Mocha Pro general and roto questions

Hi All,

I have recently purchased Mocha Pro and have been looking at the videos, manual as well as experimenting and I have some basic questions that I can’t seem to figure out and was wondering if anyone can help me out.

My questions below are based on the following scenario involves which involves Rotoscoping and the clip itself is a person dancing with the background which is perfectly still since the camera was placed on a tripod throughout the whole clip. I have read that it is easier if your track the object prior to doing the actual spline for the profile in order to help reduce the amount of keyframes to do the actual spline throughout the clip. I also read that it may be easier to break up the object [the person in my case] into sections.

So lets say I work on a leg. My first layer, call it “left leg track” is a loose spline which the leg stays in throughout the clip, achieved with a few keyframes here and there. After a sucessful track I click on the cog to remove it so I don’t accidently overwrite the track. All good so far.

I then create my spline of the leg, call it “left leg spline” and say I start this at frame one. Lets also say the spline has only 8 control points. After this spline is complete I then link this layer to the “left leg track” layer. And now for my questions:

  • Because I linked the layers, I no longer have the option to track my left leg spline layer. This means that although the spline is tracked the changes to the spline due to motion are not. An example would be the foot turning 90 degrees where the start you see the profile of the foot and the end the toes are facing forward so the profile “disapears” I expected Mocha Pro to be able to track this as well but since the cog is not an option I cannot tell Mocha to do track the spline [on top of the track it is linked to] This means I have to keyframe all of these types of changes. Is this correct? Am I missing something?

  • My next question is how Mocha Pro handles the “Points” that make up the spline. So I have 8 points around the leg and foot. Lets say Pt 1 is at the calf, pt 2 is at the heel, pt 3 is the big toe and pt 4 is at the little toe when I set up the spline in frame 1 as described above. Now lets say 10 seconds down the track I creat ea keyframe which contain the same 8 pts that I will modified in frame one. When I adjust the points let say pt 1 which was the calf I move down to the heel, pt 2 which was the heel I move to the arch of the foot and pts 3 and 4 stay the same. Does Mocha Pro react differently because it sees each pt and follows it throught the clip or are the point just tools for the spline and Mocha only works with the mathmatics of the spline alone?

  • My last question, Since my background is completely stationary, is there a way to remove it completely? I understand if the background were moving as well this could be very complicated but since it is not and it already knows how to remove the person can’t I just remove the background for this specific scenario? Either way, what is the best way to do this?

Sorry so long and I appreciate any feedback to help out. Thank you all in advance.

Jay

Correct. Consider any layer as a pairing of two data types: The track data and the spline control points.
This means that the layer can either use its own tracking data or another layers. It can’t double up and do secondary tracking based on the movement of a parented track.
In your foot scenario you would either have to track the foot separately, or adjust the foot based on the leg track data.
But we have considered adding this. Let us know if it’s the kind of thing you’d like to see in future. :slight_smile:

You can either animate via spline keyframes or control point based keyframes. Working at spline keyframe level means any change you make to a control point is keyed for the rest of the control points. You can set this in the “Keyframe controls” panel from the View menu and see the result of your animation via the Dopesheet.
But yes, all spline animation is based on the control points of the spline. Any transforms that happen to the spline happen to all control points.

The short answer to this is no. The long answer is “It’s complicated”.
Removing the person is generally no problem because we can work out where the person ISN’T and patch it with the background at any point in time.
Removing the background is not so simple, because of edges. Even with a green screen, motion blur and soft edges can be difficult to extract from a moving object. this is even harder with a subject who may have luminance or colour data that blends in with portions of the background. The result means you get a foreground object with what is called “snowy edges” which means the matte for the foreground has edges that look like it’s in a blizzard.

The best way to try this is to get a foreground-free version of the background (either via a standard object remove or by painting a static plate) and do a difference matte of the original footage over the clean plate. You can then roto out the parts of the difference matte that blend incorrectly.

First, thank you so much for not just the quick reply but the clear answers. It is much appreciated. I have a couple of followup and new questions if you could help:

1- I understand both layers and grouping but would like to confirm its effect, if any, on tracking and/or roto. So:
a- In general, other than organization and ease of “batch” effects like turning a whole group on and off by just pressing the eye at the group level are there any other effects which grouping does?
b- Ordering of layers in Mocha has an impact and the closest item in the scene should be the highest in the layers but what about if a spline is Linked to a track layer? Are both layers critical in stack order or only the spline layer? So for example, if all my splines are linked to its respective track layer can I group all the track layers that are linked and place them randomly in a group and place that group all the way at the bottom while I group the corresponding splines into one group but in its correct hierarchy will this work as expected or do the linked pairs need to stay together [and even put in individual groups]?
c- Lets say I have a group with 3 layers. Each of these 3 layers contain a spline and track data [ie. cog wheel on and tracked with keypoints added if needed]. Now at the group level I have an optiojn to use the cog wheel. What effect does tracking the group have, if any, to the individual tracks and keyframes?

I have a mac pro, fcpx, motion and in speaking with sales to confirm Mocha Pro compatability I was told there was no plugins for fcpx or motion but if I purchased the Mocha Pro standalone version all I have to do is export/import so essentially a couple of more steps. I have completed my first “test” roto in Mocha Pro [of the person I referenced above] and tried to export the matte to fcpx. I created the xml but fcpx will not open it. What am I doing wrong and how do I get this into fcpx?

Thanks again for your help.

Jay

At this stage, no. We have plans to expand this beyond just the simple grouping in future versions. Requests are welcome.

Matte Layer order matters regardless of whether the track is linked or not. This means you can keep the tracking layers at the bottom of your layer stack and just turn off the visibility. They do not need to be together, and we actually discourage this as it can cause problems for Remove etc.
Keep in mind that turning off the visibility of a layer does not mean it is ignored in further matte calculations in tracking and Remove.

All the cog wheel at the group level does is control whether processing happens on the layers in the group. This means you can keep the individual cogs on or off in the group, but universally turn them all off by turning the group-level cog off.

The Final Cut shape XML export is specifically for FCP 7. FCPX removed support for this external format so at the moment, you have to rely on rendered clips.
What you need to do is export rendered shapes from the file menu. This will output the layers as a rendered black and white clip. I recommend exporting to a QuickTime movie at the same prores format you use in FCP.
You can then use the Image Mask effect in FCP to drop the rendered matte onto your original plate. You can then use Luminance as your matte value.

Once again, thank you for the quick and concise reply. It is much appreciated. Great work and thank you!

Jay