After our camera tracking workshop, we were all given the task to camera track the room using 3D Equaliser that we were going to be using for our Group Project.
I started by opeing the footage on 3D equakiser, setting the image input settings such as the gamma level to what we were taught to set it at, and then began tracking the tracking marks.
Tracking the marks sucessfully individually took roughly a few hours, as I made sure each mark kept within the green area of bar when tracked. This meant changing the contrast, saturation level and brightness for different areas in the room, for example some areas have way more shadows within them, and the pattern of the cross and the green need to be emphisised to make tracking more accurate, so it can easily recognise the pattern through movement.
I then calcualated the tracking and viewed the camera movement in 3D.
There were no breaks in the line, and you could see where different parts of the room were. I decided to use result and then bring up the deviation browser to see the average accuracy level.
As you can see the accuracy number was quite high and needed to be brought down to as close to 0.5 as I could. Also, I was unsucessful at tracking the top of the table as it always turned out to be innacurate, these were the two factors I needed to change before exporting this scene.
I then remembered I could apply the lens distortion to the footage which would create more accuracy, I changed the lens to the correct camera option and copied the data of focal length, lens distortion and quartic distortion to be adjusted in the Parameter Adjustments tab. I then clicked adjust and this brought the accuracy to 0.7.
I then went through the waves on the deviation browser to pick out the points that were lowering the average accuracy, andn re-tracked these. I also decided to track the top of the table which was difficult but with changing the image levels such as contrast I was able to track them.
This brought my average to 0.6 which was much better than what I had started with.
I then used the 3D model tool to mesh the points together uaing ‘legacy’ and create a 3D structure of the room, camera tracked.
I then exported the lens data to a Nuke node, and exported the scene as a Maya file, including the 3D models.
However, when I opened the mel script file in Maya, there was only the tracked camera, with no models or footage.
This is something I will have to solve and re-export this file from 3D Equaliser.