Thursday, February 28, 2013

AS3: Custom Components Flicker / Mouse Event Bubbling

Problem:
While working on a project, I added some button components within another component and then had the buttons hide unless the mouse was over the entire component. Each time I would move the mouse around over the component, the buttons would show and hide quickly causing a flickering.

Solution:
I had forgotten that there is a different between MouseEvent.MOUSE_OVER and MouseEvent.ROLL_OVER. Using MOUSE_OVER defaults the events to bubble which causes other listening items to respond to that event whereas ROLL_OVER doesn't bubble the event.

Wednesday, February 27, 2013

Raspberry Pi: Remote Canon Camera Viewing and Control

This is my ongoing wireless video streaming project from my T2i using Raspberry Pi.

NOTE: I will organize this better when I have more time, I promise.

One possible option: http://www.auvidea.com/index.php/theme-styles/2013-06-22-21-23-34/raspberry-pi-hdmi-in
This would allow an HDMI in (non-encrypted) and also would provide a wifi system which could potentially encompass everything I wanted to do and do it better.

Another non-pi options: http://www.auvidea.com/index.php/theme-styles/2013-06-22-21-23-36/encoder-e100

Interesting Reading: https://plus.google.com/+RyanMatthews/posts/btdqMZqCzEA

In the Beginning:
I though it would be really nice to have some way to view my live video feed from my Canon T2i remotely from a mobile phone or tablet computer. It has crossed my mind that using this setup, I could potentially remote control the camera using PTP (Photo Transfer Protocol) and MTP (Media Transfer Protocol). I realize there will be a need to a computer to handle mediation between the camera and the remote viewing device. Only one other commercial device has been found which is for serious cameras and sells for $1,500. I'd like to do it for less than $100 using the Raspberry PI.

Some similar projects in regards to still photography acquisition and transfer:
http://islandinthenet.com/2012/08/23/hdr-photography-with-raspberry-pi-and-gphoto2/
http://davidhunt.ie/?p=2641

References:
http://blog.waynehartman.com/archive/2009/01/07/edsdk-to-remote-control-your-canon-camera.aspx
http://www.graphics.cornell.edu/~westin/canon/index.html (powershot)
http://gphoto.org/doc/manual/quickstart.html#using-gtkam
http://islandinthenet.com/2012/08/23/hdr-photography-with-raspberry-pi-and-gphoto2/

10/01/2012 (or so) 
I purchased the Raspberry PI model B (256MB) (512 wasn't available yet) online from Allied Electronics for $35 plus about $8 shipping.

10/05/2012 (or so) 
I determined the design and hardware requirements if this is to be obtainable. The Canon camera has an EDSDK which allows a developer to write programs that will interface with the canon Firmware and hardware (using APIs). The EDSDK however only appears to be supported on Mac and Windows (Linux not officially supported). One of these controls is the live view data. Using a USB connection between RPi (USB2.0) and the Canon (USB2.0) will allow the Pi to establish a connection with the camera to control it. A WiFi dongle will be used to enable the Pi to be a WiFi hotspot. Pi will use something similar to VLC or ffserver to stream image data over a wireless protocol to the listening device. The device will have a custom built solution to allow it to render out the media stream. Using a socket (or even RESTfully, commands could be sent back to the Pi to change focus, contrast, etc).

10/19/2012
Rasberry Pi shows up but I can't do anything really for it. Bought an SD card ( $8 / 16 GB Class 10 like this one) and an adapter from HDMI male to DVI female (about $5; who knows if this will work).

10/24/2012 
Adapter and SD card arrive, switched SD card with wife for 8 GB Class 4. Adapter still needs to be tested while OS formatted SD is in the Pi.

10/25/2012
Did some testing with the PI and research about Canon cameras and USB control. Applied and obtained the EDSDK from Canon. SDK appears to support control for the live view along with histogram data, etc. Pi boots up wheezy raspbian and is working great, still haven't tried it with the HDMI/DVI adapter. It's running quick though. Much better it seems then it did on the emulated version.

Preparing myself for adventures in C. Planning to try to remote access Camera data via Windows 7 desktop to ensure it will work and that I can get the data I want from the Canon. If all goes well, I will then move the programming onto the Pi and begin networking the hotspot and streaming services.

10/30/2012
Been thinking a lot about how to get the RPI setup to work better and did some more research into gphoto2. Found out Gphoto2 will allow the live view to be captured on some camera. Decided to setup a fedora box as a sandbox for testing this. VirtualBox fedora wouldn't recognize the USB devices (I might not have had it setup right, and I'd been looking for an excuse to finish setting up my old Dell with Linux).

11/01/2012
I began doing some testing now that I have a sandbox setup. Gphoto2 successfully recognizes my camera and can control it over USB. I am able to take the capture data and have it send to stdout. With ffmpeg I have figured out how to accept the stream and convert it for ffserver to send over the network.

My command is like this: 
ffserver -f /etc/ffserver.conf | gphoto2 --stdout --capture-movie | ffmpeg -f mjpeg -r 2 -i - http://localhost:81/canon.ffm

This appears to be working and ffserver is reachable over my local network. However, when I go to open the native stream in the browser ( which supports playing mjpeg streams) I get the error: This feed is already being received. This was a simple solution.


My ffserver.conf file looks like this:

Port 80
BindAddress 0.0.0.0
MaxClients 10
MaxBandwidth 50000
NoDaemon
<Feed canon.ffm>
file /tmp/canon.ffm
FileMaxSize 10M
</Feed>
<Stream canon.mjpeg>
Feed canon.ffm
Format mjpeg
VideoSize 640x480
VideoFrameRate 10
VideoBitRate 2000
</Stream>

As you can see, there is no conversion happening other than re-sizing the frame size. These setting provide a stream with the following attributes:

  • fps 13
  • bitrate 233.7kbitz/s


I assume changing my ffserver.conf will allow me to change the output format to be compressed, and allow me to up my frame rate.

11/03/12
Today I finally saw my camera stream over the network in real time. This attests that it could definitely be controlled and viewed a remote device on a wireless network. I'm about to buy a USB WiFi dongle so that I can begin porting this functionality to the Raspberry PI. I'm really excited to be seeing this working so well. I ended up buying a case for the PI for about $10, just to prevent damage from static electricity and dirt. Who knows when this will arrive though. The stream today has some lag, but I doubt it is from a wired network, so I'm going to look into how to optimize the image transfer. Maybe I will add some hooks to allow the client to control the output stream. Currently it has the following specs:

  • MJPEG uncompressed output
  • fps=15
  • q=0
  • bitrate 292.6 kbits/s


I'm going to optimize the output format to be compressed and see if I can get a 720 resolution with decent frame rate.

The stats page from ffserver look like this:


Screen capture of the video streaming

h264 links:
http://www.dexmac.com/index.php/how-to/74-streaming-with-ffserver
http://ffmpeg.org/sample.html
https://lists.libav.org/pipermail/ffserver-user/2011-January/000248.html


USB Dongles Options:
http://elinux.org/RPi_VerifiedPeripherals#Working_USB_Wifi_Adapters
http://www.edimax.co.uk/en/produce_detail.php?pd_id=328&pl1_id=1&pl2_id=44
http://dx.com/p/ultra-mini-nano-usb-2-0-802-11n-150mbps-wifi-wlan-wireless-network-adapter-48166?item=1&Utm_rid=24958662&Utm_source=affiliate
http://thepihut.com/products/usb-wifi-adapter-for-the-raspberry-pi
http://www.buyraspberrypi.com.au/raspberry-pi-802-11bgn-usb-wireless-dongle/ (w/ antenna)

11/16/2012
Been very busy and haven't been able to work on this much. Finally at 11 PM on Friday night, I am making some time (to take a break from AMS suffering).

I determined that my goals from before were somewhat mixed, I wanted to find a good output format and have low latency with very little processing power. To make this simpler I have divided the goals hoping I can eliminate type conversion all together.

My initial thought for achieving fairly low latency and still keeping resolution higher is to try and implement 3 things:
  1. Decent but not incredible frame rate, I think if I can get it to be under 18, that will help and still be decent
  2. Try to interlace the video frames to send less data overall
  3. Implement a loose packet protocol UDP (of course)

Lowering the frame rate actually hasn't added much because ffmpeg wont let me get to more than 17 or so. Also, I note that my bit-rate is incredibly high (3857 kbits/s). My resolution is 640x480.

11/17/2012
Tried to setup the same situation on the PI but ran into issues getting the same stream to get captures and transmitted. Going to install everything from source if I can to see if I am running into issues because of the packages available through the PI repos.

------ REVISITED ------

07/25/2016

After getting the Pi 3, I decided to try this again. This time I plan to take better notes.

* Installed raspbian
* Expanded to use entire sd card
* Setup my wifi by updating wpa_suppliment to connect to my router
* run all updates
* build latest gphoto2 (remove any existing, install deps)
       - $ sudo apt-get purge gphoto2
    - $ sudo apt-get install libpopt-dev libltdl-dev libusb-dev libusb-1.0-0-dev -y
    - Download the tar.gz for libgphoto2 and gphoto2 (i.e. 2.5.10) from sourceforge
       = Once they are downloaded, tar zxvf on each download, cd into each extracted dir and run (libgphoto2 must be done first):
       = $ ./configure --prefix=/usr
       = $ make
       = $ sudo make install
* downloaded https://fattylewis.com/latest-ffmpeg-compiled-for-rasbian/ (bui
* (optional, if desired, sym link the binaries in the ffmpeg download to some place on  your path (i.e. /usr/local/bin)
* Ran into an issue where gphoto2 would not detect my canon 550D on RPi3. Halting for now

Next:
* Solve the gphoto2 issue
* Install HostAPD - https://www.raspberrypi.org/forums/viewtopic.php?f=36&t=138550

jQuiz: A professional quiz plugin for jQuery

jQuiz for jQuery on Github

A little while ago I wrote a quiz component for jQuery. I wanted to make it professional and easy to use. It features: multiple choice, multiple response, fill in the blank, essay, ordered list, and now hot spots! The example that comes with the plugin is written in PHP and demonstrates the AJAX/PHP interface with a SQL database template file.

Tuesday, February 19, 2013

Programming: Calculating Ongoing/Moving Average - Average Over Time

Problem:
I am writing a platform that will track ratings of people over time and display an average star rating per person. It dawned on me that there should be a solution out there, so I went looking. If you're not familiar with calculating an average, you would normally do a mean average like this:

(n1 + n2 + n3 + n4) / n_count

Like this:

(1 + 2 + 3 + 4) / 4     // The 4 as the divisor because we added 4 numbers together.

But this doesn't work very efficiently when your programming an application that will constantly be displaying the user's average, not to mention having to do the query to get all the values and run the calculation every time would mean using a lot more in server resources.

Solution:
Use a weighted average over time (also sometimes called an ongoing average, rolling average, running average, or moving average). You will still need a couple reference points in order to do this, but you will find it will be much simpler on both the calculation and the resource usage. This will only need to be calculated when input numbers affecting the weighted average have changed (been added, removed, or modified).

new_avg = (existing_avg * existing_number_of_ratings + new_rating) / (existing_number_of_ratings + 1)

Now new_avg will be the current weighted average for that user. Just be sure to store the existing number of ratings so you can compute this again later.

Note: Mean average and weighted averages wont always produce the same result.