Process For Projecting on Cube in EMMA

Step by step process

A.)-For this first test the Audience's view wasn't taken into much consideration, but it should. (in order to view the projection correctly you need to be at projector level which in this case was lying on the floor.)

B.)-Our next step was to measure and arrange our objects in physical space.

1.The object we were projecting onto was a rectangular box that measured 77.75"x30"x18".

2.The projector was, in this case, arbitrarily placed 147.5 " away from the center front face of the box which was enough distance project on the entire surface. When considering projector placement it should in general be as Matt wrote- as close as practical for maximum resolution and minimum background bleed. I also tried to place it perpendicular to the box. At this stage it is also important to either zoom all the way in or all the way out so that calculations down the road are easier to find.

C.)Modeling a Virtual "set" that matched the Physical one.

1.I chose a global origin of 0,15,0 so that the box could be at the center of the virtual space with the bottom face flush with the ground plane (translate Y=15 since the box would be 30 units high)

2. It is important to also pick your units of measure within maya. We measure the physical in inches so chose inches for maya.

3.I built a polygonal cube that matched the measurements of the physical cube 77.75x30x18 and placed it at the origin. I left rotations at o,o,o since we would be projecting on the object perpendicular from one side only (in order to calibrate our maya camera with the projector we started simple with the hopes the same measurements would work when the objects are angled).

4. Next I created a camera within maya. Using the measuring tool I placed it 147.5" away from the front edge and center of the box.

5. After the camera positions match in physical and virtual space, we needed match the camera's attributes with those of the porjector.

-Aspect Ratio: 4:3- The easiest. The projector was set to 4:3 so we did the same for the camera.

-Field of View: 35.48: Can be found hopefully in the manual, or what we did was project a perfect square onto a perpindicular surface. Measure the length from lens to center of projection then from center to edge of projection we could figure out the angle using the Pythagorean theorem. Multiply by two since we only calulated half the field of view. I'm also pretty sure that the Aperture setting directly affects/calculates field of view so finding that in the manual would also be idea.

-Aperture: .902-.678: So we had a lot of trouble figuring this number out at first. Through some rough mathematics and some other measurements we got a projection that lined up closely. We later found out that some of the confusion lied in somewhat different understanding of what aperture meant in Maya compared to the Projector. It turns out that for a projector, aperture is reffering to the the LCD panel size, which according to some sources is the diagonal of the aperture measurements. Using the value from the manual produced bad results in Maya, as before, but produced better results once we realized we had been using the wrong end of the focal length range.(Hence why it is important to zoom all the way in or out in the beginning).

At this point, with all of our measurements and camera/projector settings matching in both maya and in EMMA, we were able project onto our cube properly without manual tweaking and the render-re-render process.

We did a quick experiment with rotating the physical box and then matching that rotation for the virtual to see if our calculations would still be effective and they were.

The animations were secondary for the most part. Just wanted to test what kind of depth you could get inside such a small space. I didn't push it to the max but I think it was pretty successful overall.

Projection Mapping Second pass WIP

In recent years, the art of 3D projection mapping has been climbing in popularity. According to Wikipedia, 3D projection mapping is defined as “any method of mapping three-dimensional points to a two-dimensional plane.” The most common form of this technique is the mapping of geometrical shapes and architectural facades. These 3D projections give artists the power “change” the objects surface and create illusions of depth, forced perspective as well as more dramatic special effects such as objects crumbling, morphing and turning into cloth. However before these artistic visualizations can be realized, you must understand how the mapping process works and that is what I have been researching in recent weeks.
Initially my knowledge of the projection process was very limited. I was not aware of the intricacies of process that were necessary for accurate projections. I was under the impression that if and object was accurately replicated in 3D software it was merely the act of matching the digital camera position with the projectors position in real space that would allow effective mapping. I soon learned that these notions were incorrect. While struggling to align our 3D rendered models with real world objects, we bypassed the process by manipulating geometries to make up for our miscalculations. However, we soon realized that while this may work for static projections, animation would likely expose the flaws in our process.
Creating an accurate and intuitive method for mapping these projections became imperative and Matt, Ben and I began experimenting. We decided to start simple by projecting onto a single flat surface in the hopes that what we discovered could translate into more complicated perspective projections. The first step was accurately measuring our rectangular box: 77.75w x 30”h x 18”d. With these measurements we were able to create a replica within Maya. The next step was to translate the distance between real-world projector and object to our digital camera and 3D object. The projector was placed perpendicular to the box and at a distance of 144.5 inches measured from lens to the box’s projector facing side. The virtual box’s position in was coordinates (xyz) 0,15,0. Since the box is 30 inches high chose its Y position 15 inches above the ground plane which put the box’s bottom flush with the floor. The camera was created and placed 144.5 in. away from the virtual box facing side which would put it 154.5 in. away from the 0,0,0 origin. The physical projector lens was measured 3.25 inches off the ground so we also bumped the digital camera up 3.5 in. in the Y direction.
After figuring the spatial relationships in physical and digital space it was necessary to adjust the digital camera’s Focal Length, Angle of View, Aperture Settings etc. to match that of our projector. These settings are projector specific and in this case we were using a Canon ______. The angle of view was calculated by projecting onto a close, perpendicular service so that the projection was square. We measured first the distance from projector lens to the center of the projection to get the adjacent side (X). Then we measured the width of the projector and divided it by two to get our opposite side (Y). Using the Pythagorean theorem we found the Hypotenuse (H). The tan ø =y/x, or our FOV angle =2* arctan (y/x).

The focal length was actually found in the Canon Owners manual which was 34.5. I believe that only one of these numbers needs to be found because their values directly correlate to one another. We used the found focal length since it was the most accurate and plugged it into our digital camera’s focal length value. However, manuals usually give you a focal length range to account for the zoom settings of the projector. We assumed that since we were zoomed all the way in, we would use the larger focal length. We discovered later that it was the smaller of the two numbers we should have used.

Learning to Align

So amidst all of our experiments and conceptual designs, properly aligning 3d objects within maya with there real world counterparts has been an arduous process. It goes something like this, Rotate, tweak, rotate, tweak, forever and ever until it is close. There is also the painstaking process of rendering out stills so we can full screen them and see if they align. Basically a long long process. Matt and Ben recently created a jitter patch/program that would allow us to isolate only our renderable area from maya and full screen it through the projector for real time alignment. As nice as it is, we still need to figure out how to accurately align camera/projector settings. Matt, Ben and I spent a few hours yesterday trying to figure this stuff out.

The scene we set up was simple. We decided to calculate a straight on projection to begin with so we only had to worry about a single face. I measured our physical box and created an accurate replica in maya.

1.77.75" wide, 30" high and 18"deep

2.Then we measured the distance from the projector lens to the box so we could properly position the digital camera.

3.Either measure the projectors angle of view or find out its focal length via. manual (AOV and FL seem to adjust to each other)

4.Make sure digital camera's resolution and film aspect ration match the projector settings. We used 1024x768 and 1.33 or 4:3.

5.The sketchiest part for me at this point is the camera aperture. I/we need to further understand what it is and what it controls. It seems to affect multiple settings in maya and needs to be adjusted accordingly. We could not find the projectors aperture setting so we found a formula online and used it to figure it out. This was much more Ben's area than mine and I still do not fully understand the process.

We came very close to aligning the box/projection yesterday. There are so many working factors that a small error in a single step could account for our miscalculations. Or there is another aspect we haven't found yet that is the culprit.

Appologies for Procrastination

Well, somehow I've allowed my documentation process to fall behind. So here is a catch up post on what I've been working on.

Recently it has been decided that a rear projection screen may not be the best plan of attack for the theater set up. Instead we have been assigned the task of designing a back wall that will be receded from the stage which will be projected on by us. I believe the idea is to move the wall far enough behind the stage so that the projections will not be interfered with by the actors i.e. shadow castings. As of now we are pulling inspiration from artists such as Cherdak in creating a structure with strong perspective and a compartmentalized cubist aesthetic. For some reason I tend to skip sketching stages and dive straight into building so I created the following model to test how certain shapes and design decisions will affect projections.


.



Using this model I utilized our awesome tiny laser projector. I mounted it on a tripod and hooked it up to my laptop and painted onto it live. After a while I came up with the the following.








Yes it is rough, but I learned a lot about designing the wall. I think that projecting from straight on would be our best bet to eliminate unwanted shadows, but I don't think that's a possibility. Current ideas suggest that we will be projecting from high with a short throw distance which, on a wall like this, will create hard and long shadows cast from shelves and protruding objects. I think we need to think of a shallower depth design and use projections to create the depth.

Prior to this experiment we had a photo shoot to prepare for. It was a proposal for the show to be picked up and needed to show aspects of how projections could be used in a theatrical setting. Unfortunately we had one week and our projections were static and couldn't explore movement and animation, but all in all I thought the setup looked good.

Our first setup was an SAE interview. I had strep throat at this time and did most of my work from home. We are adapting an older tale that involves monkeys and a narrative about the importance of total team work. I am creating the monkey model/rig etc. to be used at a later date. He is hanging out on the bookshelves.





Next was a mock discussion about animals use of camouflage. We utilized a mallard for our example. I created a 3d model in Maya of a duck then Matt Lewis took it and generated a real life model. That figure was wrapped and projected on several times to show different stages/examples of camouflage. (center of the table, can barely see it)







Finally was a experiment in painting live onto actors and an interrogation scene. This was all Nikki and Vita.