pixels. Of course low end reds and high end blues will give only 25% of the resolution, but a good algorithim might give 75% overall.

[Laurence Maher]

Say guys, I copied this from another thread:

Not sure if this will help or not but I found a codec for free that supports 16 bit per channel as well as an alpha channel making a 64 bit video codec. It works on mac and pc with just quicktime 5. Best of all it is free. They even have a lossless codec that can get 6:1 compression with no loss but that codec is $99.00. I know it isn't 12 bit per channel but it might be an easier way for people to manage files opposed to a series of stills. Besides right now the tiff files will need to be 16 bit anyways.


http://www.digitalanarchy.com/micro/micro_none16.html

[Rob Scott]

Wayne Morellini wrote:
I am currently thinking of an air tight, water tight, demountable (lego like unit) with some nifty styling, and ergonimic like features, in a shoulder mount version. A handheld version with one drive (so only single chip 720 at first) might be a possibility.

Wayne, keep me in the loop if you get anywhere with this. I've been thinking about trying to build a "shoulderable" camera out of this project. (Sort of a "Kinetta lite" I guess :-)

[Wayne Morellini]

Sumix and Steve,
and Robs you need to have a look at some of this.

Took me a while but I finally found that high speed serial interface.

HDMI - High Definition Multimedia Interface. Small like USB, is the next generation of DVI (using:

"High-bandwidth Digital Content Protection (HDCP)", "VESA’s Extended Display Identification Data (EDID) standard, Display Data Channel (DDC) standard (used to read the EDID), and Monitor Timing Specification (DMT). In addition, the EIA-861 standard specifies mandatory and optionally supported resolutions and timings, and how to include data such as aspect ratio and format information."
www.sigmadesigns.com/support/DVI_HDMI.htm )

It also will also send hi-end surround sound at the same time.

Consumer grade (so hopefully will be cheap). Supported by a lot of big companies, so it should eventually come to multimedia PC's (hopefully replacing the VGA port). The speed is 5GB/s (for one socket). So not as fast as Camnerlinks 2.3GB/s (on SI website), but looking below it should serve much of our future needs.

And it only gets better. One of the links below says that behind the scene the manufacturers are agreeing on mass to use HDMI as the default video connection ;).

You can get DVI to HDMI convertor cables (anybody what AGP graphics cards with DVI input will do HD resolutions, bandwith limitations, and cheap capture cards limitations, just dissapeared). I could not find mention of HDMI on VIA, but they have just announced a graphics/capture addin cards, I would think with DVI, I expect this to eventually carry over to the ITXEden form factor ;).

You can also get HD-SDI to DVI adaptor ($999) using an existing (DVI based) graphics cards?

www.gefen.com/kvm/products/hd.jsp

And this isn't that little single droplet SDI channel that even Firewire beats in speed tests.

And it canbe made to travel 150feet (the 1640feet DVI extender is a bit expensive)..

5Gb/s (8mp*24bits*24=4.608Gb/s, or 8mp*48bits*24 / lossless compressed 2:1=4.608Gb/s, or 16bit*8mp bayer * 24=3.072Gb/s, or 16-bit*8mp bayer *50/ lossless 2:1) Forget the comrpession for the moment, but it will come in handy for in a few years time.

Planning ahead, the shorterm possibilities are: 3.2Gb/s Firewire (also there was supposed to be 1.6Gb/s wireless firewire), 10 Gigabit Ethernet, I think I remember something about a USB like version of the PCI-express bus, but am unsure. If 8mbit/s cameras are made with one of these and an older interface we could use it for 1080 shooting until Main Boards came out with the new interface.

Just looked it up, rumour is USB3.0 will have 200-500MB/s 2005/2006 (and somehting about ultra wide band and being wireless??, but beware the source of this rumour is only the inquirer article he got from some "guy" on a trade show bus or something):

www.the-inquirer.com/default.aspx?article=5920

Could I sugegst these following configurations:

Camera with 10Gb or Gigabit Ethernet (running at Gigabit for the moment)
+ HDMI (allowing us a host of multimedia connectability and recording on the host system).

Camera with External PCI express/USB3 (if such a beast exists)
+ HDMI, or Gb or 10 Gb Ethernet.

The use of 10 Gb Ethernet, allows great ussuability as it becomes available.

We coulde ven go one further and use S-ATA 300 ;), or two of them (then where do we pout the drive ;)


Could I suggest another trick that would be great for your customers in shooting, industrail and security applications. Including upto 40:1 Wavelet compression in the camera head (see the clearspeed device I posted, and also I think Analogue devices has something). This would allow 40+ 8mp cameras for security or shooting, to be hooked to one 10Gb Ethernet trnsmission path, (or HDMI or PCI Express), or even 4 + on Gb Ethernet, which is a major advantge for your customers.

In future (if all goes well, and often it doesn't come through) we may be dealing with a lot of HDMI, what do you think?

www.hdmi.org
http://whatis.techtarget.com/definit...847398,00.html
http://www.whatvideotv.com/articles/...l/200209_2.php
www.technewsworld.com/story/32576.html

http://www.pioneerelectronics.com/pn...0798291,00.htm
www.siimage.com/press/01_09_03_3.asp

Very insteresting VIA small formfactor platform for Flat panel:

http://www.via.com.tw/en/Digital%20L...FusionMark.jsp

Robs:

Been over to the www.PCIsig.org (the pci standards org), hard to get information there (havvew to register and pay big time just to find out how many MB's per second different PCI standards do) but if you do a search I came accross people in forums trying to maximise framegrabbing to RAID baords.

There will be a mini PCI express, and I'm confused wether 66mhz bus is part of the mini-pci spec (that would give 266mB/s).

PCI express external seems to be name Newcard (that seems to be PCMCIA express card which combines pcie and usb, confusing isn't it), TI has also tried an external connection over Gigabit Ethernet. Speculatively, maybe this is the tech that is being used in the rumoured USB3.0 above.

http://www.extremetech.com/print_art...a=30763,00.asp

http://www.extremetech.com/article2/...,895785,00.asp
http://www.extremetech.com/article2/...1269435,00.asp
http://www.extremetech.com/article2/...1271868,00.asp
http://www.extremetech.com/article2/...1267936,00.asp

http://www.ad.tomshardware.com/cgi-b...%3Frf%3Dtom025


http://www.extremetech.com/article2/...1585024,00.asp
http://www.extremetech.com/print_art...a=30763,00.asp

Tape BAckup:

Just been checking Tape Backup, found stuff for upto 1.2 tbyte backup and fast rates (some not yet available), I haven't been through the pricing but I think it is too high. So this will have to be researched in the future

http://www6.tomshardware.com/storage...408/index.html

Have fun, another 14 hours down the drain.

[Rob Scott]

Including up to 40:1 Wavelet compression in the camera head...

Very interesting idea, Wayne. I have been tossing around the idea of doing a two-stage process -

• Capture the raw data quickly to disk (no processing except perhaps some real-time RLE compression)
• During "pauses" in recording, the camera would process and compress the data in the background.

Basically, after a day of shooting, you leave the camera sitting there (powered on) for a while and it automatically processes all of your footage for you.

I also ran across this open-source project: http://www.bbc.co.uk/rd/projects/dirac/. It's a BBC project to develop a wavelet-based video codec. It supports 4:4:4 (and other encodings). Not sure about 10+ bit depth.

Thanks for all your other research as well! It's a bit like drinking from a firehose, but much appreciated! :-)

[Eliot Mack]

I spent a bit of time thinking about mounting systems, but came to a couple of conclusions--

1. The lens mount will depend heavily upon the actual chip and lens system used. If SI is planning a Altasens based design, then it is probably a slam dunk for the frontend. Lens mounts, etc. are very precise components. After it comes out, the requirements will probably be obvious. I'm very curious if SI has a way to prevent dirt from clogging the CMOS chip. Perhaps a glass plate to seal the CMOS surface from dirt would help; don't know if it would disrupt the optical path or not due to its distance from the focal plane.

2. The system design will also depend heavily upon the processing architecture used. I posted some potential Mini ITX power supply and motherboard designs on the Obscuracam Wiki (great idea BTW Rob.) If packaging this and batteries results in something that can reasonably be put on a shoulder, great.

I think the Kinetta design of making the recording module detachable from the camera head is very smart, and straightforward to do given the use of CameraLink or Gigabit Ethernet to connect the camera to the recording system. The computer chassis that most interested me (the Opus Systems Mini ITX automotive mount chassis) is not available for a couple of months, which will coincide nicely with the Altasens system availability.

3. When the above two components solidify to a reasonable degree, the detailed mechanical design of the housing can start. I think we will end up with a system with roughly ENG camera proportions. I've been thinking about thin walled aluminum or magnesium castings, and will probably delve into this when things get a little farther along. Should be fun. My background is in production mechanical engineering (was lead mechanical engineer on the Roomba robotic floorvac) so it's a good fit.

Exciting stuff. Can't wait to see where it leads.

Eliot

[Wayne Morellini]

<<<-- Originally posted by Steve Nordhauser : Wayne on Bayer:
Intrinsic t the Bayer concept is that the image is 50% green pixels. Since green is in the middle and the filters aren't perfect, you get some response from part of the blue and green areas. Also, the eye is most sensitive to green. This means that for a scene with the normal array of colors, you get the majority of spatial information (edges, high frequency) from the majory of pixels. Of course low end reds and high end blues will give only 25% of the resolution, but a good algorithim might give 75% overall. -->>>

I understand all this, but it is merely a trick with res upscaling. I agree that it would be good at higher resolutions where the 4 pixel bayer pattern falls in a 150dpi area (instead of 75dpi area), so I hope to go over the specs of your 8mp sensor when it comes out. The other problem with bayer is that you are sampling one primary and discarding the other two thirds of the potential light, where as complementary scheems sample two primaries per pixel primaries making the third primary more accurate, are there any good scheems for this?

Laurence, 100% thanks, we are hoping for the ability to support any codec, but until then we could do with what we can get ;) Looked at their test comprssions, pretty impressive numbers, but the scenes are very siimple, I can archieve around 8million:1 compression on a black Imax frame myself, even much more, as I could say a single bit value could flag wherever the frame is black or something else requiring further compression. Used to get in debates at uni with my engineer freind, he didn't get it because he had been told some law on data preservation, but I was approaching it from the veiw of moving data out of the file into the codec instead. Ohh yeah, I would like to see the results from some natural scenes, I don't have anything against the idea of 6:1 lossless compression. . I personally don't believe in a limit to loosless compressiuon of 2 or 2.5:1, I would use interframe compression myself (the whole file even. I had been thinking of doing my own compression routines for some time to use in my OS, and think I can achieve 10:1 over what is regularly been done (actually this has been done by a couple of groups years ago, Israeli Intrelliegence and a Video guy down south who wanted to replace the tape decks in a video wall with a live feed over the phone from his computer, same as the Israelies who popped up and announced they had been using 600:1 compression for years a few months after this guy came out with I estimate 1000:1) they probably are using the same techniques that I am thinking of. I think it is the approach more than anything. One thing that I would do to get high lossless, would be to eliminate niose/grain becuase it doesn't add anything, detracts from the real image, and is hard to compress. I sit on an awfull lot of patentable ideas around here, thgat I have to just ignore.

Rob, I preffer everybody make their own solution for the time being, these things can take a while, but I would like sub $500 solutions (even $200 solution) but need to find a company that can do it that cheap. I forgot the extra details, now I forgot what I forgot before ;) (been up all night). Because of the flexibility of the camera the front camera and lense (don't know how to waterproof a manual lense yet, maybe have to do some follow controlls, maybe there will be patent problems with that) is detachable and folds away for treansport and desktop use. I put a lot of projects on the back burner around here because I can't afford the developement and patent stuff, one was my own PC form factor and selling it to via or something (I have about three I want to do). Another interesting thing that could bge done is to make the case its (no, I'm not going to mention that incase somebody else patents it, that is how it goes in this world).

Thanks

Weyne.

[Wayne Morellini]

It is even a bit more simple then that Eliot, a number of cases (and even formfactors) can be used in my design. Maybe we should call it the camera form factor, or VTX. Now you remind me I have probably 4 designs I'm thinking of, one a long narrow case (like the Olympus SHD camera), another a really out there design with some very nifty features. But the water housing design can be designed to slip a standard small forma factor case in there (it would be designed for anti shock cushioning, so the case would just be a shell that everything else mounts into). The final one virtually is a case/or custom made small formfactor case with camer bolter on battery in the 51/4 drive bay etc (where to put for Raid drives thogh is a problem), my origional case idea. One place the display and electronic controlls are mounted is on the side.

Do you have any reccomendations on video equipment manufactures (or somebody cheaper) that could manufacture this type of case? I know how to get the best royality deal, but I am more interested in getting value out there (and take my 1%/10$).

Rob, your idea is good, but I was thinking of breaking the interface/burst bandwidth bottle kneck restriction on higher quality data. With HDMI or 10 gbit ethernet interfaces on the camera (we could have both) we don't even need to compress in the camera head until good 8mpixel sensors come ( So PC delayed PC compression is back in vogue again ;). So what about it manufacturers a good idea?


Thanks

Wayne.

[Wayne Morellini]

New Development:

Whoo hoo, I just checked my email and a manufacturer has emailed me out of the blue about something else, unfortunately they were trying to get me interested in a Movietube 35mm adaptor for $14500 Euro, so I don't I don't think I'll be using them ;) (and this is after I told them a month or so ago I would be interested in buying one of their units if it was cheap enough, instead of building a cheap one myself).

Thanks

Wayne.