interactive scene

19
[Interactive Scene] LMC 6310: P3 Interactive Installation using Kinect Apurva Gupta MS-HCI

Upload: apurva-gupta

Post on 15-Apr-2017

311 views

Category:

Design


0 download

TRANSCRIPT

[Interactive Scene]LMC 6310:

P3 Interactive Installation using Kinect

Apurva GuptaMS-HCI

Brief

The installation tries to capture the movements of the user and translate those movements into a pre-designed character on the screen.

The installation uses Kinect for detecting user movements and translating the motion data using Processing.

Concept Design Code Iterate

PROCESS

ConceptThere will be multiple scenes which the user can choose from. Kinect will detect user’s movements and the character in the scene will move according to the user’s movements.

Schematic of the concept Imagined Scenes

Inspiration

V-Motionhttps://vimeo.com/45417241

Skeleton tracking:https://vimeo.com/110735619

Skeleton tracking:https://vimeo.com/109920497

Absolut Blank:https://vimeo.com/55014954

DesignAssets for 2 scenes was designed (Head, Body, hands, Legs, Background)1. Astronaut Scene

2. Grim Scene

Astronaut Scene

Grim Scene

Code + IterationUsing SimpleOpenNI and processing, 2 iterations were tried

1. Using user-map as a body and image of helmet/ grim’s head

2. Using skeletal tracking and pasting images of body parts on the joints

Usermap

Skeletal Tracking

Aesthetic Choices

Astronaut Scene

• Use of body parts and not stick figure lines

• Various parts are not connected but disjointed giving a feel of puppet since the motion is not smooth. Thus, parts were pasted to look like a puppet.

• Music was added to add fun factor aimed at engaging the user.

Grim Scene

• Use of stick figure lines for hands.

• Music was added to add fun factor aimed at engaging the user.

• I tried to add face recognition to Grim’s face and have burning eyes but due to limits of time could not try that out.

Computational Choices

Computational Choices

• Trigonometry and Calculus were used to calculate the angle and pasting images

• Translation of coordinate system to obtain desired anchor point for rotation

• Scaling factor used to make the character consistent with movements in Z-axis

Suggested Improvements1. Smooth of joint detection

2. Face tracking

3. Try capturing motion in 3D

4. Use of user maps instead of skeletal tracking.

AcknowledgementsProf. Brian Magerko

Devin Wilson

Mikhail Jacob

Mehul Agrawal