4:4:4 10bit single CMOS HD project

[Rob Scott]

Wayne Morellini wrote:
The hotspot, can you get rid of it by reducing it just below max pixel value, by reducing the gain or putting it in reverse, or just using ND or iris, or is it a contrast thing with surrounding pixels?

I haven't done a lot to try to reduce the smearing, but your idea of reducing the gain is a good one. I'll try it.

[Rai Orz]

Lets thinking on differnt ways.

@Robs FPGA problem list:

1. it would need cameralink or gigabit ethernet in (Juan gets other kind of signals)

a.) Why not? But...
b.) Take the next step. Look inside camera heads. Most CMOS Sensors have digital, par. out, so why not connect the CHIP direct to a FPGA. (Okay, if you can do this, you need only a sensor and a few parts, not a camera head form manufactor xxxx)

2. it would need firewire 800 out (which Juan's FPGA solution seem to have)

If you can build a FPGA system, it will be easy to build also firewire ports or others. There are lots of chips for this work. But, look at point 5...

3. we need a viewfinder/monitor out (Juan already has this since it is on the host camera), since there is no host camera.

a) optical viewfinder?
b) second FPGA or PC based system for that job

4. we need controls to choose settings (Juan already has this since it is on the host camera), since there is no host camera

Write software on PC, test it out and translate the ini and control function to a MC.

5. we would need some form of RAID support either in the FPGA (either make it ourselve or have it builtin) or in the harddisks attached (ie, the Lacie RAID 0 drives??)

Take your brain free from all PC stuffs. The A/D Unit, inside (or outside the CMOS Sensor) have 10, 12 or more output pins (each pin is one BIT). The output rate is 33-66MHz. FPGAs can handle those data speeds in realtime. Shift or translation to 16Bit words are simple works. Next part is the Harddisk. Not a controller or interface, you can write direct to disk. A HDD needs words, not bytes. There are 16Pin (one pin = one byte) and you can connect it also direct to the FPGA or MC.
Also you can split the datas to more than one hdd. (First word to first HDD, second to second...)
But some sensor chip manufactors do the work for you. Most chips (2M pixel, or more) have two, or more output ports. Your FPGA can handle this and you can write to multiple HDD, without software logic. (First chip output go to first HDD, and so on)

Okay, FPGA is a new world. But since i thinking about a 3 chip, 1920x1080 camera, i think its the only way we can do it now. Also there are a lot of people with FGPA know how in special forums.

[Rob Scott]

Regarding FPGAs ... When we discussed this a few months ago, Steve Nordhauser indicated he'd be interested in embedding a FPGA solution inside of the "box" camera. In this case, the FPGA would be connected directly to the sensor and would output a compressed signal, period. Presumably the "box" camera would output Firewire, CameraLink or Gigabit Ethernet, which a separate system would them capture and stored onto hard drives. This "capture" system could be simpler than before, however; for example, a single hard would presumably be enough to store the compressed data stream. (The viewfinder would be a bit more challenging, since it would have to decompress the stream in real-time.)

[Steve Nordhauser]

Eric: Progressive scan is the sequential reading of the image from top to bottom. Interlaced is doing a frame of the odd, then even lines. All the sensors discussed in this forum are progressive. The only global shutter sensors discussed are the IBIS-5A for 720p.

Wayne:
The Altasens can read out at up to 60fps for a full frame, letting you drop every other frame at 24 or 30fps. It is still rolling shutter.

Marin:
We are shipping GigE cameras *now*. 100MB/sec of continuous video in 8 bit, 12 bit packed and 12 bit normailized formats. We haven't seen the need for direct connectivity to a hard drive since you need the camera control, the preview and other features.

[Régine Weinberg]

Hm
thats not crazy at all, but there is a lot of money involved I do guess. It is exact the way the big ones are thinking and if it would be easy to do this as DIY, Arri or Dalsa would have done this as they are not stupid at all and Dalsa is nothing else as a Philips child, so FGPA knowhow, the chips, emulators, tools all is with Philips. A nice idea but like with all the boyz in the Audio world nice discussions lot off fun, a year or so and then, nothing.
going Pc components is not so sexy at all, but do get the Data in and out at disk is not all and not as easy as it sounds with an FGPA

[Rob Scott]

Ronald Biese wrote:
thats not crazy at all, but there is a lot of money involved I do guess. It is exact the way the big ones are thinking and if it would be easy to do this as DIY...

Well, you know what they say ... if it was easy, everyone would be doing it :-)

What I had thought about is a FPGA camera head outputting to Gigabit Ethernet; this connects to an embedded PC capture body with a single swappable hard drive. The head and capture body could be dockable for a one-piece system. The FPGA would compress the signal to a standard open-source codec such as Dirac (wavelet-based near-lossless).

Of course, this is a pipe dream at the moment, and may never be worth it since the Kinetta camera is already doing similar things.

[David Newman]

<<<-- Originally posted by Wayne Morellini : Very good, as long as it is a good alternative to Mpeg2 50Mb/s stream (or was that 36Mb/s) I think this will give everybody some excitment.

From what you said the comrpession for Bayer is ^-4:1 and 3chip 4:4:4 6-10:1, is that right?-->>>

Yes.

<<<--How does your codec compare to the Avid codec used in the Ikegema camera, it is something like 145-220Mb/s. Is it true lossless or just megp2 like with about the same editable quality as your cinform codec (but at much bigger daya rates)?-->>>

The Avid DNxHD codec is a DCT based solution much like MJPEG. It will have higher bit-rate than the CineForm codec for the same quality. I have not had access to that codec for testing.

[Flax Johnson]

<<<-- Originally posted by Rob Lohman : Marin: most
3. we need a viewfinder/monitor out (Juan already has this since it is on the host camera), since there is no host camera
aboard DVInfo.net! -->>>

Rob,

Juan has a vga output on his mod.

The pana lcd gives a "so far from truth" picture since he bypasses the pana processing so he (we ?) need to know the picture look like.

[Régine Weinberg]

huh
What Juan does
and somebody wrote here, have a disk to get the frame out of the camera, drive me totally mad.
network storage device, it's bulky but I do guess there are smaller ones, Steves cam with Gig ethernet out or Camaralink to Gig ethernet, and than something like this

GIG Ethernet storage device http://www.iomega.com/na/products/pr...=1100189941751

with the rolling shuter for god's sake take the Vance Cam idea, a mechanical shutter like good old S16 and 35mm days.
for the viewfinder maybe another crazy thing pops up
The Steve Altasens wits an Argus taht's poor mans 35mm

voila found the pipe dream!
http://www.broadcom.com/collateral/pb/91125E-PB01-R.pdf

that's a evaluation board has 2 GIG Ethernet a chip with realtime OS and Hyper transport only to tell you that's a very fast I/O and that could take the data out to a disk controler chip I do guss

http://www.amd.com/us-en/Corporate/V...1^4752,00.html

imagine a very tiny motherboard there something runs to control the cam one PCI slot such a card in from this card to
a, or an disk array sound not so crazy and no FGPA could take the full bandwith.

[Rob Lohman]

<<<-- Originally posted by Rai Orz : Lets thinking on differnt ways.
1. Take the next step. Look inside camera heads. Most CMOS Sensors have digital, par. out, so why not connect the CHIP direct to a FPGA.

2. optical viewfinder?

3. Take your brain free from all PC stuffs. The A/D Unit, inside (or outside the CMOS Sensor) have 10, 12 or more output pins (each pin is one BIT). The output rate is 33-66MHz. FPGAs can handle those data speeds in realtime. Shift or translation to 16Bit words are simple works. Next part is the Harddisk. Not a controller or interface, you can write direct to disk. A HDD needs words, not bytes. There are 16Pin (one pin = one byte) and you can connect it also direct to the FPGA or MC.

4. Also you can split the datas to more than one hdd. (First word to first HDD, second to second...) -->>>



All good points. But you missed mine. I (and others) KNOW this
can be done, that is not the problem. The problem is DOING it.

1. I don't know how to do this, neither do I know of someone here who does (except perhaps Juan [who is busy with is mod])

2. I don't want an optical viewfinder. I want a digital one with B&W option, zebra stripes and histograms, but I guess an optical (although I'm wondering what kind of optical engineering that would take) solution would be okay to start with

3. Are you saying hook an SATA or IDE harddisk DIRECTLY to an FPGA design? I don't think this is going to work without some sort of controller since the harddisk needs all sort of commands being send before it will read/write

4. Yes, that is a software "RAID" solution we have already implemented. I was only saying this also needed to be done in FPGA which is harder to do

I think most of us know what the possabilities are and how such
systems could be designed. We just lack the specifics for much
of the realtime embedded hardware design.


Ronald: "with the rolling shuter for god's sake take the Vance Cam idea, a mechanical shutter like good old S16 and 35mm days"

Yes, that was one of my questions as well. I've asked it multiple
times but never got a clear answer from anyone whether this
will work, will remove the artifacting and how this would be
connected to a camera head (the chip and shutter would be
needed to sync I think). So unless someone has some realworld
info on that....


Flax: hmmm, interesting. Thanks!

[Régine Weinberg]

Hm
Vance has done this on his PAL cam
do google for Vance Cam

[Rob Lohman]

Yes I know. I've read all I can about the VanceCam. However,
what works for him might not work for us. Remember that his
head had an external shutter option. And it still doesn't say
whether it will actually REMOVE the effect (he wasn't using
CMOS technology I think). We need to hear about this from
the guys who make these CMOS chips and camera's. However,
Vance is also a member on this board, so I'll shoot him an e-mail.

Also on your "pipe dream solution". Do you know a price? From
what I've seen thusfar those systems are very expensive
(think $5-10K easily).

[Marin Tchergarov]

Rob L.
Thank you for the warm welcome!

Steve, Thank you for that information-it's important

Originally posted by Rob Lohman:
-----------------------------------------------------------
So unless someone has some good answers FPGA wise for these
kind of "issues" I don't see this happening for now.
-----------------------------------------------------------

This someone could be Juan! I hope! I've never seen an FPGA chip ,so I just assume it's a programable "many pins device"...

...and start dreaming:

1. You get from Juan a little metal(not plastic! :)) box with the following ports: 2xUTP - 1GigE-in(to connect to Steve's camera) ,1GigE-out for camera control (see Rem#), 1xSATA (or PATA) a HDD for Recording images, 1xVGA - Originally posted by Flax Johnson: "Rob,Juan has a vga output on his mod.", 1xPower connector -Battery

Rem#: (a part? of the FPGA should act as GigabitNetWork Switch)

2. You get from Steve a Camera with GigE out

3. You get 2 very very fast HDD's - WD Raptor (10000RPM, 16MB cache ...or similar) when the disk is full a "Camera Reloading" is occurred and the assistent swap the drives and the full one is attached to host computer (near the video village) and then copy files to host- then delete them :) from the Raptor.

Some clarifications:

The "1xVGA" is for a view finder(LCD) and this device (a wireless PC->TV convertor): http://www.grand.com.tw/en/product_f...-090&lwk_page=

To the "1GigE-out" is connected a small WireLess Access Point(BRIDGE) with a 100Mb UTP connector-
this is for camera control from the host computer

Best Regards

Marin

[Barend Onneweer]

It's always good to dream (mechanical shutter, optical viewfinder etc.)

But at the same some of us need to keep both feet on the ground, and make it work. That's what Rai and Markus apparently achieved, and that's what Obin and the Rob's are working so hard at.

Of course there are many ways to skin this cat (no harm intended...), but with all these ideas about FPGA's and other solutions, there's nobody around here that can actually build a camera that way.

On the other hand, the solution of working with a small PC connected with a cable to the camera seems to be very feasible, although difficult enough. But if the two Robs would get a reliable application that can write lossless RAW 24p1080 sequences to a PC with 2 disks, I'd be willing to put up with the limitations of having a cord connected to the camera. Probably a dual cord: one for the cameralink/ethernet from the camera to the PC (in the video-assist cart), and one back with a VGA signal so the cameraoperator can see what he's doing. I tend to have cables running to the camera anyway, video-assist etc. Not a big deal. And 9 times out of 10, I'll have power on set - so that's not a big deal either.

If this results in a clean 10 - or 12 bit uncompressed HD image, I'd personally be happy as a clam :-)

On another note: an optical viewfinder needs a prism to split the beam of light in two beams. Thus only half the amount of light hits the CMOS. Not good. Same goes for groundglasses etc. A 1-chip bayer image is not ideal, but if we get it to disk without compression, or other image polution it should be more than sufficient.

Just my way of supporting the efforts here - since I'm no good at programming or any other useful skills.

Bar3nd

[Régine Weinberg]

Bolex did it 80% to 20% 20 to the viewfinder