Comments

6 comments

  • Avatar
    Ondrej Trhan CR

    Hi Mark,

    by FACS do you mean Facial Action Coding System? (I found more meanings for this acronym)

    If so, the best way is to have a camera rig around the head with known cameras positions and fixed head. You can take a video of facial movement in high video quality or take images in same time. This could be very similar to a full body scanning, but you will scan just a head.
    You will need to have some cameras around the head on a fixed rig. And also placed some control point with known coordinates or distances between these points. Then you can use XMP's workflow to process your images into models.
    For some inspiration you can watch these tutorials: https://www.youtube.com/watch?v=5VyYLaNxHz0&t=61shttps://www.youtube.com/watch?v=E2QCtisvdDc&t=21shttps://www.youtube.com/watch?v=dKX48fLuyv0
    For each expression you need to create an own model which can be then merged into movement in 3rd party software.

    0
    Comment actions Permalink
  • Avatar
    MarkRuff

    Thanks Ondrej,

    I should have been more specific.

    It's a full body scan system 170 cameras using around 50 cameras for the face/head.

    There are 4 supports for the head to minimise movement of the head between captures.
    The rig is mobile and has a 'door' of cameras for entry and exit, so consistent alignment of some cameras is not possible.

    I have no issue creating the models with scaling, and use plenty of markers, rather, consistent alignment between the models in RC since there are no ground control points.

    Should I have a master model with XMP info and apply to all other models?

    0
    Comment actions Permalink
  • Avatar
    Ondrej Trhan CR

    Hi Mark, if it is a movable system, then it will be hard to manage same camera position for each set-up. But for one person it should work. Then you can use the XMPs, but you will need to create a new one for each of your model.

    If there are markers, you can export coordinates of those markers and use them as GCPs in next process https://support.capturingreality.com/hc/en-us/articles/360012872579-Export-Control-points-with-XYZ-coordinates

    0
    Comment actions Permalink
  • Avatar
    MarkRuff

    Thanks again Andrej,
    Yes, once the door is closed then cameras should not move for all the required takes.

    Not much of a coder, so I am a little lost with that link.
    Can you teach me a little more?

    M

    0
    Comment actions Permalink
  • Avatar
    MarkRuff

    May have to re ask the question:)
    Sorry about any confusion.

    I am after consistent 'orientation' of each model in the series of captures.

    0
    Comment actions Permalink
  • Avatar
    Ondrej Trhan CR

    If you want to export control points and then use them as GCP, copy the example from attached link, then go to C:\Program Files\Capturing Reality\RealityCapture and open the controlpoints.xml file. And replace the 43rd line with this. You can also use it as separate option, but then you need to change formatID and descID to have exceptional IDs.

    If you do so, you will be able to export coordinates of your control point, which will be the same in next model processes and therefore all created models will be in the same coordinate system and basically oriented (if the model won't change its position).

     

    0
    Comment actions Permalink

Please sign in to leave a comment.