Rocky Movie Computer Fight Simulation

The 2006 Sylvester Stallone movie Rocky Balboa featured a scene where a computer simulation estimates the outcome of a hypothetical fight. Can computers do that?

In this movie, Stallone's heavyweight boxer character Rocky has retired. In one scene, Rocky is watching ESPN news, and is startled by a broadcast. It features a computer simulation depicting a fantasy fight, and predicts the outcome of how Rocky would have fought in his prime against the movie's present-day heavyweight champion Mason Dixon, spurring his return to the ring. A real pro boxer plays Mason Dixon's character - Antonio "Magic Man" Tarver is a southpaw from Florida, and former light heavyweight world champion.

Computer generated fights that generate real probable outcomes in real time 3-D are not yet possible outside the movie industry. What can real computers do?

An actual "fantasy fight" computer simulation was done in 1970. It was the SuperFight between Muhammad Ali and Rocky Marciano. Rocky (Rocco) Marciano was heavyweight champion of the world from 1952 to 1956. Muhammad Ali was three-time World Heavyweight Champion in the 1970s. Marciano and Ali fought in different eras and never fought an actual bout.

To make the SuperFight, probability formulas were entered into a computer. No drawings, just numbers. Ali and Marciano met in real life on a filmset to film numerous short segments showing possible parts of a fight. Marciano was already retired 13 years and wore a toupee. The short segments were then spliced together to match the already done computer outcome to make a movie that looked like a real fight or computer-generation of one, but was not. The predicted outcome had already been generated by computer, but the fighters and movie were the real people, not computer generated. The outcome may or may not have reflected actual ability of the fighters or the real outcome.

In the mid 1980s, I was investigating which differences in human movement determined injury potential and athletic performance. In one study, I wanted to know what made the difference between the punch of a black belt martial artist and the same punch by an athletic person without training.

In present day, a camera can be hooked directly to a computer, which picks up the locations of the person's joints at each point in time, generating a computer image of the person as they move in real time. Software automatically calculates, draws, and records the image on the screen.

Back when I did these studies, we didn't have any of that. I did it all manually. I filmed two subjects using 16mm high speed filming. An athletic man who had never done martial arts was subject #1. My husband Paul, who had earned his black belt a few years before that, volunteered as subject #2. I put markers over the center points of their major joints, and bands around joints which initially faced the camera but would rotate during the punch, so that the joint center would still be determined. Both executed a front reverse punch with their dominant arm. (Paul had to use traditional hyperlordotic position to match the untrained subject, rather than healthier neutral spine position, just for this comparison. We have done other studies comparing my neutral spine adjustment and found it to be a stronger punch - try it here.)

After waiting a week for film developing, I went into a darkened lab and used a film projector to throw the image of each of the thousands of frames, one by one, against a large computer digitizing tablet hung on a wall. I then digitized each joint point of each projected image, in each frame, of both subjects, frame by frame, with a digitizing Graf-pen. I sent data points from each frame by (300 baud acoustic coupling) modem to a text editor on a mainframe in another building at the University's new computer center. I wrote my own FORTRAN programs to generate data summaries and used packaged International Mathematical and Statistical Libraries (IMSL) cubic spline programs and subroutines for data smoothing. This was all to get each knee, hip, ankle, shoulder, wrist, elbow, neck and other filmed joint points into a computer to see exactly where and how fast they moved. Projecting each frame against the wall also allowed me to trace the subjects' outlines to make series of line drawings of their punch, and to make stick figures showing joint center placement. Here are some data and the actual drawings I made:
























































Not long after, with improvements in automating this process, action video games were flourishing. I was invited to a computer-generated imagery (CGI) development studio to be their "movement representation figure." They put the dots on my joint centers and filmed me using high-speed 3D computer graphics modeling as I did martial arts and tumbling moves. Not just one punch, painstakingly done, but jumping, spinning, flying all over the studio, and up and down walls.

The software automatically generated a mathematical, "wireframe" 2-D representation of my three-dimensional form. From it they animated a wild female warrior action figure for their fighting/mission genre arcade and video gameplay. They also used skeletal animation for when I would morph (on-screen) into various animal forms. I never got royalties but it was fun.

More:

• This is a big fun topic. For more, click the label "martial arts" under this article.
• I can write more about motion capture analysis of various sports if anyone is interested.
• Learn how to adjust to neutral spine for a stronger punch without lower back pain.
  
• The Great Muhammad Ali has been diagnosed with "Pugilistic Parkinson's syndrome" of tremors, muscle rigidity and slowness - with the possibility, still not fully determined, if due from the damage of a boxing career. See Rocky IV and Head Injury.

Labels: computer, injury, lordosis, martial arts, movie/media fitness, neutral spine, Parkinson, speed

Permalink | Email Post