5 ms·

What TV manufacturers are doing is unacceptable. It introduces strange artifacts and the motion looks very artificial compared to a native 60fps video. So it re

by korm 7y ago

What TV manufacturers are doing is unacceptable. It introduces strange artifacts and the motion looks very artificial compared to a native 60fps video. So it really doesn't solve anything. The argument that it's the filmmakers' fault for using 24fps is flawed. What about animation? What about stop motion? Those may never be produced at a higher framerate. Movies with high VFX budgets will be much more expensive due to the additional frames when rotoscoping and compositing. So if it is inferior and only makes sense for a subset of productions, why is it the default?

mrob 7y ago
In many cases the bulk of the VFX budget is artist wages, not CPU time. If you already have the motion paths set up then it doesn't take much more human effort to render more intermediate frames. You might even be able to reduce the cost per frame by exploiting the smaller differences between frames, e.g. with temporal noise reduction. Hand drawn or hand posed animation is the rare case where increased frame-rate really would take more human effort, but it's often already produced at extra-low framerate (e.g. 12fps or 8fps), with frames duplicated for playback. The same duplication can happen with higher playback framerates. Modern video codecs handle duplicated frames well.
- korm 7y ago
  That's why I specifically referred to rotoscoping and compositing, not rendering. The artist has to edit frame by frame.
- dreamcompiler 7y ago
  And yet many fully computer-generated movies today are converted to 3D in post, rather than being created in 3D in the first place. For CGI movies, doing 3D at creation time gives better results and requires zero human labor (it's just a translation of the camera's matrix), so why isn't it always done?
  - Mathnerd314 7y ago
    CG movies are created in 3D, they use 3D modeling software. There's nothing to "convert" to 3D unless they lose the original models and have to retrace them or something. Which apparently is a quite common occurrence, according to some articles on the 2D to 3D conversion of Toy Story and Shrek. So collateral damage from capitalism is the main explanation for old movies. For new movies, where there is proper data, the 3D camera for each scene still requires more than "zero human labor"; you have to specify the inter-ocular distance, convergence plane, and tweak the scene a little for cinematic effect or 3D weirdness. And since there are thousands of shots, and the work is fairly tedious, it makes little sense to work on it in parallel with 2D during production when shots might just be cut entirely. So you end up with a bias towards 2D or 3D during the creation process. Dreamsworks apparently has a made-for-3D animated movie division (https://animatedviews.com/2007/dreamworks-goes-3d/ https://animatedviews.com/2007/dreamworks-goes-3d/), where presumably they are flattened to 2D in post. Reviews of their movies (https://www.3dor2d.com/reviews/how-train-dragon-hidden-world-3d-3-d-movie-review https://www.3dor2d.com/reviews/how-train-dragon-hidden-world...) unsurprisingly say "great use of 3D". Since that's the goal of the division I'd expect them to continue doing 3D-first movies. In contrast Pixar so far has stuck to a more 2D style. In this interview (https://www.youtube.com/watch?v=HepIGDJK98s https://www.youtube.com/watch?v=HepIGDJK98s) their stereo guy says they do 3D in post. The 3D in their movies is described as "well done but a minor aspect" (https://www.3dor2d.com/reviews/Toy-Story-4-3D-3-D-Movie-Review https://www.3dor2d.com/reviews/Toy-Story-4-3D-3-D-Movie-Revi...). In the interview the guy basically says that he could use more 3D but he's more concerned with keeping the director happy so he doesn't. Pixar is now focusing on originals so maybe one of their next films will be 3D oriented. But they're making tons of money anyway so who knows if the right director will come along.
    - dreamcompiler 7y ago
      > CG movies are created in 3D, they use 3D modeling software. Of course. I should have been more specific; I was talking about stereo conversion. You addressed that. Thanks for your explanation. It's not as labor-free as I had assumed.
amelius 7y ago
Isn't MPEG or any other encoding doing similar things?
- korm 7y ago
  MPEG does motion compensation for compression AFAIK, which is different to motion interpolation, which is what is being discussed.

by korm 7y ago

mrob 7y ago

In many cases the bulk of the VFX budget is artist wages, not CPU time. If you already have the motion paths set up then it doesn't take much more human effort to render more intermediate frames. You might even be able to reduce the cost per frame by exploiting the smaller differences between frames, e.g. with temporal noise reduction. Hand drawn or hand posed animation is the rare case where increased frame-rate really would take more human effort, but it's often already produced at extra-low framerate (e.g. 12fps or 8fps), with frames duplicated for playback. The same duplication can happen with higher playback framerates. Modern video codecs handle duplicated frames well.

korm 7y ago
That's why I specifically referred to rotoscoping and compositing, not rendering. The artist has to edit frame by frame.
dreamcompiler 7y ago
And yet many fully computer-generated movies today are converted to 3D in post, rather than being created in 3D in the first place. For CGI movies, doing 3D at creation time gives better results and requires zero human labor (it's just a translation of the camera's matrix), so why isn't it always done?
- Mathnerd314 7y ago
  CG movies are created in 3D, they use 3D modeling software. There's nothing to "convert" to 3D unless they lose the original models and have to retrace them or something. Which apparently is a quite common occurrence, according to some articles on the 2D to 3D conversion of Toy Story and Shrek. So collateral damage from capitalism is the main explanation for old movies. For new movies, where there is proper data, the 3D camera for each scene still requires more than "zero human labor"; you have to specify the inter-ocular distance, convergence plane, and tweak the scene a little for cinematic effect or 3D weirdness. And since there are thousands of shots, and the work is fairly tedious, it makes little sense to work on it in parallel with 2D during production when shots might just be cut entirely. So you end up with a bias towards 2D or 3D during the creation process. Dreamsworks apparently has a made-for-3D animated movie division (https://animatedviews.com/2007/dreamworks-goes-3d/ https://animatedviews.com/2007/dreamworks-goes-3d/), where presumably they are flattened to 2D in post. Reviews of their movies (https://www.3dor2d.com/reviews/how-train-dragon-hidden-world-3d-3-d-movie-review https://www.3dor2d.com/reviews/how-train-dragon-hidden-world...) unsurprisingly say "great use of 3D". Since that's the goal of the division I'd expect them to continue doing 3D-first movies. In contrast Pixar so far has stuck to a more 2D style. In this interview (https://www.youtube.com/watch?v=HepIGDJK98s https://www.youtube.com/watch?v=HepIGDJK98s) their stereo guy says they do 3D in post. The 3D in their movies is described as "well done but a minor aspect" (https://www.3dor2d.com/reviews/Toy-Story-4-3D-3-D-Movie-Review https://www.3dor2d.com/reviews/Toy-Story-4-3D-3-D-Movie-Revi...). In the interview the guy basically says that he could use more 3D but he's more concerned with keeping the director happy so he doesn't. Pixar is now focusing on originals so maybe one of their next films will be 3D oriented. But they're making tons of money anyway so who knows if the right director will come along.
  - dreamcompiler 7y ago
    > CG movies are created in 3D, they use 3D modeling software. Of course. I should have been more specific; I was talking about stereo conversion. You addressed that. Thanks for your explanation. It's not as labor-free as I had assumed.

amelius 7y ago

Isn't MPEG or any other encoding doing similar things?

korm 7y ago
MPEG does motion compensation for compression AFAIK, which is different to motion interpolation, which is what is being discussed.