Motion Tracking & Object Tracking inside Cinema 4D: Limitations - Troubleshooting Volume 1

Share this video
  • Duration: 12:15
  • Views: 3625
  • Made with Release: 18
  • Works with Release: 18 and greater

Trimming, is a way to tell the Tracking algorithm to ignore a tracker for a specific number of frames, rather than risking the tracker "picking up" a random pattern by mistake.

Let's take a look at a some of the motion tracker's idiosyncrasies you need to be aware of, to save you from unwanted surprises. Some of the following have already been covered in other videos, but it's good to have a "centralized repository of annoying things", as I like to call it.



Let's take a look at some of the motion tracker's idiosyncrasies you need to be aware of to save you from unwanted surprises. Some of the following have already been covered in other videos, but it's good to have a centralized repository of annoying things as I like to call it. I'll begin with something quite important. So here we have a final solved scene for the camera and the object. And you can see we have our user features and the auto features and everything else we need. First of all, I'm going to go to the motion tracker and make sure that the reconstruction lock solve data is on. And in the object tracker, lock solve data is on. And you can see we have the user features and when you select either of these we can actually see the automatic tracks or the manual tracks. Now what I'm going to do, is go to the motion track of footage, and change the re-sampling. So from 50, I'm going to make it 56.25 and you can see that although the actual features, the nulls are retained because these are physical objects, all the trackers manual and automatic have disappeared. Any change you do that has to do with the footage regardless of if you're reloading the same footage again, this will force the tracker to delete all the automatic and manual trackers. And in the case where you don't have this or this enabled, doing the same thing will actually get rid of the features as well, which means you need to do your tracking again. Anyway, you can always undo but keep that in mind. Let's continue with the luminance problem as I call it. And this is that our tracker tracks the luminance value. So in order to show this, I'm going to add a hue saturation, and I'm going to take the saturation all the way down. So basically, this is what our tracker sees when it's trying to track the features. And as you can see although the white ones have a quite good contrast with their background, the green ones and the red ones don't. So you need to be aware of this if you start wondering why aren't my red trackers tracking nicely against the gray backgrounds and so forth. So it may be essential depending on the color of your footage and the choices you've made to shoot your footage or if the footage was given to you, you don't make that choice. You may need to go and change certain colors so you can add some contrast, brightness or whatnot just to enhance the contrast between the features and the background. Now in some cases, you may choose to go and select a particular channel to work with. So if I get rid of this, delete, and I deleted the wrong thing, but that's not a problem. I'm going to go to my channels. And sometimes you can see that one of the three channels contains good information. So the red channel has good information for everything including the red and the green has very good information except for the red. You can go to your composting program and create one composite black and white image that has these red features overlaid as white and the green ones and so forth so that they all look somewhat like this. So it's up to you, but knowing that this is the way it works will allow you to make better decisions. So we have this particular manual track and you can see in the graph view that we have some areas with high error, but that doesn't mean it's not a good track. And that's because I've actually placed most of these by hand. And you can see that the pattern size is a bit over-sized. I did this on purpose. This is something you need to be aware of. If you have already tracked a tracker with manual keyframes and automatic paths, what will happen if you change either the pattern size or the search size? If I release this, take a look. It's processing. It's not only processing the frames on each side of this particular key frame, it's actually going to process the whole of your track. So there's a good chance that certain areas which were good up to a certain point now are not going to be good. And there you go. The rule is that because there is no way... Let me undo lock down the actual 2D track, unfortunately. If you change either the search size or the pattern size, then the whole tracking for that particular feature will be recalculated based on this. So make a point on choosing your pattern size and your search size from the beginning. Otherwise, you will have to go back and see what has changed and go and fix it. The actual manual keyframes don't change. This only applies to any of the frames that have been calculated based on a pattern and a search area. In this scene, we have all our manual tracks visible in our tracking graph view. And you can see that the way it's sorted, is from where the features starts, and where it ends. So you can see that this one that starts from a later time, it actually is placed up on top. And this is so that we can have a nice spatial arrangement of our features. Now with manual trackers, this shouldn't be a case, but unfortunately, it is. And this is what can happen sometimes. Let's take this particular feature here and let's go to frame 20 and just add a keyframe. Automatically you saw it switch from up there down to here. So sometimes you may be zoomed in. So let me undo and let me zoom in here. And I'm focusing on this particular one. And I just do this, and I move it. There you go. And you can see that it will change position and very frequently it may disappear from your view. So be careful with that, because it will jump depending on where it starts and where it ends. You may be surprised sometimes if it disappears from your graph view. Just make sure you press H or something like that and make sure you can see where this brown-orangey line is. And that means it's your selected tracker. Another thing that bugs me sometimes about the lock view and tracks is that if I have this track selected and you can see it's locked properly. I can move it around, and it stays locked and that's fantastic. If I move to another tracker, just click here and scrub, you will see that it stays locked on the previous one. So this is a bit deceiving sometimes because you can see the lock is activated and you can see a tracker selected. You would assume that it's activated because of this. Anyway, the solution is quite simple. You just click on this one more time. If you click a second time, it will just deactivate and yeah, just click one more time and it's active. So every time you change tracker, just click on this once, and it will lock the specific tracker. Deleting trackers sometimes may cause some odd things. So for example, if I go here and select a tracker and press backspace, then it deletes that particular tracker. So everything is fantastic. Okay. Let me undo. Now if I select this, which means I'm putting my focus in the object manager, then select a tracker over here. And then press backspace, it will delete everything. And when I undo, it will actually take me to the motion tracker, and I need to activate the camera to go back where I was. So the safest way to do it is, if you're generally selecting things in a viewport, you can use your backspace to delete them. But if not and you want to activate and select these, just go and use the delete user tracks over here so you have less headaches. It's very important to remember that when you create a new user track, and you adjust whatever you adjust, whatever of these three buttons you press, that is going to be the direction that is going to track regardless of which one of the buttons you press afterwards. So if this is off, and I press track forwards. It will track forwards and let's bring our motion tracker view. Dock it here, and let's go and select this one. You can see it up here. It only tracked forwards. Excellent. So now if I go here and I say, okay, I want to go here. I want to track backwards. You can see that it doesn't track backwards. If I say insert get track position just to reset it and track forwards, you'll see that it will track forwards. So always remember my proposal is always use bi-directional and for the majority of cases, auto-update tracks is what you need to have selected. I've prepared a very specific scene here. Now although I did track this object successfully, you can see the particular object has keyframes from the beginning to the end. And this is what I used to create my fantastic animation. What I've done, I've changed the way the manual tracks are calculated, and I've deleted the frames here in the beginning. I only have three manual keyframes here and four here. So I know by definition that from this point onwards, and from this point and before, there is no way for the object tracker to calculate the position of the object. So let's see what happens if I tell it to run 3D solver for object in this case. So it's going to run the solver. And what you will see, if you look down here is that it only created keyframes for the range where it feels that it has some information. We already know that we can go to the motion tracker footage and click this button to create a background object. And this is going to create a background object, and it's going to create a material, where the image sequence is loaded in the color channel and is set to animation and here are a number of frames. Now there's a bit of a limitation here. If for any reason, your footage doesn't start from zero, zero, zero, zero. And it starts from one or two or something like that, then creating the background object will cause some problems with the animation. So my advice...because I don't know why it happens and it's a bit too complicated anyway to understand this. My advice is that if you're going to use the create background object, just make sure that your footage, if it's an image sequence starts from zero, zero. If it's a movie, that doesn't make a difference because the frames are allocated automatically. But if you have an image sequence, make sure it starts from zero, zero. There are tools for both Apples and Windows machines where you can renumber your sequences. And I think it's just a two-minute thing you need to do to make sure that everything works perfectly. Another thing I need to remind you of is... Let me delete this background object and delete this. If you have the motion tracker selected and you have zoomed, let's say over here, and you say create background object. The background object will be created fine here, but the texture tag applied to the background will have all these offsets. So if I delete all these and render, you'll see that I get a render which is similar to where I had zoomed in. And of course, this is not going to match your 3D view. So make sure before you create your background object, that your motion tracker is set in full footage. And then create your background object.
Resume Auto-Scroll?