Advanced, Experimental VFX Animation and Techniques.
Blog Week #14 (May 4), Tracking Component of Term Project
Blog Week #14 (May 4), Tracking Component of Term Project
Tracking: Taking a Step Back
Object tracking has a long history that has been described by Michael Seymour (2004) as being derived from efforts within the military for tracking of missiles. As further summarized by Seymour (in his blog The Art of Tracking, Part 1: History of Tracking), the initial application within the VFX industry dates back to 1985, efforts from Lewis and Brigham at the NYIT, where a fast fourier transform technique was implemented to provide a track for movement of the sun across a clip in the National Geographic "rising sun" commercials. From then onward, a long history of many developments within the VFX tracking industry has resulted in highly automated and sophisticated techniques that have further found application within the AR/VR field (Loraas 2018).
As a complement to the development of the mathematical and engineering tracking methodologies applied to different ends, many VFX artists have secondarily enjoyed the use of tracking technology from a practical software-user perspective. This second hand borrowing is generally associated with enthusiasm for the practical relevance of the application, and then understandably, less enthusiasm for acknowledging some of the most salient conceptual nuances and maintaining correct use of the associated terminology. In my opinion, this phenomena can lend itself to problems when there exists a range of similar applications for overlapping, yet different techniques. That is, indiscriminate use of the terminology can readily become associated with unclear application of the techniques, -and this issue may be of greatest relevance for the student. It's my impression that many things are lumped under the common umbrella of 'tracking' and 'match move'. Persons extremely familiar with all the applications and the underlying techniques are able to navigate thru murky waters, but students and recent graduates may apply techniques incorrectly or describe their applications in an incorrect manner (which can be problematic when presenting one's efforts to more experienced artists). The following paragraphs address the terminology lumped under the common umbrella of 'tracking and match move'. Because our term w Aldo didn't address the simplest tracking, i.e. one point tracking, the following investigation of the language will be conducted alongside a review of one-point tracking from the first term.
Match Move: Not Always Camera Tracking
One of the ways in which this issue is perhaps most apparent is in regards to the term 'match move'. As defined in his text 'Matchmove: The Invisible Art of Camera Tracking' (2013a), Tim Dobbert provides the definition: "match move is the process of matching CG elements into live action footage". He then further details 6 core components to the process of match move, the main aspect of which is the creation of a virtual camera. It is notable that other authors have proposed that creating a virtual camera is the core feature of the match move process (Barber et al, 2016).
As detailed in the IMG below, the Foundry (creators of the Nuke compositing program), appear to embrace the more general or encompassing use of the term. The following IMG summarizes how one, two, three, and four point tracking each serially implements additional features for the task of following an object in a scene (i.e. X-Y position, then position and rotation, and finally, position, rotation and scale). Subsequently, the further PS edited images highlight how simple object tracking does not incorporate a virtual camera (Dobbert 2013b).
Object tracking has a long history that has been described by Michael Seymour (2004) as being derived from efforts within the military for tracking of missiles. As further summarized by Seymour (in his blog The Art of Tracking, Part 1: History of Tracking), the initial application within the VFX industry dates back to 1985, efforts from Lewis and Brigham at the NYIT, where a fast fourier transform technique was implemented to provide a track for movement of the sun across a clip in the National Geographic "rising sun" commercials. From then onward, a long history of many developments within the VFX tracking industry has resulted in highly automated and sophisticated techniques that have further found application within the AR/VR field (Loraas 2018).
As a complement to the development of the mathematical and engineering tracking methodologies applied to different ends, many VFX artists have secondarily enjoyed the use of tracking technology from a practical software-user perspective. This second hand borrowing is generally associated with enthusiasm for the practical relevance of the application, and then understandably, less enthusiasm for acknowledging some of the most salient conceptual nuances and maintaining correct use of the associated terminology. In my opinion, this phenomena can lend itself to problems when there exists a range of similar applications for overlapping, yet different techniques. That is, indiscriminate use of the terminology can readily become associated with unclear application of the techniques, -and this issue may be of greatest relevance for the student. It's my impression that many things are lumped under the common umbrella of 'tracking' and 'match move'. Persons extremely familiar with all the applications and the underlying techniques are able to navigate thru murky waters, but students and recent graduates may apply techniques incorrectly or describe their applications in an incorrect manner (which can be problematic when presenting one's efforts to more experienced artists). The following paragraphs address the terminology lumped under the common umbrella of 'tracking and match move'. Because our term w Aldo didn't address the simplest tracking, i.e. one point tracking, the following investigation of the language will be conducted alongside a review of one-point tracking from the first term.
Match Move: Not Always Camera Tracking
One of the ways in which this issue is perhaps most apparent is in regards to the term 'match move'. As defined in his text 'Matchmove: The Invisible Art of Camera Tracking' (2013a), Tim Dobbert provides the definition: "match move is the process of matching CG elements into live action footage". He then further details 6 core components to the process of match move, the main aspect of which is the creation of a virtual camera. It is notable that other authors have proposed that creating a virtual camera is the core feature of the match move process (Barber et al, 2016).
As detailed in the IMG below, the Foundry (creators of the Nuke compositing program), appear to embrace the more general or encompassing use of the term. The following IMG summarizes how one, two, three, and four point tracking each serially implements additional features for the task of following an object in a scene (i.e. X-Y position, then position and rotation, and finally, position, rotation and scale). Subsequently, the further PS edited images highlight how simple object tracking does not incorporate a virtual camera (Dobbert 2013b).
|
One Point Tracking w TRANSFORM Match Move
One point tracking is distinguished by its simplicity of tracking merely one point in the original footage. As in the frames on the right and below, the tracker is simply targeting the back aspect of the boat (the section that might carry cargo has been targeted to receive a checkerboard). Because a CG asset will be imported into the scene, it is still generally correct to refer to this as a match move process, i.e. "match move is the process of matching CG elements into live action footage" (Dobbert, 2013a). Using the phrase 'match move' in this very general manner however, is somewhat confusing, i.e. it violates a more targeted use of the phrase to involve the production of a virtual camera and then matching the movement of that camera to the real world camera (Barber et al, 2016). While possibly beyond the scope of a research blog for VFX, it is briefly noteworthy that equivalent conflicts addressing varying implementations of language have long been the topic of struggles within language focused philosophical schools, e.g. logical atomism (Bergman 1957, Holbert 1983). |
|
Transform Match Move: Applied Skills
