Lecture 17: Intro to Animation, Kinematics, Motion Capture (50)
kevkang
With the previous slide listing out the pros/cons, which motion capture method is the most expensive? Are some methods now cheaper because we have more processing power?
jeromylui
I found this link to be useful in going in detail on Optical motion capture and how it works: http://physbam.stanford.edu/cs448x/old/Optical_Motion_Capture_Guide.html
ellenluo
Couldn't the issue of occlusion for optical motion capture be fixed by having cameras in all perspectives? (from the front, back and sides, etc.) The footage could then be transformed as necessary.
knguyen0811
Inspired by the first comment, I thought I'd look into the pros/cons of the other types of motion capture:
Mechanical motion capture experiences no interference from light or magnetic fields (which may be the case with the other two), but the motion is less realistic due to the restrictiveness of the equipment.
Magnetic motion capture can be used in smaller spaces and is relatively cheaper than optical motion capture; however, it can experience noise due to magnetic fields and can be difficult to move.
In lecture professor mentioned that one con of the first set of equipment is that some markers might be occluded sometimes. Instead of taking a movie of the scene and extract the positions of markers in space, why not make those markers report their own position sequence? There should be more efficient ways to do locating other than optical method imo.
tristanburke
I am a huge marvel fan, and watching behind the scenes footage this week, I discovered that some motion capture equipment (particularly body motion), can render real-time CGI of a particular character. For Thanos for example, they used these real-time renders (of basic quality), so that they Russo brothers could direct the scene with the basic render in mind. I thought this was interesting breakthrough, as most CGI is usually done after the shooting.
AronisGod
Notice how the images look progressively more vintage, thus saying something about the associated technology.
rpottle
To add to the comment about the motion capture used for Thanos, machine learning was also heavily used to save time by the Marvel VFX team both for the real-time renders as well as full the final images. In fact, they think they would not have been able to capture the performance with the level of detail that they did without these advancements. https://variety.com/2019/film/awards/machine-learning-saves-avengers-vfx-artists-time-1203140163/
hershg
Modern blockbuster films like Marvel/Star Wars/DC/etc... extensively use motion capture - you can find tons of footage of images looking like those to the leftmost. These films are generally praised for their VFX/CGI, and the motion capture is a major part of this (several superhero character actors almost never wear an actual suit - they spend all their shooting time in a motion capture suit! Hulk is a good example of this).
A pioneer in this were the Lord Of the Rings movies in the early 2000's with their motion capture for the character Gollum (portrayed by esteemed motion capture artist Andy Serkis)
With the previous slide listing out the pros/cons, which motion capture method is the most expensive? Are some methods now cheaper because we have more processing power?
I found this link to be useful in going in detail on Optical motion capture and how it works: http://physbam.stanford.edu/cs448x/old/Optical_Motion_Capture_Guide.html
Couldn't the issue of occlusion for optical motion capture be fixed by having cameras in all perspectives? (from the front, back and sides, etc.) The footage could then be transformed as necessary.
Inspired by the first comment, I thought I'd look into the pros/cons of the other types of motion capture:
See this link for further info.
In lecture professor mentioned that one con of the first set of equipment is that some markers might be occluded sometimes. Instead of taking a movie of the scene and extract the positions of markers in space, why not make those markers report their own position sequence? There should be more efficient ways to do locating other than optical method imo.
I am a huge marvel fan, and watching behind the scenes footage this week, I discovered that some motion capture equipment (particularly body motion), can render real-time CGI of a particular character. For Thanos for example, they used these real-time renders (of basic quality), so that they Russo brothers could direct the scene with the basic render in mind. I thought this was interesting breakthrough, as most CGI is usually done after the shooting.
Notice how the images look progressively more vintage, thus saying something about the associated technology.
To add to the comment about the motion capture used for Thanos, machine learning was also heavily used to save time by the Marvel VFX team both for the real-time renders as well as full the final images. In fact, they think they would not have been able to capture the performance with the level of detail that they did without these advancements. https://variety.com/2019/film/awards/machine-learning-saves-avengers-vfx-artists-time-1203140163/
Modern blockbuster films like Marvel/Star Wars/DC/etc... extensively use motion capture - you can find tons of footage of images looking like those to the leftmost. These films are generally praised for their VFX/CGI, and the motion capture is a major part of this (several superhero character actors almost never wear an actual suit - they spend all their shooting time in a motion capture suit! Hulk is a good example of this).
A pioneer in this were the Lord Of the Rings movies in the early 2000's with their motion capture for the character Gollum (portrayed by esteemed motion capture artist Andy Serkis)