September 30, 2012 § Leave a comment
This week we had another large group meeting . Farah and I used this opportunity to practice our poster presentation for the Grace Hopper conference. The final version of the printed poster looks great and Farah has made a terrific carrier to get the poster to Baltimore. We are ready to head to Baltimore in a couple of days. We also reviewed the use of the Sign Transcriber and continued to work on creating new signs.
We did not have a chance to discuss our work with the ProOptimizer modifier with Dr. Wolfe. From my experimentation it looks like we can use the ProOptimizer fairly successfully on the hair to reduce polygon count. The ProOptimizer settings allow you to maintain UVs so we would not need to retexture the hair. The biggest potential issue that I see at this point is that the lower poly hair has a much more angular profile than the rest of the model.
The hair before using the ProOptimize modifier
The hair after using the ProOptimize modifier.
However the straight edges in the hair may not be a huge issue in the waist up view that we will primarily be using for signing.
From the experimentation I have done so far with the ProOptimizer on the tongue and teeth model, it looks like it will be harder to get good results. However only a small section of the teeth are visible at a time. When viewed with the rest of the model the lower polygon mesh of the teeth may look better.
September 27, 2012 § Leave a comment
This week Farah and I met with Dr. Wolfe and discussed what we had found out about the polygon count in various parts of our model. It turns out that there are two major types of geometries in Paula. Most of the model is composed of Geometry class objects. These are poly meshes, editable meshes, boxes, spheres, etc. However the eyelashes are made of shape objects specifically spline shapes and lines. Among the geometry objects the largest are the tongue and teeth model, the body, the head, the hands, and the hair. The tongue teeth and the hair both appear to be good candidates for eliminating polygons. The hair is not skinned so we only need to worry about transferring the uvs between the original and revised model. The tongue and teeth model is skinned but it is fairly simple and should not require too much effort to reskin.
We started discussing how we will measure and compare the effects of changes to the geometry on the ability to run in realtime. It looks like the best way will be to compare render times for various versions. A little research found a script that records the time required to render.
Farah and I are looking at the effect the ProOptimizer modifier has on the geometry and runtime when applied to the tongue teeth model and the hair this week.
September 16, 2012 § Leave a comment
This week we had a large group meeting. Farah and I talked briefly about our upcoming presentation at the Grace Hopper conference. We will practice presenting our poster with the group in a couple of weeks before heading to Baltimore for the conference. The main focus of the meeting was an in depth discussion of how to use the Sign Transcriber, our group’s custom 3DS Max interface for building signs. We each tried building a sign with the program. The sign transcriber makes tweaking the positioning very easy, however it is very difficult to get an accurate sense of how the motion actually looks because the playback within max is slow and skips frames. Working with the program showed Farah and I the accuracy and speed a revised model will need.
Farah and I both started working on a MaxScript program that outputs the number of polygons in each part of our model. I created a working first version that gave us some first ideas on where to start reducing the number of polygons. Areas such as the tongue and teeth, the feet and the hair have high numbers of polygons, however they are generally not animated or seen. Over the next week Farah and I are going to revise our scripts and make sure we are not missing parts of the model. We are also working on preparing for the poster presentation.
September 9, 2012 § Leave a comment
I am continuing my work with DePaul University’s American Sign Language Synthesis project this year. I will be working with Farah Thomas and Dr. Rosalee Wolfe on developing a real – time display for our signing avatar. Our current model is too complex for real-time playback within3DS Max our modeling and animating program. To view animation in real-time it is necessary to render the animation. This time consuming process makes the fine tuning of the motion slow and cumbersome. We will adapt our avatar so that a visually acceptable real-time display is possible. Our first step is to analyze our model and figure out what elements are computationally expensive and unnecessary. We will begin by looking at the number of faces in the different sections of our avatar and at potential ways to reduce the polygon count.
At the end of last year we learned that the poster proposal Allison and I submitted to the Grace Hopper Women in Computing conference was accepted. Farah, Dr. Wolfe and I will be traveling to Baltimore for the conference in the beginning of October. Over the summer I worked on the poster design and we are also currently researching printing options for the poster.