It has been weeks of experimenting and searching for software and solutions for cheap motion capture.
Motion capture was necessary to avoid falling into the uncanny valley when using 3d scanned characters that I wanted for my project. However the quality and detail reqirements were far from cheap.
First off: Object tracking with Blender.
This required at least 8 visible trackers for each moving object. That means 8 markers for each head, lower arm, upper arm, hand and so forth. A tremendous task of making a physical rig with a useless result as the actor would not be able to move freely.
For facial tracking, Blender is not streamlined at all, and although tracking points were easy it was also only 2d and would not produce good enough results.
Then I looked into everything from how they made Benjamin Button http://www.youtube.com/watch?v=52JqQkx_VDc to 4d capture like from L.A.Noire http://www.youtube.com/watch?v=aL9wsEFohTw.
Also cheap soultions like IPI Soft http://ipisoft.com/ for Kinect and so on.
The last thing I found was a rather easy going software called Mycap that could give me just about everything I needed with just two cameras. A HUGE drawback here is that I could not find a tool for reparing broken tracks. Sometimes they moved too fast for the cameras, or got mixed up in each other, and if it was Blender I could easily reposition them, but not here. The software requires two cameras set up in stereo to capture depth, and would automatically look for markers. Needless to say, markers would pop up everywhere and be removed manually before starting the automatic tracking procedure.
Another thing was that I had a pretty new web cam, but need an additional one. I bought the same brand and type of camera, but they were very different. Both played off recored videos at 30fps, but one of them played the videos at almost double the speed than the other. Also the colors were different and Logitec did not support any kind of color correction. It was useless to rematch the speed of the cameras as the videos would still not match.
When I get two cameras recording at the same quailty with descent framerate and minimal blurring, I will try again, but now I will go back to key frame animation and regulate my project for a more hand drawn approach.
Motion capture was necessary to avoid falling into the uncanny valley when using 3d scanned characters that I wanted for my project. However the quality and detail reqirements were far from cheap.
Testing maximum contrast from Blacklight and UV paint :)
First off: Object tracking with Blender.
This required at least 8 visible trackers for each moving object. That means 8 markers for each head, lower arm, upper arm, hand and so forth. A tremendous task of making a physical rig with a useless result as the actor would not be able to move freely.
For facial tracking, Blender is not streamlined at all, and although tracking points were easy it was also only 2d and would not produce good enough results.
Then I looked into everything from how they made Benjamin Button http://www.youtube.com/watch?v=52JqQkx_VDc to 4d capture like from L.A.Noire http://www.youtube.com/watch?v=aL9wsEFohTw.
Also cheap soultions like IPI Soft http://ipisoft.com/ for Kinect and so on.
The last thing I found was a rather easy going software called Mycap that could give me just about everything I needed with just two cameras. A HUGE drawback here is that I could not find a tool for reparing broken tracks. Sometimes they moved too fast for the cameras, or got mixed up in each other, and if it was Blender I could easily reposition them, but not here. The software requires two cameras set up in stereo to capture depth, and would automatically look for markers. Needless to say, markers would pop up everywhere and be removed manually before starting the automatic tracking procedure.
Another thing was that I had a pretty new web cam, but need an additional one. I bought the same brand and type of camera, but they were very different. Both played off recored videos at 30fps, but one of them played the videos at almost double the speed than the other. Also the colors were different and Logitec did not support any kind of color correction. It was useless to rematch the speed of the cameras as the videos would still not match.
When I get two cameras recording at the same quailty with descent framerate and minimal blurring, I will try again, but now I will go back to key frame animation and regulate my project for a more hand drawn approach.
