April 29th, 2013:
The interpolation code is finally merged with the rest of the software (camera and GUI code) and can process the video clips for the slo-mo replay. Unfortunately, the interpolation processing time on a Linux VM on a laptop is substantially slow. A 1/2x or 1/4x slow down (interpolating 1 or 3 extra frames per existing frame) exceeds our desired max processing time (1 minute) by a lot! The pandaboard seems to take 5-10 minutes to interpolate a 1/2x slow down, which is unacceptable. We are now planning to wirelessly offload the recorded video to Michael's desktop computer (acting as a server) for interpolation and transfer the interpolated video back to the pandaboard. However, this is slow process due to transfer speeds, so we are now only planning to interpolate the center 60 frames around the trigger time (0.5 secs in normal time), which will return a 4 sec (1/2x slow down) or 8 sec (1/2x slow down) "bullet-time" at the center of the video.
GUI updated but still under development!:
Our live camera preview working:
GUI can replay saved video clips and play them at different slo-mo speeds.
April 28th, 2013:
We have added a gallery section in the GUI: we can view recorded videos!
The task involved having to load multiple videos from a directory and display them with thumbnail pictures. We ran into several issues playing the videos in loops as well, but it is now very functional. This code has also been incorporated into the camera/flycapture codebase.
Buttons for viewing the selected video in different slo-mo speeds have also been included (not shown in the image above).
April 27th, 2013:
The GUI code has started to be merged with the camera/flycapture code. The live camera preview now works with the Re.Play GUI, so that camera can be focused before capturing a video
We can also use Re.Play to actually record and replay clips, but the replay is only in normal speed so far.
April 26th, 2013:
Basic GUI has been implemented and can be controlled with the touchscreen.
Main Menu and Preview (Focus) screens. The preview screen does not have a live camera feed yet:
April 25th, 2013:
The interpolation code is more or less complete. However, it is not yet merged with the rest of the software. In addition, there are many parameters in the interpolation code that can still be played around with to optimize the quality of the code's performance.
Live video preview is also enabled
We began the development of the Re.Play GUI with C++ OpenCV. Graphics for the system's GUI were created in photoshop (different screens and buttons). So far, the GUI code can perform simple button clicks for screen transitions. At this point the GUI includes:
-Camera focus screen
-Main menu screen (Buttons: Record, Replay, Settings, Focus)
April 21st, 2013:
Carnival was great.
April 19th, 2013:
Michael is continuing to work on the interpolation code. Though, he will be taking a break for most of carnival due to booth and buggy!
Re.Play is now nicely packaged in a portably-sized enclosure!
From the outside:
Can also open it up if we need to fiddle with any of the hardware. The pandaboard is all wired up inside the case, with switches, a potentiometer for brightness and USB ports connecting out of the case (as shown on image above):
The tripod is yet to be attached!
April 5th, 2013:
We are looking into object detection algorithm to improve our current interpolation algorithm.Still working on transferring all FlyCapture codes to the pandaboard, but running into complications! (problem not exactly known at this point)
In parallel, we are designing the GUI interface for the touchscreen user interaction. We are looking through FlyCapture source codes to figure out how to include a camera control dialogue into the GUI, which will allow the user to configure camera's properties such as exposure and lighting. However, this has proven to be more complicated than anticipated. Hopefully we will hear back from Point Grey's technical staff to have this resolved, as this is an essential part of the GUI...
Also in parallel, we are beginning to design the mount for our device (camera, pandaboard, touchscreen and other necessary electronics) onto the tripod. Below is a rough sketch of a possible design, which is rather pretty simple. The design will be subject to change:
March 29th, 2013:
Developed Test and Quality Assurance plan. Presentation has been uploaded.
Main things that need to be tested so far:
-Interpolation processing speed
Mar 20th, 2013:
All parts have arrived! Screen has been tested and works, camera will be tested soon.
Feb 26th, 2013:
Preparing for the midterm demo! We have basic video interpolation using linear blending. and 2 second video capture to work using FlyCapture SDK. Unfortunately we have ran into some problems running the program on the pandaboard, and have contacted Point Grey for solution. Also waiting for camera....
Feb 20th, 2013:
Preliminary website up. So far we have the pandaboard, and got team repository set up.