Using the footage that Edward recorded, I imported it into PFTrack,
With Edwards Footage Imported into PFtrack, I added an Image manipulation node and adjusted the contrast,brightness,saturation and gamma of the footage to make it a bit brighter and a little sharper in an attempt to make the footage easier to track.
Footage before: Image Manipulator Node
Footage After: Image Manipulator Node
After the image manipulator was added, I applied an Undistort node in an attempt to remove and blur or unwanted noise from the footage
I then started to track the footage, to begin, I applied an auto track node to the footage as a basic foundation for the tracking footage.
With each tracker, I edited its deformation attributes to pick up on Rotate,Scale and skewing happening within the footage (considering there was a bit of all 3 happening within the footage) hoping to obtain a better track and a failure threshold of 0.700.
Auto tracked footage
After which I overlayed a User Track Node in order to obtain a more accurate track.
User tracked Footage
After the footage was tracked I applied a Camera Solver Node in order for PFtrack to determine what would the best camera animation based of the footage.
And then used a orient scene node to determine the axis orientation and scale of our scene.
In order to make sure the orient and perspective of my track was correct I used the test object Node
Test objects in real time
Finally I exported our tracked PFTrack scene as a Ascii file to continue working with in in Autodesk’s Maya.
With My tracked footage imported to Maya implemented my 3D object into the maya scene and adjusted the tracked camera settings in its attribute editor.
I didn’t alter the camera’s focal length as it keyed and changed the Near Clip value to 0.001 and the Far Clip value to 1000000.
When adjustments to the camera had been made and the 3D object matched with the tracked scene I started to implement lighting, consisting of a few directional lights, since the rubrics cube had a MIA preset material allied to it, the values were quite important. I ended up going for a lighting value of 0.200, this made the material visible with a render but also didn’t produce intense shadows.
After which I assigned my 3D object to it’s own layer and called it beauty
The image plane (which I would use to render out shadow) onto it’s own layer called shadow.
And a layer consisting of all the objects within the scene to obtain an occlusion pass, with a sample rate of about 300.
Thinking I was finished, I rendered out the layers and composited them together however I soon realised that the tracked footage was fine but the shadows were way off.
So, returning to maya I altered some of the lights attributes and positions and this time rendered out single frames and composited them together within Photoshop to ensure everything was perfect and that I was happy with the end result before I hit render.
In this test I was trying working mainly with the shadow layer, but I thought the shadow was too dim for the scene and also strangely sharp around the edges, which confused me.
I didn’t like the results I was getting with the shadows and started looking into alternatives like using the Occlusion layer as my shadow, I preferred It must more to the shadows because of the way it naturally diffuses both on and off the cube. But I felt like it was too much for the scene.
Playing about with the intensity of some occlusion pass attributes. I found a result I was much happier with and created the illusion of shadow I was looking for.
I then rendered out the beauty layer and the Occlusion Layer Separately.
With both my Beauty layer and Occlusion layer rendered out it was time to composite everything together, in this case I used after effects.
As I learned from the tutorial mention previously: (see blog post here)
We can ensure the frame rate of our footage and rendered sequences matches up perfectly by interpreting the footage and applying the same frame rate to all of them. In my case I set everything to 24 frames per second.
With both my footage and rendered sequences interpreted to match I then started the process of compositing everything together within the layer editor.
As seen above I applied a multiply layer effect onto my Occlusion sequences this layer effect overlayed the alpha values of the occlusion pass layers to create the soft shadowed effect seen below.