2

I am trying to see if there's a term for what I'm about to explain or a better way of doing this. Basically, my company is filming an actor from the waist up for short, 1 or 2 sentence clips. These clips need to be strung together to look as seamless as possible.

The main problem is that the model has to rely on muscle memory to remember where her head (specifically eyes, nose, mouth) and shoulders are in the frame.

My solution would be to use a mirror with a hole cut out for a camera lens. On the mirror would be marks showing where her features should return to after she says something.

Is this typical, or is there a better way to do this? Or possibly to have another monitor with an overlay of the original position on it?

I'd just like to try to make sure she doesn't constantly look away from the camera to achieve this.

Any thoughts?

Improve this question
12
  • This seems incredibly difficult. The mirror-with-a-hole option is smart! I'd like to find out more about how this footage will be used and why the clips need to be strung together to look nearly seamless. As for a solution, one option would be to have 2 cameras on the actor: one directly in front, and one off to the side. Cut back and forth between the two, so it appears seamless but disguises variations in her position.
    – BrettFromLA
    Commented Dec 1, 2016 at 21:06
  • Thanks @BrettFromLA, want to hear something funny? We're good friends IRL! The footage would be used on a mobile device, basically talking to a person, so we can't have cuts like that...
    – ntgCleaner
    Commented Dec 1, 2016 at 21:10
  • @ntgCleaner If all the footage has to be head-on, you could start each new segment with a quarter or eighth turn into the shot. Everything spoken will be face forward, with any small change disguised in the turn. As Brett says, anything else is likely to be imperfect -- even a very small discrepancy will be noticeable. Good luck, though.
    – Jim Mack
    Commented Dec 1, 2016 at 21:40
  • @JimMack, Thank you. I'll take this into consideration, though what I didn't mention is that we have this "idle" clip where the actor is "waiting" for you to talk to her and that's pretty static.
    – ntgCleaner
    Commented Dec 1, 2016 at 21:54
  • 2
    @NoahL, It's not about remembering lines. It's all a part of machine learning that we're adapting to use video responses, instead of TTS responses. It's basically a canned response, but we're going to have hundreds (eventually thousands) of videos that should all feel like they line up together.
    – ntgCleaner
    Commented Mar 3, 2017 at 3:24

4 Answers 4

Reset to default
2

One option, if you don't mind the actor keeping her head still, is to be partially lying down. She could be in a recliner, leaning back, with her head against the headrest so she never moves it. She would feel the headrest behind her so she could tell if she'd turned her head, and it wouldn't "drift" left or right because the friction of the headrest would keep it in place.

Of course you would want to shoot all her lines in as short a session as possible so she didn't get uncomfortable - like 5 minutes for all the lines and retakes.

Improve this answer
2
  • 2
    You could have something like a headrest but for the model standing up—like a bit of foam on a C-Stand or something, which you could matte out of the final shot.
    – stib
    Commented Dec 2, 2016 at 1:11
  • Great suggestion!, maybe an incline wall too? just enough to lean on. and @stib, also a great suggestion!
    – ntgCleaner
    Commented Dec 2, 2016 at 13:05
0

Back in the "golden age" of movie production, they had a "continuity person" or a "script supervisor" whose job it was to keep everything matched so that shots could be inter-cut without discontinuities. They used a Polaroid camera to record exactly what the actor looked like, what facial expression, angle of gaze, tilt of head, how empty the cup or half-burnt the cigarette, etc. etc. etc.

These days with video, it is easier because we have instantly available video recordings (or still frame captures) which can be compared by half-dissolving them on a monitor to exactly match the new shot with a previous reference frame.

Improve this answer
1
  • That's what I was thinking about too! We're going to need a continuity person in there while filming at the very least. Thank you! I will add this to the list!
    – ntgCleaner
    Commented Dec 2, 2016 at 13:06
0

(This idea is similar to the one in Richard Crowley's second paragraph.) If you can view your spokesperson on a full-sized monitor, have her take her position, and then put small pieces of opaque tape on the screen: one just to the left of her head, one just to the right of her head, one at the top of her head, one for each elbow, one for each hand, etc. Then it will be easy to tell her how to realign after each take. (It would not be easy for her to realign at the end of each take.)

Improve this answer
0

An oldie, but let me do some brainstorming.

  1. If you are going to have thousands of clips, you definitely need to construct a rig, not only for the models but also for the lights and cameras.

    For the model make one structure to be adapted on a comfortable chair that gently directs the orientation of the head.

  2. Prepare a digital projector to shot a grid or a silhouette from one side of the model (be careful with the eyes) as a reference image.

  3. Or prepare a set of orthogonal cameras, that can Livestream to a computer. If you are using windows, there is a small program called "Ghost win Real transparent windows" that can overlay one application over the other, a reference image. These cameras need to be also attached to the rig.

  4. You could use a flat background, so you can align the clips in post pro.

  5. Combine all the above.

Improve this answer

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.