4 ms·

3D Face Reconstruction with Position Map Regression Networks

111 points by austin_kodra 8y ago

bwang29 8y ago
Next - select your own cast for the movie you're about to watch.
- John_KZ 8y ago
  Sooner than later you will be able to spontaneously generate films from books or scripts. You will then do some high-level editing by clicking through latent spaces, and voila, you have your own Hollywood-grade film. It's gonna be weird.
  - simcop2387 8y ago
    Then it's just a short step to creating new stories in a known style: LAFORGE: Computer, in the Holmesian style, create a mystery to confound Data with an opponent who has the ability to defeat him. COMPUTER: Define parameters of programme. PULASKI: What does that mean? LAFORGE: Computer wants to know how far to take the game. PULASKI: You mean it's giving you a chance to limit your risk. LAFORGE: No, the parameters will be whatever is necessary in order to accomplish the directive. Create an adversary capable of defeating Data. We're steadily approaching science fiction when it comes to creating things like this.
  - rl3 8y ago
    This is why Disney's casting of real actors for the new Solo film struck me as odd. In Rogue One, ILM already conjured the younger forms of two actors using CG and achieved damn near photorealism in doing so. Furthermore Disney's TRON: Legacy (2010) featured a slightly uncanny but nonetheless impressive young Jeff Bridges. A fully photorealistic and believable young Harrison Ford in about ten to fifteen years doesn't strike me as entirely beyond the pale.
yodon 8y ago
To my human eye the facial landmark placement in this paper looks much better than the 3D reconstructions (which look pretty poor compared to the facial shapes a human eye or human artist would interpret from the photos). The authors argue that their approach isn’t limited by the explicit finite number of facial basis functions problem encountered in most prior 3D reconstruction work, but they seem to neglect that because of the finite size of the ground truth 3D scan dataset their approach is still limited in exactly the same way, it’s just limited implicitly rather than explicitly. If you don’t have enough ground truth scan data for black or Asian or Slavic or etc. faces, the network can’t learn to represent them correctly and you end up with the same low-dimensional results as the prior work, you’ve just hidden the low dimensionality from view by burying it in your neural layers rather than making it explicit in a set of easily identifiable basis functions. Sure, if these authors had a larger ground truth dataset they could do better, but so could the explicit basis function methods. Again, the facial landmark detection looks great, my concerns are purely with the inaccuracy of the 3D model reconstructions.

111 points by austin_kodra 8y ago

bwang29 8y ago

Next - select your own cast for the movie you're about to watch.

John_KZ 8y ago
Sooner than later you will be able to spontaneously generate films from books or scripts. You will then do some high-level editing by clicking through latent spaces, and voila, you have your own Hollywood-grade film. It's gonna be weird.
- simcop2387 8y ago
  Then it's just a short step to creating new stories in a known style: LAFORGE: Computer, in the Holmesian style, create a mystery to confound Data with an opponent who has the ability to defeat him. COMPUTER: Define parameters of programme. PULASKI: What does that mean? LAFORGE: Computer wants to know how far to take the game. PULASKI: You mean it's giving you a chance to limit your risk. LAFORGE: No, the parameters will be whatever is necessary in order to accomplish the directive. Create an adversary capable of defeating Data. We're steadily approaching science fiction when it comes to creating things like this.
- rl3 8y ago
  This is why Disney's casting of real actors for the new Solo film struck me as odd. In Rogue One, ILM already conjured the younger forms of two actors using CG and achieved damn near photorealism in doing so. Furthermore Disney's TRON: Legacy (2010) featured a slightly uncanny but nonetheless impressive young Jeff Bridges. A fully photorealistic and believable young Harrison Ford in about ten to fifteen years doesn't strike me as entirely beyond the pale.

yodon 8y ago

To my human eye the facial landmark placement in this paper looks much better than the 3D reconstructions (which look pretty poor compared to the facial shapes a human eye or human artist would interpret from the photos). The authors argue that their approach isn’t limited by the explicit finite number of facial basis functions problem encountered in most prior 3D reconstruction work, but they seem to neglect that because of the finite size of the ground truth 3D scan dataset their approach is still limited in exactly the same way, it’s just limited implicitly rather than explicitly. If you don’t have enough ground truth scan data for black or Asian or Slavic or etc. faces, the network can’t learn to represent them correctly and you end up with the same low-dimensional results as the prior work, you’ve just hidden the low dimensionality from view by burying it in your neural layers rather than making it explicit in a set of easily identifiable basis functions. Sure, if these authors had a larger ground truth dataset they could do better, but so could the explicit basis function methods. Again, the facial landmark detection looks great, my concerns are purely with the inaccuracy of the 3D model reconstructions.