Incompetence At It's Finest

Over the course of the weekend I decided to do some tests with tracking. I however ran into numerous problems, all of which I can attribute to my own incompetence; or so I've convinced myself.

There isn't much to update in the form of words. Simply put I had put my hopes on using the tracking software suite Boujou. I got a type of trial license at an expo a year back, remarkably enough it was still valid. After shooting some quick shots for tracking and wrestling with some of the known issues of Boujou, such as the constant fps changes across different steps of the camera solving process. I ran into a few problems with the scale and positioning of the world axis of the camera solve versus the internal solve of the particle system I'm using. 

After a few fail attempts I finally got the depth issue worked out. But a simple way of setting up the orientation of different elements is still beyond me. Clearly this is something that has to do with me not having used this kind of technique before. I've done some tracked shots with AE's internal tracker as well as Mocha. The shots that I'm going to be shooting however are going to need a tracker like Boujou.

Hopefully I'll actually make some progress beyond what I have now. I've made a small test shot, which looking back is a tad grainy thus making the track a rather big mess. It does however get the concept across however crudely as it might be at this stage.

The first of the two passes is obviously a failed attempt. Upon exporting the camera solve I was unaware of some of the orientation and size settings. Making the camera and the tracking nulls extremely narrow and close to each other. The second was a sad excuse for success. I am now however aware of some of the pitfalls and it should make things far more easier going forward.

Now it's a matter of doing some test shots with a decent setup and proper lighting to eliminate grain and issues with the shutter speed. Till then, this will have to do.

Harsh Lessons Learned

Today was supposed to be a day of simply painting maps. 

I however decided that I'd try to figure out a good way to generate a map zoom. I decided to simply rotate/scale/position the depth map's in a pre-composition. I however didn't anticipate this when setting up the grid when I started. Thus I had to generate a workaround which would proof difficult as the map would shift outside of the bounding box I had set up. Re-adjusting the grid to compensate would completely ruin the look so I had to set up a additional track matte to subtract everything outside of the bounding boxes. This however left me with a set of solid layers within my particle grid. I decided to take a extremely ghetto approach and generate a rotating and scaling mask to go along with the zoom. Tweening the mask didn't net me accurate results so I was forced to once again go in manually and fix it.

At this point I had wasted a fair amount of time, but had in fact also gotten it to work. To much dismay however I realized that due to how I generate the depth map's, the pre-composite process had thrown off my anchor points and all three maps were shifted over and through each other. This left me with two options.

• Manually re-align each map based on the street map.
• Roll back to a auto-save and just scrap the work for now.

 I've decided to simply revert back to an old auto-save project file and continue work from there. This is when I realized something I had overlooked when setting up and configuring my AE CS5.5 setup. While I had turned on the auto-save feature, I had forgotten to specify the increments properly. Which left me with auto-saved project files an hour apart. This carelessness set me back 3 hours of work as apposed to what I expected to be no more then a hour at most.

So today I painfully got reminded that one should lock and check their layer properties when pre-composing. As well as making sure you set up and check all your preferences when updating to a new version of your software suite. But again, these things happen and you've just got to roll with the punches.

I've started conceptualizing the UI on the side, I personally like the idea of a 3D mouse in a holographic panel. Direct physical feedback and smoothly tracking a hand and/or finger gesture will be interesting. I'll definitely have to make a small test as a proof of concept to see what looks and feels best. Something to maybe do tomorrow if I get bored with depth-map painting.

I have a list of things to do still, among them is figuring out what the smartest way would be to generate a accurate reflection. But I realize that most of the things I have on my mind at this point are just going to be educated guesses at best. Simply because the main asset isn't done yet and it all revolves around that.

I will leave you with a updated version of yesterday's video. Beyond this I will probably not post any more video's until this version of the map is completely populated.

For now however I'm going to call it a day.

-Niotex

Holographic Introduction

This is the first project I will be posting on which will hopefully become a habit and motivator of sorts. I've started work on the holographic representation of Downtown Los Angeles. It is to be tracked and incorporated in a 2/3 minute short I plan on shooting. Right now however I'm focusing on getting the main assets done. 

Initially the idea was to generate all the buildings from a set of solids. This however was quickly discarded for a more stylistic and efficient approach by generating geometry based on particle hight. This offered up a range of new possibilities for when it came to animating the hologram as well as give it some nice depth characteristics. A problem you run into with that, is when you're displacing particles in their base z-space with a depth map. The alpha range is hard to control over a spectrum of a building with only one floor to a downtown building with 50+ floors. If you want a nice resolution to work with you end up having to create independent depth maps for different geometry. As such I've created 3 independent maps for 3 different heights and layered those on top of each other in the final comp.

Thus far I have established the streets, highway and parking lots completely. I'm slowly generating more buildings along the edges and working my way to the center. A good chunk of the center piece skyline has also been completed and the grid is filling up nicely as you can see on the right here.

While I have started work on the initialization and boot up sequence, I have yet to come up with a clean and intuitive UI that is to be used in the short. I've been toying with the video of having holographic 3D track pad or two at the edge of the table. Or something like a holographic keyboard. I simply just don't want to resort to some kind unpractical yet strangely accurate hand gesture interface without a sensor in sight.

From a shooting standpoint I want to create a set of props and scout for a location which is something I'm hoping to do sometime in the next few days. I have some idea's as to what kind of setting I want but as of this moment I'm very open minded based on some of the locations that I can find. If a suitable location is found and it is possible I want to try and actually create a table for this to be tracked on. I might however have to scrap that if I come up short.

 

I have included a rough "initialization and boot-up" sequence along with a slight camera pan to show of what I currently have done. It is hard to say but one can assume that as of yet the map is only about 40% populated. I'm hoping to update and add to this daily and hopefully all the digital assets of this project will be done over the weekend.

For now however, it is time to call it a day.
-Niotex