Sorry, I have noticed that I use the words interface, front end, and application interchangeably. Really they all mean the same thing. A front end is the client side of an application. Its what a user will see and interact with. Generally the term front end is associated with a website.

I just tried the buttons, did not realize it was flash. Thats pretty neat, looks like you through it through. You could easily interface this with the 353 by making calls to PHP files.

http://www.kirupa.com/developer/actionscript/flash_php_mysql.htm

I know that I can interact with php with flash :-) but not already how so thanks for the links
I think I have also to understand how php really works before :-)
also I want to leave the flash web interface and make a simple html one so that it would be faster and "really open source"

I've made some progress with the viewfinder. Making everything adjustable so I can focus everything right will be difficult, but you can see it should work out.


It seems that I have to cut a little bit from the side of the body too.

For a while I worked on a simple input device using an AVR atmega128...
It may be even possible to plug such a device directly into the usb ports avaliable on the 353.

Hi Daniel,

I've been thinking about that as well - and using this:
http://www.arduino.cc/

I bought the USB version and its very easy to program.
You download/install the compiler, write a few lines of code, hook up the USB connector and just click the flash button...and a few secs later your code is running on the ATmega board. I made a time lapse controller for my 20D in about an hour. :) - In the end though I decided it would be too much work and I think I'll be controlling my Elphel with a touch screen mobile phone running a web page from the Elphel via wifi from a USB dongle in the Elphel. That means no building stuff...keeping it a software solution. That in turn means whatever I end up doing is so much easier to share with others...and that makes it a lot more fun. //O.

Yes, we do.


Many thanks for your quick reply Andrey! I have a few more questions regarding the purchase but I'll put that in an email to you instead. Thanks!! //O.

I had some great tests with the viewfinder. Very clear and bright image. Perhaps it'll even be good enough to focus most shots. I still have to put everything together though and make a robust housing around the tubes.

Daniel, what are your thoughts about an internet tablet and the Elphel+HDD? (skipping a Laptop/PC) I'm thinking about a combination of the optical viewfinder and 'only' a 2x or 4x focus aid on the tablet screen. (not necessarily at full frame rate)

The two ways to get a preview image from the 353 is to either use the imgserv which will provide a jpeg image of the last captured frame, or to use the real time RTSP stream.

The problem with both options is that they are going to be very high resolution images of 2000x800 for example. At the moment the camera has no ability to downres images for preview. This is overkill for small devices such as cell phones with lcd screens of ~64x64, PDAs 240x320, or internet tablets of ~640x480 - 800x400.

Not only does it become a problem to display such an image, but its a lot of data to process and the CPU capabilities of such devices are limited. Furthermore, when displaying the full frame most of the data is processed and disregarded.

One of the options I was considering is to have a device server. Clients such as a PDA, cell phone, internet tablet, laptop, or desktop pc would connect to this server. The server would then format the RTSP stream from the camera optimized for the connected device.

For example, the DOP may be using an internet tablet or a similar device to get a preview of the picture. He would turn the device on, it would connect to the server over WIFI and he would get a properly formated image of 800x600 for example. If he wants to zoom in for focus aid or detail all he has to do is use the touch screen, and the server would send a new, or reformatted video feed/image. Using this device he can also configure the camera and more.

Or here is another example. The visual effects supervisor would like to get an idea of how well a scene matches a miniature shot, key out the bluescreen, do a basic comp, overlay videos/images, or would just like to simply give the director an idea of things are coming together. Using the handheld device connected to the server he can do this by the touchscreen, it would be computed server side, and receive the realtime or image preview.

The possibilities are endless.

The disadvantage of using such a server is that its going to make the system more complex. Additional power will be required, and it would make the camera package bigger. But you did ask about my thoughts :)



It is not imposable to just use a high resolution image from 353 directly on the hand held device. Devices with wifi abilities are best suited and would take minimum work to implement. You could possibly get the imgserv jpeg at ~6 FPS. The RTSP stream requires a lot of cpu resources. I believe the recommended was 3ghz for optimal performance, so I don't think thats a realistic option. It is also possible that requesting an imgserv image at a high rate will take a toll on the camera CPU and lower the recording frame rate.

For the viewfinder mode it will be rather easy to make FPGA to provide 1/8 (or even 1/16) resolution (so 1/64 or 1/256 number of pixels) of uncompressed data in parallel to the video, virtually without using additional resources (so not slowing down the recording).

There is also another feature in the FPGA - not yet used (and needs some code to connect it to the output) - the amount of high-frequency components fro each of the 16x16 macroblock used in compression. This data can be used to provide info for the quality of focus if you average that data over selected region of the frame that you want to be in focus.

This data can be used to provide info for the quality of focus if you average that data over selected region of the frame that you want to be in focus.

Andrey, ..this goes a bit above my head I guess. With data, do you mean in the form of a histogram or something?

What I meant with focus-aid, was only to show a the center of the image zoomed in 2x (like used in the ajax interface) and show that on the mobile device. The whole image would be unnecessary, because I would have it on the optical viewfinder. This is just an idea, because I understand that not everybody would be able to make the viewfinder... but than again, it would be a simple starting point.

Oscar,

It is more difficult to implement zoom window (w/o slowing down the main data) than to provide (per 16x16 tile) "focus quality" data, it is a by-product of the compression process, most of it is already implemented. Next time I'll get into FPGA code I can add output of such value over the specified window, we can display it as a bar in the UI.

For the viewfinder mode it will be rather easy to make FPGA to provide 1/8 (or even 1/16) resolution

Excellent!!! That would be VERY useful!

The resulting preview image would be perfect to ensure framing and make it very easy to control the camera using a wifi enabled device...like a cell phone with touch screen or an internet pad. Resulting in a very compact setup.


There is also another feature in the FPGA - not yet used (and needs some code to connect it to the output) - the amount of high-frequency components fro each of the 16x16 macroblock used in compression.

Even better! I was hoping for a feature like that but thought it was impossible. (due to CPU power) This would be an excellent way to check image sharpness.

This data can be used to provide info for the quality of focus if you average that data over selected region of the frame that you want to be in focus.

Exciting news!

If possible - please implement this! It would make it totally possible to not just control but also frame/focus the camera with a wifi device.

I'd love to build a PHP web page where I can switch between a 1/8 or 1/16 sized image stream for framing...and be able to switch to a "focus quality" image to check for sharpness. Add to that a few controls for exposure/timing etc and we have a really cheap and compact camera controller. very cool

//O.

great that would be heaven, will it be ready some days ??

great that would be heaven, will it be ready some days ??

Yes, sure. And focus quality in a specified window can be really soon.

What do you think - maybe this can be used to control the camera?
http://linuxdevices.com/news/NS2593777857.html

The optical viewfinder is almost ready to be mounted on the Elphel. I will make it detachable, so I can always adjust things. And I'll probably make another one without the image inversion, so I can see the image from the 35mm adapter right way up.

Andrey, I'm very exited about implementing the focus quality. How can it be used in the control-panel (how should you read the focus quality)?

Oscar, that will just be a value that will be updated after each frame compressed. The focus will quality will be controlled fro specified window (in 16x16 pixels increments.

It has some resemblance of the JPEG image size (but that depends on all the frame, not just the center area). If you make a series of images with different focus settings (all the rest being the same) you can compare image size - the larger, the better is the focus.

cool the focus aid!
maybe ca be used the magnification area
1- position the magnification area in the point of the picture we want to focus
2- press the button focus aid
3- roteate the lens focus wheel to the right
4- press button focus aid
5- roteate the lens focus wheel to the left so that the software can calculate back the image size of the same (more or less) frame of the right rotation of the lens focus wheel and when the size of the frame is similar to the higher achieved a green indicator on the gui will tell you!

nice work oscar!

The viewfinder is about ready and working great. I haven't had time to put the sensor and main board back in the camera body, but that'll be ready tomorrow.

Yes, sure. And focus quality in a specified window can be really soon.

What do you think - maybe this can be used to control the camera?
http://linuxdevices.com/news/NS2593777857.html

That would be awesome!
..might be tricky though as it doesent have wifi...? ..or maybe there's a way to hack it and use the USB port?

Maybe this one would be easier (although not linux based :( )
http://www.youtube.com/watch?v=lbdmMpwgErE&feature=related
Its got opera, wifi and fast mpeg-2 playback.

Err - I just realized - archos is linux based and uses Qtopia... http://trolltech.com/products/qtopia/qtopiainuse/qtopiadevices

Hi all,

I made a little framerate calculator...and thought you guys might find it useful too. You can enter the sensor size and full frame framerate and it will give you an approximation of the framerate att smaller image capture sizes.

Please let me know what you think.
If you need any other features - ask and I might put it in.

http://www.oddpixel.com/fps.html

//O.

cool the focus aid!
maybe ca be used the magnification area
1- position the magnification area in the point of the picture we want to focus
2- press the button focus aid
3- roteate the lens focus wheel to the right
4- press button focus aid
5- roteate the lens focus wheel to the left so that the software can calculate back the image size of the same (more or less) frame of the right rotation of the lens focus wheel and when the size of the frame is similar to the higher achieved a green indicator on the gui will tell you!

nice work oscar!

or better :
a upper green arrow if the current frame is bigger than the previous 2 and a red down arrow beside the magnification 2x, 4x square if the current frame is smaller!
:-)

or better :
a upper green arrow if the current frame is bigger than the previous 2...
:-)

Hmm since it's just a value representing the sharpness of an area...I could write a little flash app that plots a little graph + does the arrow business on a continous basis.

It will not be of much help on a mobile device (maybe a small piece of php would work on those) but on a pc it should give pretty fast response!

I bought my 353 yesterday and I'll make that app as soon as it's arrived!! :)
(and post here of course)

cheers
//O.

That's great to see more people using the 353.
I have to put everything together again to use mine, but I have to get a new IR filter (I couldn't prevent it from braking when I cut out the piece for the viewfinder.) .(
In the meantime, I'm building a rod support for the Elphel and all other parts (adapter or lens, batteries and things like that.)


Elphel camera's come powered with 48v DC (unless you get a custom made version with 12v). What are the possibilities to use battery power?
Is there some affordable voltage doubler(or 4x when using 12v) available?

Elphel camera's come powered with 48v DC (unless you get a custom made version with 12v).

Ah! I didn't know I could get one custom made for 12v...hmm too bad.

The 48v issue should be possible to fix using a DC to DC converter...I'll have to read up on that been a long time since I did electronics.

We should probably ask Andrey...

I don't know of any existent ones for the correct power, this is something I was planning to build - a simple 48VDC power injector that will take power from lower voltage source. It is even possible to make one that will take 5V power from 2 USB ports (1 is not enough)

Maybe we can keep in mind, lots of the batteries for 'consumer' DV camcorders are 7,2v. Is it also possible to convert that to 48v ?

I don't know of any existent ones for the correct power

That explains why I couldnt find any when browsing the net ;)

this is something I was planning to build - a simple 48VDC power injector that will take power from lower voltage source.

..is it still in the works - just postponed?

It would be real cool with a solution that doesen't involve modifying the camera itself, but instead something to hook between the camera network connector and a battery source.

If you decide to make that power injector Andrey, please put me on a pre-order list ;) hehe I'd love one thanks!!

Has anyone seen a good solution to hook up a Canon EF lens to C mount? The only one I found was birger.com's _very_ expensive solution...

I have a few C mount adapters & EOS extension tubes on order from Ebay, hoping to find a way to make one - with or without a groundglass.

Anyone tried this?

I saw the shots of your EOS experiments Andrey and that got me inspired! Btw I got the camera today!! Thanks for the excellent packaging and the extra body! Much appreciated! Doing my first tests now...

Odd,

Large (EOS) lens makes sense _only_ with the ground glass (or similar) adapter. If you just use EOS-to-C-mount adapter you will get quality worse than with the modern C/CS mount lenses, the sensor pixel is just 2.2 um x 2.2 um. The sensor size is about 5 time smaller than 35mm frame, so to get 5MPix resolution you need a 35mm lens capable of (5 x 5) x 5Mpix = 125MPix resolution

The sensor size is about 5 time smaller than 35mm frame, so to get 5MPix resolution you need a 35mm lens capable of (5 x 5) x 5Mpix = 125MPix resolution

Ah! But of course...doh! Thanks for the clarification!
This is very much a learning process to me. :)

What I really want is to do a groundglass solution as I want the depth of field you get using the whole lens... - the EOS directly to C mount was more of a starting point.

Well you made it quite clear I better go for the groundglass solution right away. I better go read up on optics. ;)

Odd,

As the lens is removable, I would recommend to use a fancy one with a ground glass and just a C-mount type one as an alternative. To get the full resolution you need a really good one, but just to experiment with the camera operation, you can probably get 1/2" format C/CS-mount lens for the same price or cheaper than the adapter.

Definitely!
I've already put a bid on a cheap c mount lens just for tests :)

I really want a groundglass solution. I do photography with an Canon 20D camera and have a few real good lenses that would be real cool to use with the Elphel.

I need to read up on macro lenses & how to make a simple solution to focus on the ground glass...any ideas Andrey? ..or know of a good site that I should have a look at?

The EOS mount part & groundglass is pretty much solved via the info at http://www.jetsetmodels.info

I'm sure you can find more info on it even in this forum, but if I had to build something like that I would use two standard lenses on the camera side, not just one. The closest one to the camera (1/2", C/CS-mount), let's say, f=5mm, the second one - five time longer - f=25mm (perfect ratio you can get as a ratio of a 35mm frame size to the sensor size). The first lens will go to the camera thread in normal way, the second - attached other way - facing outer side (designed for infinity) to the first lens. This second lens should be a 35mm format - the ground glass will be on it's back side. Maybe it is better to start with somewhat longer focal lengths, some 8-12mm for the first one and proportionally more - for the second one.

Such arrangement will reduce aberrations - each lens is used as it is is designed to - one side - to infinity, other - to an object in focal plane. That might turn out to be important as I think most ground glass adapters were designed for camcorders, not for the high definition (forgive me if I'm wrong - I am not an expert in this area).

The space between faces of 2 lenses is an ideal place for an IR filter - we tried to use it as thin as possible (t=0.3mm), but still spherical aberrations introduced by a glass plate can be eliminated completely if the filter is in the parallel beam.

The space between faces of 2 lenses is an ideal place for an IR filter - we tried to use it as thin as possible (t=0.3mm), but still spherical aberrations introduced by a glass plate can be eliminated completely if the filter is in the parallel beam.

I'll be able to test this... because I broke my IR filter. I'm very curious if it'll get better results. I'm rebuilding/improving my 35mm adapter right now, so I can choose/experiment where to put the IR filter.

Odd, I'm using a microcrystalline wax screen. I prefer this because of it's image character and lack of electronic parts (no power needed)
A normal (non moving) ground glass would be far to grainy for a high resolution camera. Also, a moving ground glass, can also get a small movement in and out of the focal plane and blur the image a bit.

Do they use ultrasound actuators to vibrate the ground glass too?

Or maybe it is possible to use running waves (probably surface ones as compression one will have too high frequency for required wavelengths)?

Andrey... and others, this doesn't seem so different than the one I've made: http://www.pstechnik.de/en/digitalfilm-si2k-optical-viewfinder.php
It seems that my optical viewfinder will also prove to be perfect for focusing, but I can only check when I can compare with HD footage from the Elphel, focusing on a focus pattern.

Now, can anyone tell me how a viewfinder extension works? it shouldn't be to complicated. Could it be home-made?

(I've finished my Elphel rod support. I'll post pictures tomorrow.)

Here just an update. I nearly finished a rod support suitable for the Elphel and all the things that need to be attached from time to time, like the 35mm adapter and bigger lenses than normal cs-mount ones.
Here is a picture of the support and the new version I'm making for the 35mm adapter (need to make a smaller (still 35mm) wax screen for this one)

Dear Oscar hat is tooo god to be true,
since I'm here I've been fighting, crying that rails withe lenses, compendium and a REAL viewfinder are needed as you have to look thru the lens and with the other eye the scene what ever it will be to have a re true feeling of film making and get the right focus the lens.. Sorry sounds better than sex !!!
A digital poor mans Arri is finally arriving. Exit to fiddle around with diva like used 16 mm equipment. Wow.
Dear Oscar is there any way that me
OT !!!!!!!
now Régine, since years,yes I've been Intersex just the day born; so now a petite blonde with brains but
Ot End ......
still film making and a Linux geek can build with more then two left hands quite good home tools such a beauty.
What about how to do guide for bonafide idiots.

You did exactly what is needed out in the field.
Pump it in a smallish 104 Pc with big disks, Cinarella and co will do the rest.
A Gnu project of impact not seen every day.
I' m fascinated and dreaming a great Forum.

Hi all,

Can anyone tell me how to buy an elphel 353 ?
Is it already available for sale ?

Thanks,

MGC

Ronald, thanks for you kind, but... strange words. I guess I know what you mean, and I am encouraged.
I really like combining the 'best of the old' and 'the best of the new'. One should respect traditional technology.

Marie, you should contact Andrey Filippov at Elphel to buy a camera. http://www3.elphel.com/en/contacts

Hi all,

What are the real specifications of the elphel 353 ?

Can i shoot at 24/30 fps with a 2K resolution ?

MGC

Marie, I don't know yet, the wiki says they tested 1920x1072 video at 24 FPS.

I was discussing the rolling shutter artifact with Andrey and Daniel. I never really tried virtualDub deshaker, but my first tests (with older 333 footage) was pretty good.

Before (http://community.elphel.com/videos/original.avi) and after deshaker (http://community.elphel.com/videos/with%20deshaker.avi)
I just thought I'd post it here. In my opinion, the effect isn't a big issue when you bear it in mind while filming (use a tripod, don't do fast panning etc.).
Hand held filming is not really an option with the current Elphel models, for me that's no problem.
If people have thought on this, lets discuss it.

Marie, I don't know yet, the wiki says they tested 1920x1072 video at 24 FPS.

http://wiki.elphel.com/index.php?title=Image:1920x1088x25fps.jpeg

Oscar, it looks like you've accomplished a lot with your camera. Well done!

Since it appears that you are working with a beamsplitter block or plate between your OVF and the camera silicon, I'm wondering how you are collimating the OVF to assure that your eye and the silicon are at the same plane of focus. Do you have a varifocal OVF, and have you been testing it against a target? Not a point of criticism, I'm just curious.

I ask because I have an old 16mm Kinoptic orientable VF from an Eclair ACL that I may attempt do something similar with.

Andre: Will the 353 take an external TTL pulse control?

Andre: Will the 353 take an external TTL pulse control?

That will be a part of the 10369 board I'm working on now. It has a 4-pin modular (like telephone) connector for sync I/O (with the optocoupler, nominal 5V input). It will be possible to wire multiple cameras and get precise lock (to a pixel period).

Andrey

I'm sure you can find more info on it even in this forum, but if I had to build something like that I would use two standard lenses on the camera side, not just one.

Thanks for your feedback Andrey!
..it made me realize two things:

1. What you said is what I need to do.
2. I need to get the harddrive to work, get used to manual focus, make a gui that works for me...and THEN think about a ground glass solution ;)

I got my lens today and now I'm learning the camera...exciting!!

Odd,

there is a cool program (mostly FPGA code) in the last software (7.1.7.11) on Sourceforge for focusing. I believe it is useful and fun to play with.

Andrey

...working with a beamsplitter block or plate between your OVF and the camera silicon, I'm wondering how you are collimating the OVF to assure that your eye and the silicon are at the same plane of focus. Do you have a varifocal OVF, and have you been testing it against a target?

Brandt, I've made the VF adjustable (making sure I it's roughly in the same range of the sensors focal point.) I've just began comparing the picture from the sensor with the image in the VF. The eyepiece lens will capture the final image (functioning in a way as the focal plane) so I can ultimately fine adjust the focal point with the eyepiece. There is also a rough adjustment possible in the 'flipping' tube part.

I ask because I have an old 16mm Kinoptic orientable VF from an Eclair ACL that I may attempt do something similar with.
It would be a very nice starting point, but I choose for the 8mm camera VF, because the sensor size is nearly the same. You'll have to find a way to 'zoom' into the 16mm image to match the sensors image.

One other thing I'm working on is a viewfinder extension, making it easier to look through the VF. If possible, I'd make a separate extension, which flips the image again, so I'd see the 35mm adapters image right way up. I'm consigned about the light loss though...

...I'm very anxious to play with that new feature, Andrey. I think it'll prove to be a perfect combination with the optical viewfinder.

Oscar, it now combines two things - destructive (embedded into image) visualization of what the program is looking for (on the screenshot you can see - the brighter - the more high-frequency components are in each 8x8 pixel block. When you have foreground, background (and many in between, like with your lens adapter) you will see highlighted the features you focus on.
Other (green diagram) shows the total amount of hi-frequency components in the WOI regardless of if the image shows the focusing window, or not. Current value is displayed at the bottom as a horizontal line, the whole image moves up so you can see the history.

With USB port in the camera it is possible to attach audio output and provide different pitch tone dependent on that value :-)

The visualization will later be non-destructive, it will be output as one of the options for the low-res viewfinder DMA channel independently of the main images.

Andery

For anyone who isnt bold enough to build your own optical viewfinder...
..maybe this would be of interest:

http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&ssPageName=STRK:MEWAX:IT&item=320210604120&_trksid=p3984.cWAT.m240.lVI

Similar can be found by looking for "pan cine" or "VIEW LEG" or "ANGENIEUX"

I might get one myself...was about to bid on the above one but decided to wait until I'm sure that's the route for me...

//O.