Project Tasks

Progress

Weekly progress reports:

Run tests using LBS, DQS, and a mix of the two, and gather visual perceptual results (videos + screenshots) as well as the FPS.
Perform small user study. Each member finds 3 people to show the different results to, they decide which looks better.
Play around with mixing LBS with DQS to find which setting gives the best results (eg. using 60% LBS result with 40% DQS result).
Continue timer and camera work if not quite done.
Clean up and complete documentation, including instructions on how to use, and also list of files what were changed including description of what was changed.

Meet with Nassim to go over dual quaternion skinning. Implement dual quaternion skinning in pinocchio (done), and fix the wierd artifacts happening, which may be the cause of some bugs.
Even if dual quaternion skinning isn't looking great, find a way to blend the results of linear blend skinning with that of dual quaternion skinning.
Take a look through the scripts and programs we found looking around (list available at https://github.com/CMPUT414W15TeamF/CMPUT414-Skinning/wiki/Open-Source-Tools-Available ) and try them out to see if we can also use them for comparison, or learn something from their implementations.
Implement pausing pinocchio and / or showing the time during the animation so we can fairly compare different screenshots of different techniques. (Basically create a visible timer).
Implement setting fixed camera angles to use for fair comparisons of LBS and our own changes.

Get Pinocchio compiling. If unable to do so, emergency meeting for a change of plans.
Find a metric to measure quality other than human perception. Do by Saturday. Ideas: performance, ie. time taken for skinning; mesh comparison using tool Polymeco [Silva et al. 2005] to measure the geometric and normal deviation between each of the meshes generated. Code in this / these metrics(s) (eg. if skinning time is a metric, modify the code to record skinning time and output it).
If the code is able to compile, change the spot in the code we think is the LBS algorithm to see if we were right. Slightly modifying the algorithm should produce messed up results. Take a screenshot of the changed code and the resulting effect.
Decide on one of the algorithm changes from last week, and modifying the code to implement it. Record the results, and write what changes to the algorithm were made.
For each of the four test models, each member skins it, and takes screenshots at specific moments during the animation where candy wrapper artifacts are occurring. These moments should be predefined (maybe code to take a screenshots at specific times if possible?) These will be the base images used for comparison. This should be done by Tuesday, March 9. Each member should be using the model they chose during last week's project tasks.

Use pinocchio to skin a character. Each group member must skin one character and show the TA.
Read over the code and understand what each module does. All members must do this. Then find the module(s) in which the character skinning is done (the code we’ll be modifying).
Each member will choose one test model to be used for testing (4 models in total). Make sure they comply with input needed for pinocchio. Add them to the repo.
Skin these characters using default LBS. Make a short report on findings to later be used as a base for comparison against other techniques. This should include test models chosen, results of skinning the test models, screenshots, problems noticed or possible improvements.
Decide on changes to be made for next set of testing. This could either be modifications to the LBS algorithm, or completely switching it out and using a new algorithm. Make a list of 4 possible experiments to try, accessible here.