I'm trying to make the posing of a 3D character so that it covers perfectly the actor's body behind that's mapped on a 2d plane :


as you can see I haven't been able to cover it perfectly. It seems easy but it isn't. I would like to know about a set of tips and tricks to do it better and faster.

For example,is there a way to make the mesh transparent ? It maybe easier to match the mesh borders with the actor's body borders. Infact if I put the mesh in wireframe mode,the texture of the plan behind it goes away. How can I keep it displayed ?

4 1 2 3

  • 1
    $\begingroup$ What you're trying to do here is a tall order for the inexperienced so you will need some simple exercises to get you there. I would be doing this with a form of "rotoscoping" in reverse. Instead of cutting a hole in the video to remove the original character, have your 3d model a little larger than the video actor, and animate (pose) the model frame by frame. Ordinary rotoscoping can be a laborious task but it's not difficult so long as you don't use shapekeys. They tend to get unstable when making adjustments later $\endgroup$ – Edgel3D Mar 17 '19 at 2:17
  • 1
    $\begingroup$ A further suggestion - You could try using rotoscoping to make masks (one for each frame) that will make a hole in the background video. i.e "cut out" the person in the video. Instead of placing the 3D model in front of the video, try placing it behind the hole. That way it will also be masked so it will always fit the video character's position, even if limbs and so forth are a little bit out. The video around the hole will be untouched. $\endgroup$ – Edgel3D Mar 17 '19 at 2:30

Instead of using an image on a plane use the image as background and composite later using the compositor.

Put the background image in Front and set the opacity to a value where you can see the image and your object.

enter image description here

Consider a few more things: The settings for the camera in blender will make a difference. If you are trying to match the placement of 3D elements on an image, the point where things match reality will only be valid for a single point in space. If the camera is a bit higher or lower or has a different perspective, then everything you do will be off. In other words the camera settings (field of view, height and rotation) should also match those of the original camera.

Use Blam or similar tools to determine the settings and placement of the camera in blender.

Create very basic proxy auxiliary objects that might guide you (like the couch) and help you visualize things when you are not looking through the camera view.

NOTE: Blam is no longer mantained and has been replaced by Fspy: https://fspy.io/, but there is no importer for blender 2.79 or previous versions.

For a detailed post on how to use blam and Fspy read: How can I recreate geometry using a photograph?

| improve this answer | |
  • $\begingroup$ can u suggest to me a video tutorial that can help me to understand the workflow that you talked about ? $\endgroup$ – Marietto Mar 12 '19 at 0:05
  • $\begingroup$ Start with the link on the answer, then look for information on camera mapping $\endgroup$ – user1853 Mar 12 '19 at 0:13

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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