Matrix frozen in time effect, can Mocha Pro help do this?

I’m a youtuber. I review gadgets. I’m new to vfx, and want to add more of them to my youtube channel.

I am currently reviewing a bug gun, the ‘Bug A Salt’ and trying to create two special effects. The first is this one:
Tutorial: Bullet Time - YouTube (Frozen in Time).

I am using Da Vinci Resolve and Mocha pro (MP) plugin trial on a mac.

I have a motion control camera and have filmed two scenes. One with me in it, and one with just the background.

I have two layers in resolve. Top layer is me. Bottom layer is the background without me.

Here is the clip of me. BorisFX forum help - YouTube

Clearly I need to isolate/mask me (or the workbench and supports I am resting on)

I have watched quite a few of the BorisFX training videos. My understanding is that the tracking plane would be me (nothing else co planar). I needed quite a precise mask because the light was slightly different on each layer, which makes the border quite obvious unless it is tight in on me.

I was hoping the new resolve (DR) magic mask would do it, but no.

So then I tried to manually rotoscope, which I found difficult in DR. The gun is a complex shape and the tweening of the mask was visible.

My question is whether, in a situation like this, where the subject is static, and the camera moves around, whether MP is able to track the changing shape of the mask as the perspective changes, or whether the ONLY way to do this is to manually rotoscope, and given the complexity, It would be better if I try and reshoot with more precisely identical lighting, so my mask can be much more approximate (and feathered).

Does that make sense?

For a bonus point, am I right in saying that in a situation like this, with camera movement only, the best thing is a planar track with skew, and that the best place to track would be some parts of me, like my leg and arm.

Thanks in advance for any help - I’ve been trying to crack this for days!

Hello Arlo!

Yes, Mocha can definitely help with that!

Absolutely right - when dealing with human bodies, you should track every joint separately. Your shapes should look like a paper doll.

In general, as your camera moves around an object, at some point it starts to see parts of the object that weren’t visible at the beginning, and the other ones may become hidden or obscured trough time.
Because Mocha is a planar tracker that looks for groups of pixels moving in the same direction - your shape will follow the overall surface movement, but not the edges of the object. Meaning, you will have to do some adjustments to the shape.

However, the footage you provided is a very simple task for Mocha. Because the speed of the camera is consistent, and the movement is linear, you will only need to make 2-3 keyframes to do your roto:

  • first at the beginning
  • another one when camera stops
  • and probably one more somewhere in the middle of the shot (depending on the shape).

This shot has some perspective changes, but they are so slight, that trying to track it with Shear+Perspective or Shear only wont make a huge difference. I was able to track that with Shear only, and just made slight adjustments to the shapes. Please check the video I attached below

Hope I was able so answer your questions! Let me know if something isn’t clear

1 Like

Wow, @elizabeth.postol, that’s so helpful, thank you. Having watched some of the training videos, I wouldnt have thought to break it (me) into so many body parts! I thought I would be one plane.

I’ll give this a go on one body part to start with, and if I can make that work, then I think you’ve got me off to the races.

This is what I think I am supposed to be doing (based on what I have watched of the training videos).

a) x-spline an area within the shape I want to track (ie my foot)
b) Track it back and forth (because it is not a nice rectangle, it is not really clear to me whether it has tracked correctly, except that it seems to follow my foot ok). I thought I had to do a camera solve on each tracked plane (body part), but it wont do that.
c) Draw more precise shape around foot at start point.
d) Move to the end and correct the end point.
e) Look in the middle and if that needs correcting, then correct that too.

I’m finding its a huge learning curve with this stuff - so thanks again for your help, much appreciated.

2 Likes

Well, it was all going so well with the foot that I tried the leg. That too seemed to work, but once I had created the leg tracking layer, then the leg shape, I went to look back at the foot and now it’s got a life of its own. Not tracking. Or its tracking something else. No idea whats going on. Frustrating. Tried to re-track it, but it just keeps terminating the track (weird because it was fine first time). Going to take a break before I throw the laptop out of the window.

Are you drawing your shapes on the separate layers? Can you provide a screen recording? Then i can say what’s going wrong

I was drawing my shapes on separate layers, yes.

I think we are in different time zones! But I’ve made more headway today, and your help really got me over a hurdle, so thank you very much for that.

Reading instructions, I thought it was necessary to define the tracking area and the shape / mask separately.

But it didn’t look like you did that, so I’ve tried just tracking the shape, and its working really well. Much better than I expected, and a much better and less time consuming result than I was getting with DRs key framing tools

So I’m persuaded that Mocha Pro is worth the money. Just going to wait for the Black Monday sale!

Thank you again for your help.

2 Likes

Hi Arlo,
Glad Elizabeth has been able to help you dig into Mocha!
I saw your channel and thought maybe you’d like to become an affiliate. I will email you directly.
Cheers!
Ross Shain (Boris FX team).

Hi Ross,
Sounds interesting!

@elizabeth.postol hi again, I wonder if you can give me a steer in the right direction for something else with this scene.

I am trying to recreate: Tutorial: Bullet Time - YouTube

I now have me cut out OK, and the muzzle flash frozen in time.

Two other effects I would like to add, to emphasise the 3d effect, are the dust particles in that film, which actually look more like strands, and the smoke effect.

The main thing is that the camera is appearing to move around the smoke and dust particles in the same way as it rotates round the body, and I am trying to think how I achieve that. Is it track the footage, resolve camera, then create a new layer with dust or smoke, and attach the effect to the resolved camera track?

Glad you achieved the masking part!

Sure I’ll help, the workflow would be following:

  • solve the camera
  • export to After Effects
  • drop all your desired smoke assets into composition.
  • turn smoke layers into 3D. To do that, simply select the 3D layer switch (it looks like a cube) in the Timeline or go to Layer > 3D Layer.

You do not need to link it to the camera. As long as you have one camera layer in your composition, all 3D layers will automatically follow it. However, you need to position your 3D layers in the scene. You may want some layers to be closer to the camera, and push back the other ones.To do that, you need to use navigation gismo that will appear after you turn layer 3D, and move layer closer or further along the Z axis.

@elizabeth.postol sorry, forgot to mention, I use da vinci resolve!

@elizabeth.postol final version here: The Bug-A-Salt Review: The most fun you can have killing a fly - YouTube
Thanks again for your help.

Hey, Great job Arlo!
You put so much effort in your reviews!

1 Like