Hardware...
 

Thanks for your thorough answers Andrey - much appreciated!

Yes - zsolt, I was hoping to use the Nokia 770 (or a similar device) as a viewfinder...and record the hi-res imagery direct to disk, in-camera.

..however using USB for audio I think would be unnecessary...?

I usually record audio separately using this:
http://www.m-audio.com/products/en_u...2496-main.html

..as it gives much more flexibility.
(plus it has phantom feed if needed - and can be found on ebay much cheaper than rrp mentioned on m-audio's site)

Another thing:
Andrey you mentioned you have EOS support on your larger cameras...could you please tell us more about that? How much support would find its way into the 353 and what are those larger cameras?? *curious*

..and matteo? Great work on the wiki!!

thanks

//O.

Yes, that can be a good solution fro now, but I would still like to have a camera with audio in it. First step - be able to use off-the-shelf USB audio.


It is our current model 323 camera and the 363 under development (using 35mm format CCD sensors). We are able to control motorized lenses (at least most commands) and 10347 board has a connector for it. I also plan to have such connector on a planned daughter board in 353, together with IDE connector, USB and possibly a sound chip.


Thanks from me too. I'll contribute when I'll get to a milestone of first images in 353.

For the view finder it would be easier to use low-res MJPEG (still needs modification of the memory controller to implement decimation in FPGA). Doing that for Theora may be to expensive.

You may have that data if you put your module between the sensor and SDRAM (were "gamma" processing is now).

Using USB1 is as easy as connecting to pins on the CPU chip. USB2 needs a major development around it - not as a simple add on. And we already have Ethernet port for video streaming with most of the hardware and software included with the ETARX FS.

Has anyone tried this to reduce the rolling shutter effect?

http://www.guthspot.se/video/deshake...tter%20setting

When I have some time to test it, I'll post the results.

Anyway, correcting the rolling shutter is like a crazy thing if you think deeply about it.
Yes you can correct a "general" "global" distortion, but if your camera is not moving and things inside a frame move fast, how will you correct the wobled moving object?
Most of object movements take place on the X axis, so I guess even having the sensor turned 180 degrees would give a better end result.

Aside from the simplest solutions I already mentioned, now I'm planning to redesign the memory controller around the lossless encoder. It will generate two output streams: full resolution bayer to the hard disk and a quarter resolution 8bit grayscale with 5fps which can go to the Nokia 770 through wifi with the current web streaming architecture. Hope I'll have the time to do this. If somebody wants to do the streamer that would be of great help tough...

Yes, I forgot that we can plug wifi into the ethernet connector also.

Zsolt

look at the gui I'm working on
http://www.webalice.it/teo.poz/elphel_GUI/
a layer over the normal interface :-)
seeya
Matteo

Matteo, the problem I could see is that it is (or at least was) not possible to have transparent objects (like non-rectangular buttons) over the live video - That is why I had to have these controls below the preview window in camvc.

I'm not a flash programmer, I know flash cause I've used it to make a cartoon
but I've read this http://www.adobe.com/cfusion/knowled...fm?id=tn_14201
maybe I'm in mistake but I think that is possible to use a background html page
...one thing that I haven't understand is how can I link a button to a setting variation of the parameter of the camera without changing the html page
I need to learn javascript to do a real time variation?!?

Flash GUI...
 

Hi Matteo!

check out this page:

http://www.echoecho.com/flashbuttons01.htm

Its a tutorial on how to make a flash button with a piece of flashscript that sends parameters when requesting a html page...without necessarily changing the current browser web page. (read the end of the article)
I dont think you can put video in the background and flash on top...I would suggest maybe using frames?

I would gladly have written it for you but I do not have the camera...yet!
However I intend to code my own GUI for it as soon as I have access to a 333 or a 353...

Any questions, feel free to PM me.

good luck :)

//O.

Odd,

I don't think that button will work - as I understand the problem with non-rectangular objects on top of video pluging is with plugin (or plugin mechanism - I don't know) rather than button itself. PNG with javascrit can work the same.

BTW, in camvc software I used a single big png file for all the buttons/states to reduce number of different files on the server (camera). Code just sledes that file behind a small square window. Here is the file:
in the camera - http://camera5.elphel.com/images/camvc_buttons.png
on Sourceforge project CVS - http://elphel.cvs.sourceforge.net/*c...vc_buttons.png

There are multi-layer sorce files also - just replace .png with .xcf

Here is a Makefile that contains list of all the camvc-related files in the camera:
http://elphel.cvs.sourceforge.net/*c...camvc/Makefile

Andrey,

Would it be possible to make a separate HTML frame on top or bottom of the exiting interface and save it to the camera?
And what software will be suitable to edit the html page?

EDIT:
Note that the HTML is discussed in it's own thread: http://www.dvinfo.net/conf/showthread.php?t=85568

The 770 is being replaced (or already has been) see www.brighthand.com archives. If PSP had Linux (through mod should be possible) that would be good. There is Moko Linux phone from FIC. Pity, there is 720*480 Windows phone going at Iphone mentioned recently.

About the internal and external view screen. Can I present there different alternatives to this problem, from the one that requires the most work on your part to the one that requires the least because it just uses the existing photo-jpeg:


Why not produce two related streams, so that the external stream builds on the hard disk version.

The Hard disk version could be differential bayer compressed intra frame compression. Using buffer registers on FPGA you could keep an significant amount of processing off the memory bus. Taking the bayer produce (I've had 3 hours sleep in recent days now I can't remember the terms) but do culling as in Jpeg in the process. Use the culled version for the external view adding intra-frame. This can be aggressively culled to fit wireless or 100mb/s.

You can give more detail to the center of the screen, or only when an small focus area box is brought up. You could pass the hard disk frame out at low frame rate for focusing assist when user requests it.


Now, the simpler way, use Photo-Jpeg grey-scale (normalised to one primary, as I discussed before, can be done in FPGA before goign to memory) at highest quality to disk. At the same time do an more aggressively compressed version to the external device. You could make the external version an normal photo-jpeg, whatever is easiest for the external device to process. This way you largely just use Photo-jpeg and share some processing.


Now the simplest way. Do normal Photo-jpeg at high quality to disk, at same time do the aggressive compression for external device (which should have photo-jpeg support). The external version is based on an subset of processing done for the disk version.

Ogg might be totally unnecessary, if you have enough bandwidth to the hard disk. Ogg might have some advantages intra frame.

Have an good day.

With HTML, isn't it possible to overlay the buttons and the maps, with transparent pixels to make rectangular buttons look differentially shaped?

Wayne,

I believe you could see it on our online camera. The problem is when such object is above video plugin, not above regular image/text