High Definition with Elphel model 333 camera

[Robert Schiebel]

Hi,

nice Sensor, but expensive!
At Framos/Germany 1pc. 1567Euro

[Wayne Morellini]

Andrey, what about that lowlight hires color sensor fro CCTV I mentioned a week or so ago, apart from that low light 170db low light one I mentioned last year. Is there any chance we could see these, or the Foveon X3, appearing for the camera.

Originally Posted by Andrey Filippov

.
10357 - a board with 8 CF slots - with 16GB cards it will be possible to have 128GB solid state "disk";
10359 - splitter/preprocessor for up to 3 sensor

128GB thats great, would be good if it had some frame to swap it in and out. How fast of an data rate can it handle, could it handle raw 720p or more?

This is going to sound silly, but I am interested if the multiple sensor option can be used, so that you can swap to an low light sensor during filming, or between visible, infrared and ultraviolet, or three chip?

[Oscar Spierenburg]

How about starting a new thread on the software and HTML programming part?
This thread has too many discussions going on at the same time.

I've made a new thread here: http://www.dvinfo.net/conf/showthrea...571#post618571

[Odd Nydren]

Hi all!

I've been following this thread for some time and I find this community and elphel _very_ exciting!

I have a few questions for Andrey:
(please forgive me if these have been answered before)

1. What sensor will be available for the 353 at first release?

2. When will it be possible to buy the 353 + that sensor?
(I dont mind being a beta tester with all it means)

3. With the first release, will it be possible to record video to a harddrive connected directly to the IDE connector? (or will it require custom 353 hardware programming not available at first release?)

4. What format will the recorded data have?

5. What frame rate / image sizes can be selected?
(..or just tell me the sensor name and possible 353 restrictions)

6. If it needs custom programming, where do I find info about tools / language etc? Just need a pointer. Any free compilers?

7. Will the camera casing have space to accomodate a small harddrive?

8. Any idea of power requiements?

9. Will the camera have a C mount?

10. If I use a simple SLR mount (nikon or canon) to C mount adapter (instead of a groundglass solution), what lens multiplier would be in effect?

11. Any idea what the sensor and 353 will cost?

Phew! quite a few questions...

What I would like to do is to come up with a small and portable solution as possible...hopefully with the camera having the HD inside it and controlled by something similar to a Nokia 770 (however the 770 only has wlan...no ethernet connector) using a custom flash/html interface. Im a 3d graphics designer / programmer and do not mind spending serious time to get this to work.

Any feedback would be most appreciated!!

Thanks!

//O.

[Odd Nydren]

Possible camera control devices having wired LAN:

OQO2
http://www.dynamism.com/oqo02/main.shtml

Sony Vaio UX
http://www.dynamism.com/ux/main.shtml

..however quite pricey.

Any suggestions?

//O.

[Zsolt Hegyi]

Odd, this thread has answers to most of your questions. Read carefully, you'll find them.

Zsolt

[Andrey Filippov]

Originally Posted by Juan M. M. Fiebelkorn

BTW if someone is interested Micron is sampling a 4 Mpixel 250 fps sensor.

And it is rather easy to build a camera (not so small) for that sensor - take 16 of the 10353 boards (one per output), powered gigabit switch and design a simple board to mount the sensor and have 16 connectors fro the 10353 boards.

[Andrey Filippov]

Originally Posted by Wayne Morellini

Andrey, what about that lowlight hires color sensor fro CCTV I mentioned a week or so ago, apart from that low light 170db low light one I mentioned last year. Is there any chance we could see these, or the Foveon X3, appearing for the camera.

I just can not work on more projects than I currently do

Originally Posted by Wayne Morellini

128GB thats great, would be good if it had some frame to swap it in and out. How fast of an data rate can it handle, could it handle raw 720p or more?

It is designed fro particular application where hard drives are not appropriate. And flash will not be removable - you may always use Ethernet connection. As for speed - it will depend on ETRAX FS ISE speed - 10359 has an FPGA on it.

Originally Posted by Wayne Morellini

This is going to sound silly, but I am interested if the multiple sensor option can be used, so that you can swap to an low light sensor during filming, or between visible, infrared and ultraviolet, or three chip?

That board is designed to simplify FPGA development for some image processing (no need to bother with the one that has a complex video compressor on it). ANd it can be used as a simple multiplexor, of course.

[Andrey Filippov]

Originally Posted by Odd Nydren

Hi all!
I've been following this thread for some time and I find this community and elphel _very_ exciting!

I have a few questions for Andrey:
(please forgive me if these have been answered before)

1. What sensor will be available for the 353 at first release?

Micron Mt9P001

Originally Posted by Odd Nydren

2. When will it be possible to buy the 353 + that sensor?
(I dont mind being a beta tester with all it means)

I hope - early April

Originally Posted by Odd Nydren

3. With the first release, will it be possible to record video to a harddrive connected directly to the IDE connector? (or will it require custom 353 hardware programming not available at first release?)

We do not have any adapters to actually connect the HD. We also do not have a camera body that will have room for one, that will come later. As well as the software fro recording - but it can be really simple to write on a disk from the softwaer point of view.

Originally Posted by Odd Nydren

4. What format will the recorded data have?

Same as camera can provide. Will start with MJPEG and I will restore Ogg Theora (ETRAX FS should be enough to handle some processing needed). All over formats - recode on a PC.

Originally Posted by Odd Nydren

5. What frame rate / image sizes can be selected?
(..or just tell me the sensor name and possible 353 restrictions)

Originally it will be up to (125/3) MPix/sec - the current FPGA code fro 333 can do that. Then - increase to hacve full speed of MT9P001

Originally Posted by Odd Nydren

6. If it needs custom programming, where do I find info about tools / language etc? Just need a pointer. Any free compilers?

To be able to do development for these cameras you do not need any expensive (or even non-free-as-in-beer) tools. There are 3 levels (not to count client/PC software)
1 - "Web design" - developing of custom user interface - just any web design tools - you can FTP results to tehy camera
2 - Software applications to run in the camera - you need a PC running GNU/Linux (any flavor) and software from our Sourceforge project page plus additional - from http://developer.axis.com - our software will tell you what to download.
3 - FPGa code development - free (for download) tools from Xilinx - http://www.xilinx.com/ise/logic_design_prod/webpack.htm

Originally Posted by Odd Nydren

7. Will the camera casing have space to accomodate a small harddrive?

It should have - in the future :-)

Originally Posted by Odd Nydren

8. Any idea of power requiements?

Model 333 needs a little under 3W. 353 will need about the same - not to count additional modules - hard drive or flash.

Originally Posted by Odd Nydren

9. Will the camera have a C mount?

CS-mount as it is more universal for cameras - you can install both C and CS lenses

Originally Posted by Odd Nydren

10. If I use a simple SLR mount (nikon or canon) to C mount adapter (instead of a groundglass solution), what lens multiplier would be in effect?

That depends on particular sensor that will be used - each sensor usually has info on the Internet. And, BTW - we'll probably have control for EOS - at least we do now have it for our large cameras

Originally Posted by Odd Nydren

11. Any idea what the sensor and 353 will cost?

I hope it will be about the same as 333.

Originally Posted by Odd Nydren

Phew! quite a few questions...

What I would like to do is to come up with a small and portable solution as possible...hopefully with the camera having the HD inside it and controlled by something similar to a Nokia 770 (however the 770 only has wlan...no ethernet connector) using a custom flash/html interface. Im a 3d graphics designer / programmer and do not mind spending serious time to get this to work.

Any feedback would be most appreciated!!
Thanks!
//O.

Yes, Nokia 770 is a good candidate and it will be possible to add USB WiFi to the camera. Currently we do not have a design for I/O board - main board has IDE, USB (host) and some GPIOs in high-density connectors, so some daughter board with connectors, usb audio, etc. will be needed.

[Andrey Filippov]

Originally Posted by Zsolt Hegyi

Odd, this thread has answers to most of your questions. Read carefully, you'll find them.

Zsolt

We need to organize data more like FAQ - thread is so big. I'm very busy now so any help in organizing data (starting with this therad) on our WIki (wiki.elphel.com) will be very appreciated.

[Matteo Pozzi]

Hi to all
I have listened the request of Andrey ....looking at the problem of this thread so I'm making this page on the elphel wiki:
http://wiki.elphel.com/index.php?tit...face_.28GUI.29

for now is in an early stage but is just something for an hour of work! :-)

[Zsolt Hegyi]

Andrey,

I'd like to send you a mail but your mailbox is full...

Zsolt

[Zsolt Hegyi]

Originally Posted by Andrey Filippov

but it can be really simple to write on a disk from the softwaer point of view.

And is there a chance to output two streams from the fpga? One to the network and the other to the disk?

I was thinking of some multiplexing stuff: every frame gets encoded into the lossless format but only every n'th frame gets encoded into theora. We might do this by buffering some theora frames and do the encoding parallel to the lossless encoder. The problem is, if we mix up the two streams how can we demultiplex them for the appropriate two-way routing?

As far as I know the 333's memory controller is not capable of addressing as it writes the output blocks in an incremental order (correct me if I'm wrong). We might need a one bit addressing scheme with 0 for hdd and 1 for the network then we could demultiplex the data. But I don't know how much the memory controller builds on the ram chip's capability to write larger blocks of data incrementally which is probably more effective than continous random access.

As for the Extras applications I suppose an additional hard disk writer utility wouldn't be a big overhead because all transfers go through dma.

(My encoder won't need much space in the fpga so that's not a problem.)

Originally Posted by Andrey Filippov

it will be possible to add USB WiFi to the camera.

Do you mean a usb2 host?

Zsolt

[Andrey Filippov]

Originally Posted by Zsolt Hegyi

And is there a chance to output two streams from the fpga? One to the network and the other to the disk?

I was thinking of some multiplexing stuff: every frame gets encoded into the lossless format but only every n'th frame gets encoded into theora. We might do this by buffering some theora frames and do the encoding parallel to the lossless encoder. The problem is, if we mix up the two streams how can we demultiplex them for the appropriate two-way routing?

I believe such dual output can complicate things a lot. Why do you really need both at the same time?

Originally Posted by Zsolt Hegyi

As far as I know the 333's memory controller is not capable of addressing as it writes the output blocks in an incremental order (correct me if I'm wrong). We might need a one bit addressing scheme with 0 for hdd and 1 for the network then we could demultiplex the data. But I don't know how much the memory controller builds on the ram chip's capability to write larger blocks of data incrementally which is probably more effective than continous random access.

Memory controller is different for MJPEG and Theora - Theora data structures are carefully built into the controller to provide efficient bank interleaving access for all 6 types of accesses for Theora (line scan order from sensor, Hilbert to comressor, r/w previous frame coefficient, r/w DCT coefficients. Each block access wastes some clock cycles on start/finish, so average (not peak) data transfer rate is about 95% of the theoretical (125MHz, DDR, 16bits - 500MB/sec). Different codecs will require different data structures and memory controllers - it is a significant part of the codec design.

Originally Posted by Zsolt Hegyi

Do you mean a usb2 host?
Zsolt

No, USB 1.1 - I thought I mentioned it. The main application - audio. And Axis have nice documentation on their ETRAX FS at http://developer.axis.com/products/etraxfs/index.html (including USB port)

[Zsolt Hegyi]

Why do you really need both at the same time?

Because during the hard disk recording we'd like to see a real time viewfinder image so we can move the camera while shooting. I think the guys here were talking about the Nokia 770 believing that the two types of streams are actually possible. (It's another question if somebody wants to record the theora stream on the hard disk because that's possible.)

Different codecs will require different data structures and memory controllers - it is a significant part of the codec design.

The encoder I'm planning doesn't rely very much on special data structures: for each pixel it needs the previous pixel value on the same coordinate and the output goes directly into the output stream. Therefore what I'm hoping is that I can use the current controller for these two fundamental functions.

No, USB 1.1 - I thought I mentioned it.

You did, but I thought usb1 is already included on the board. I asked usb2 support because wifi cards are usually work with usb2 (54Mbit/s) and the speed of usb1 (12Mbit/s) is hardly enough for real time video streaming. For sound, yes, we might find some external modules that record on usb1, altough the newer models are too using usb2.

Zsolt