Doyle Rockwell
Forum Replies Created
-
Howdy,
Looking at the output, it doesn’t appear that any of the other scene elements move in front of the text objects, so you can just do a series of separate passes for the text and comp them together in FCP (or Motion). It also helps that the text is white, so you can even do a simple comp using a blend mode (such as Add or Screen) and not worry about alpha.
Render out a version without the contact info, with all the quality bells and whistles on, then save off a copy of the project. In the copy, disable everything but the contact data and the object you’re using as your reflective floor. Try dropping some of the more expensive render options, such as float and shadows, if you can get away with it. This leaves you with the identical camera move, but you don’t have to render any of the other objects. Do render passes with each of your contact text versions. Finally, slap each text pass over the clean background and you have your finished spot.
If the render time for the complete scene wasn’t too long, then I’d suggest doing a little scripting (i.e. text find and replace) on your project file, which could easily generate 65 Motion projects, each with its appropriate contact info, ready to be batched through Compressor. In your case, however, your scene is too expensive to render so redundantly, so you need to do some baking.
-
-
Howdy,
I’m not sure why your GPU (or the drivers) are getting overloaded, but that’s what it looks like. Motion’s motion blur is full-scene, i.e. a frame is rendered for every sample, so if you have it set to 8 samples, it renders 8 frames to blend for each finished frame. It sounds like your GPU is wigging out under that load.
If you can decrease the number of samples without compromising the look, you might try doing that. It’s set in the project settings (Cmd+J).
-
Howdy,
As Zak said, by hand is the way to go. Even if you could drive select bezier points with the audio levels, they would be just that: levels. Even really basic mouth animation should have a few phoneme shapes (wide open, narrow open, pursed, tongue under teeth, etc.) and those different sounds don’t directly relate to the amplitude of the voice.
Unless you’re going for something really simplistic like a flashing light (i.e. robot), you’ll get much better (and surprisingly easier) results just doing it by hand. Take a look at the mouth animations for South Park or Yo Gabba Gabba. They’re very basic, effective, and always done by hand in a short amount of time.
-
Howdy,
Quality morpher/warpers, especially spline-based ones, have been scarce for years. Even as recently as a few years ago, I knew people who still maintained old systems that they could run Elastic Reality on.
That being said, CHV makes a couple of FxPlug plug-ins that do mesh-based morphing and warping. As FxPlugs, they run in both Motion and FCP: https://www.chv-plugins.com/cms/FxPlug/Morphing/Morphing.php
-
Howdy,
All Motion project files are in XML format, but they should sport a .motn extension, not .xml. I’m not sure how those files got a .xml appended to them, but try changing it to .motn and open them in Motion (assuming that they really are Motion documents).
You can also open the files up in a text editor (such as Text Edit) and check out the opening tags. Motion documents always begin with:
<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE ozxmlscene>
<ozml version="4.0">In this case, the project file was from version 4 of Motion, as is noted in the
ozmlversion. -
Howdy,
It sounds like the 3D group that you probably have your cube walls in is getting rasterized. This means that it’s forced to comp into the scene in layer order (i.e. the order in the Layers List). It will still respect the camera and be depth-sorted unto itself, but it will get layer-ordered with the rest of the project. This is where knowing the order of operations comes in handy, as well as looking over the “rasterizing” section in the manual, which lists the various things that trigger rasterizing.
Did you happen to drop the group’s opacity, change its blend mode or apply a filter to it? One way to get around the issue would be to put the cube’s contents inside the same 3D group as the cube’s walls, if they aren’t already.
-
Howdy,
What you’re asking to do has 3 solutions:
1. Hand animate the corner-points that move out of frame, because they can’t be tracked when the feature (i.e. object being tracked) is no longer visible. This can work fine, depending on the length of the shot and how prominent (or not) the screen-replacement is.
2. Use the Offset Track option to select a new tracking feature for the point(s) going offscreen. This works best if the feature is moving perpendicular to the camera (i.e. 2D).
3. Use a 3D camera solver, then you can position objects wherever you want and they move or sit still properly.
If your shot is short and the camera move isn’t too 3D (i.e. a lot of parallax shift), then you can use the Offset Track option. If the move is too 3D, then you can hand animate the offscreen trackpoints, but the quality will depend on how much elbow-grease you’re willing to put in.
As for a 3D matchmover adding complexity…they’ve gotten so laughably cheap in the last few years that it’s worthwhile to invest in them. It’s functionality that neither the Studio nor any other product suite (i.e. Adobe) has built-in, so it’s handy to have in your arsenal when it can save you a lot of time.
-
Howdy,
This is the kind of situation where a good 3D track makes the shot a no-brainer, rather than a potentially-tweaky 2D four-corner pin. PFHoe, which is a “lite” version of Pixel Farm’s 3D tracker, is cheap and can export 3D matchmove for Motion (as well as a bunch of other apps).
In PFHoe, you can just tag the position of the TV screen and it will give you a Motion project with a solved camera and a 3D group that is located at the 3D position of the TV screen. Then you just drop your replacement image into the group and you’re pretty much good to go.
-
Howdy,
For what it’s worth, I’ve found dvMatte Pro to be excellent for broadcast (non-film) work. It’s an FxPlug, so it runs very fast and the same version loads in both FCP and Motion. Give the demo a try, found in the ‘Samples’ section of the link above.