First Test in Space

This past week I had my first in-house test with my actor running on the mounted Christie projectors.

Below are two pictures, one with the full image fill and the other with the mapped image.

The map is not exactly precise since I was using a piece of white cloth draped over the box.

Anti-Aliasing time-lapse

Preliminary tests on the actor made clear the necessity for anti-aliasing within the program.

Patches within Isadora vary in density based on the amount and size of media within it. A patch that shows still images with be slightly faster than those that have video files. The same is true for patches with a combination of both along with various effects. Having a high level of anti-aliasing can slow the computers processing and cause lag. It is imperative that in live performances cues trigger at the precise moment they are meant to. Too much lag in a heavy media show could lead to eventual crashing of Isadora or the computer itself. To help compensate for the diverse types of performances and machines, 8 different levels of anti-aliasing were incorporated into the actor depending on the needs of the event. The levels range on a scale from 0-7. Zero meaning no anti-aliasing is being applied to the map. From there the number of tests that are performed in the code increases exponentially. The number of tests are based on jitter point arrays for 2,3,4,8,15,24 and 66 jitters.

Anti-aliasing is only performed when the map is initially drawn and when there are any subsequent changes to it. When mapping a still image this effect can not be observed. However, if anti-aliasing were constantly being applied, mapping a video file with changing frames would dramatically slow down the playback because the map would have to redraw every frame. Even though the anti-aliasing is applied only when the map is drawn or redrawn, it is recommended that when mapping in Isadora the anti-aliasing be set to zero to avoid any issues. Changing anti-aliasing value temporarily slows the media down to allow for recalculation of the alpha mask. At the very least, mapping at 7 optimization is discouraged since the number of tests being performed on the pixels is substantially higher than the other levels.

ú Source Code: http://www.opengl.org/resources/code/samples/redbook/jitter.h

Here is a picture comparing the different angles of an image with no anti-aliasing and fully optimized level 7 anti-aliasing with 66 jitter points.

Here is a time-lapse video showing the 8 different states of anti-aliasing used in the Isadora Actor.

AutoCAD Drafting for Construction

Construction on the installations pieces is underway and going relatively well considering the limited experience I have in carpentry. I put together a cut list based on the design and have completed about 35% of the construction.

Below is an image from the AutoCAD draft.

Installation Final Design

Here is the final designs for the installation pieces.

Full stage view with REVE PICT walls.

Both Stage Left Installations

Stage Left Center: AKA "The Tall One"

Stage Left: AKA "The Arch"

The video below shows the different configurations of the SR installation.

Stage Mouse Watcher Editing

Several advances have been made on the actor side of my thesis project. First, I discovered a way to use the stage mouse watcher in conjunction with the polygon mapper actor to map individual vertices. This differs from previous iterations because I can map both the x and y on the stage itself instead of using a control panel 2D slider box for each point. The issues with the previous control panel were that each vertex was located in a different box, as seen below.

With the new iteration, the need for a control panel is eliminated and all mapping is done on the stage itself. This cuts down the time it would take to map an object. It also allows for a more precise mapping since there is no disconnect between the object and the vertex points. The downfall of this process is no outline map is generated for the object. It would be beneficial to have a final picture of the boundaries of the installation for later use.

In addition to the mapping being done on screen, there are several other functions the actor now has. First, the edit mode can be toggled on and off by right clicking while on the stage. There is no need to physically change an input in the patch. This creates a more efficient work process. However, when working on multiple maps in one patch right clicking to turn the edit mode off will turn the edit mode on or off for other maps in the same patch depending on its current status. The only way to avoid this is to disconnect the line between the trigger and the toggle on the maps when you are not using them. To easily see the edit mode status a series of green and red dots appear when editing and disappear when not editing.

Another issue that was first encountered when using the stage mouse watcher was the vertex being edited would automatically jump to the location of the mouse. This is not a huge issue when first editing a map. However, it creates a very time-consuming issue when fine tuning a map. The vertex would jump to the arbitrary point of the mouse thus limiting the ability to slightly adjust a vertex quickly. This issue was resolved by creating a "sticking" effect with the mouse and edit vertex. Once a vertex was being edited the "dot" will stay in place until the mouse moves within a 5 x 5 pixel radius of the vertex. At that point, the "dot" will attach itself to the mouse and follow it to the desired location. This function prevents the vertices from jumping to random and unwanted coordinates. Further testing on a large scale revealed that using a 5 x 5 pixel ration is too large and should be scaled down for more exact mapping.

Moving from one vertex to another when editing has been simplified. The left and right arrow keys move back and forth from one edit dot to another in a complete circle. This also streamlines the mapping process. A video of the edit rotation can be seen below. After speaking with a professional projection designer Matthew Haber about this aspect, he brought up a potential problem with using the left and right arrow keys to rotate through the edit dots. In certain situations, designers will use the left and right arrow keys as triggers for the show itself to move through cues. The problem is minimal since the edit mode controls the ability to move left and right. Only when the edit mode is on do the arrow keys function in that way. The triggers for cues need to be put in place once the mapping has been completed to avoid an overlap in use. More practical testing will have to be done to ensure the potential problem is minimized. One easy way to avoid the issue is to change the left and right arrows to letters on a keyboard, or let them be chosen by the user.

It is imperative that a designer not be limited to the number of vertices for any given map. This is the case in most video projection tools that are out there, map map map for Modul8, VPT, etc. They often give you only 4 - 16. With the polygon mapper tool there is an unlimited amount of vertices that can be added at any point in the mapping. Currently, there is an initial value of 0,0 all vertices in the top left corner of the stage. More testing is needed to determine if that is truly the best place for an initial value.

This video shows:

~ editing with the stage mouse watcher (the mouse can't be seen)

~ rotating through the edit dots

~ adding vertices

~ turning the edit mode on and off (edit dots disappear)

Art Installation piece

This image represents the stage left - center installation piece.

Stage right installation piece - standard configuration

Stage right installation piece - throne configuration

The center box will moved by an actor and secured to ensure the video map is accurate.

Live Design Projection Master Class

About 10 days ago I registered for the Projection Master Class in Orlando on October 26th and 27th. This is the day before my thesis defense on Friday Oct. 28th. This is probably ill-advised but the fact the class is only in Orlando compared to Las Vegas last year, I think I will take my chances. Not to mention, a large portion of the class will focus specifically on projection mapping, my thesis. I'm sure we will learn about all these programs that allow you to wrap projections, etc, which will in turn make me question everything I've done on my thesis. Thankfully, the resources that I have are limited so I can't feel too bad. It's like I'm doing projection mapping on a shoestring.

I'm excited about all the techniques and software I will be exposed to at this conference, but mostly I am looking forward to meeting people who are actually working in the industry. The world of projection design is small and in my research I have come across works by all the presenters: Mode Studios, Vello Virkhaus, Jake Pinholster (a UF Grad). I feel this opportunity is invaluable to my career and I will just have to be extra prudent to get everything completed in time.