Characterizer offers 3 way of building a puppet:

1) Automatic Webcam where you sit in front of the camera and follow the instructions

2) Manual Webcam where you sit in front of the camera and follow the instructions (but instructions are more detailed and numerous)

3) Image where you submit one single image with minimalistic rigging

I am suggesting that Characterizer should offer a fourth option that sounds quite simple to me which is:

4) Manual multiple images where you can enter several images or photos following the same instructions as Manual Webcam but instead of having to sit yourself in front of the Webcam, you can feed Characterizer with Pictures/Photos. You can feed Characterizer with:

• Pictures of a puppet you already created (so that you can apply the various styles offered by Characterizer)




• Photos / Screenshots of a celebrity or friend you want to puppetize

Characterizer system is awesome but, it has trouble recognizing and masking a couple of things like beards and certain skin tones even with lighting.The capture process is incredibly useful though and it would be awesome to be able to send those initial captures to photoshop to do things like edit what is masked out, manually draw faces based on the captures, and manipulate the captures themselves.
OR
at a minimum add a step between the captures and characterization where you can tell characterizer what it should see as face, head hair, facial hair, etc. Similar to the process when adding a source for characterizer to use on the image captures.

It would be great to have a Teleprompter feature that prompts the person speaking with a visual text of a speech or script.

This could take the form of a Teleprompter tab or an onion skin over the scene as it's being recorded (that is, the script is superimposed over the scene as it's recorded).

Since I do most of my recordings on an iMac, I could see undocking the Teleprompter tab and placing it at the upper middle of the Desktop, just below there the built-in web cam is located. Or, reading the text as is scrolls by while in a text field that floats where I place it in the Scene tab would also be a good implementation.

If producing animations for animated shows, one would probably still have their actors learn the script and speak from memory. But, not all of us can commit a script to memory that quickly.

The problem i'm having right now with creating different emotion sets (sad, confused, happy etc.) is that it's something very basic and important, yet time consuming and complicated to do on a live performance.

my idea is:
if we could somehow for example, connect between the "smile" behavior that the camera tracks from the face, and connect it to other elements created for a smile set like eyebrows, eyes, cheeks etc. which will be triggered automatically.

if you make a sad/smile/surprised face, the camera detects it and automatically triggers sad eyes, eyebrows, maybe even body poses!

basic emotions that are very important for the performance, automatically by the camera, without needing to create complex sets of triggers, key binds etc.

Just a thought :)

I would like to incorporate multiple backgrounds and characters during a live presentation, to create the look of a two-way conversation as seen on TV and film. I want to be able to switch backgrounds every time a new character speaks, and have both the new character and background appear onscreen.

I can currently achieve background switching by stacking multiple backgrounds in the timeline and then manually fading the opacity of the background(s) on top, but it is slow, manual, and very clear to the audience what is going on.

On a similar note, switching out the characters is currently possible by moving them out of frame, but again, this is not keyframe-able or instantaneous.

Ideally, I would like to be able to create keyframes for the background and character Transform values (in this case, opacity and Position X, respectively) and assign them to a keyboard shortcut or MIDI keyboard so I can instantly toggle between them during a stream.

Would this be possible to program?