Cloud Based Components

Cloud Based Components

Seems like an odd title since components are those discrete gizmos on a circuit board.  Truth be told, the aggregate processing power of the cloud is not fully realized.  This cloud of potential is increasing with each passing moment.  Everyday, some individual, group, team, research facility, corporate entity, or government agency adds another bit to it.  It’s more than the data stored on it.  It’s more than the hardware it runs on.  It’s more than the software that supports it.  It’s more than the ideas of those, that operate it today, envision it to be.  It’s a promise.  It’s a resource for the future.

I had heard the term so many times in the last 10 years that it was mind numbing.  I really had no connection to it.  I just figured more disk space, faster processes, always available.  Along the way, some purveyors had troubles.  I welcomed the deflation to the hype.  Snake oil just doesn’t cut it for me.  When mystique is put on something as ordinary as a paper napkin, I find relief from hecklers shouting Bull Shit.

What finally opened it up for me was a cheap $40 camera that I would use as an action camera on my bike helmet.  The 720p quality video was impressive from a novelty item.  I enjoyed the ability to record rides and review segments for future rides.  The only problem I noticed was it was shaky or too dark during night recording.  Image stabilization came to mind.  I figured someone offered a video editor that corrects this problem.  I was able to find it in an evening.

vReveal from MotionDSP had what I was looking for.  The demos and descriptions I found looked impressive.  They kindly offered a demo for download so I could put it through the paces.  I downloaded it, installed it, ran it, and was blown away by what it could do.

I’m lucky because MotionDSP stopped selling the program not long after I purchased it.  This was the bridge to my understanding of what the cloud might offer.  It’s sad to see it gone, but I suspect that fears of nefarious use and the vendors target audience prompted the discontinuation.  Enabling a crook to use a cheap camera to get night vision surveillance is disconcerting.  It’s disheartening that bad behavior has led to limits placed on the general population. With that said, I’ll get back to my experiences with the post processing power of vReveal.

The vReveal software made use of a computer resource that I hadn’t paid much attention to, graphics processing units.  The GPU would be central to my cloud realization.  The software made heavy use of the CPU to analyze, calculate, and render the videos I fed it.  I didn’t have a video card that offered GPU support.  From what I gathered online, the process time would be a fraction if I had GPU support.  Some demos even showed real time processing results using beefy GPU systems.  I shelled out the cash and ordered a GPU video card specifically supported by vReveal.  When it arrived, I installed it and it proved to be worth it.

During my research for a GPU card, I looked into GPU’s and some of the things vendors were doing.  nVidia had just announced cloud based GPU at a conference.  This is when the fog cleared and I could get a glimps of the cloud.

One other item offered by MotionDSP was image mosaicing.  Stitching images together to create a map or a virtual environment was nothing new to me.  I had used stitching programs many years before working with vReveal.  It wasn’t something that sparked my interest at the time.  Now looking back, having a modeling available from video files is something worth looking into.  This is done to some extent and demonstrated by Flight Riot’s video on UAV photo mapping.

Now an entire new look at the material captured with the cheap $40 action camera is available.  By creating spatial objects for study, I could benefit more from my rides.  The real benefit from this is I can use the video files I already have.  The old becomes the new.

So how does this cloud become a component?  The answer is simple, it is the anti-component.  It reduces the need for bulky processing hardware in the field.  This simplifies the hardware, making it smaller, lighter, and more energy efficient.

Comments are closed.