Filming at an angle to fit more in frame or shooting head on?


(Andrew Gant) #1

Hey all!

We’ve talked about this in Slack in the past, but I wanted to bring it back up again as I always get paranoid when I’m shooting. The main issue here is I’m always trying to keep the subject as close to the sensor as possible in order to get maximum detail on the scan. The closer they are to the Kinect, the more detail there will be.

When shooting someone head-on, with the Kinect lined up flat against the subject (no tilting) it will often be hard to fit the person in the frame all the way. The frame may also have excess head rood that could be going to waste. Tilting the Kinect down and moving the talent closer will enable me to move them closer and fill the frame easier. Even if you are doing a closeup of someone’s head, tilting the rig will still allow someone to come a bit closer in certain situations depending on what you are trying to capture.

The result of this can, however, cause your talent to be distorted or tilted in the end. I think I remember @Alexander telling me once to use tilting sparingly, but it is still okay to do in certain situations. Does anyone want to provide a bit more insight on this, and refresh this topic a bit here in the forums? I still would love any help on this.


(jillian morrow) #2

Hey Andrew! I second @Alexander’s view of using tilting of the rig in moderation. It is key to fill the frame with your subject, but tilting too much will cause some distortion in the body. Not a lot of distortion, but the legs can warp a bit if doing a full body shot.

I’d love to get some other opinions in here too to see what others think!


(Paul Grey) #3

What about mounting the Kinect/Sensor vertically?


(jillian morrow) #4

We recommend this too, Paul! Thanks for mentioning it. I will note that, if you are relying on body index for capturing, recording vertically will break the body detection from the Kinect. So if using the body index data for background removal, you will just need to capture horizontally.


(Paul Grey) #5

Ok I was confused by this so did some testing. depthkit heavily relies on body index, so if you’re recording vertically you must use a custom mask.

It would be great to see vertical recording integrated in to the current workflow, perhaps by adding a rotate by 90 degrees option.