Jump to content

  • Log In with Google      Sign In   
  • Create Account


Member Since 23 Feb 2001
Offline Last Active Jul 21 2016 07:38 PM

Posts I've Made

In Topic: you all say oculus rift but why not google glass?

20 July 2016 - 01:31 AM

I don't know how the hololens screen projection system works. I suspect that they have three very thin layers of glass and each layer contains the red, green and blue channel and they rely on the additive properties of light to render a colored image.


Rendering an imagine is the first of many big challenges you'd face if you're going to make your own AR device. Remember that augmented reality is... augmenting reality! That means you have to know a bunch of things about reality before you can augment it! You need to know where things are in reality space, so that means you'll need to have two cameras that are constantly doing image capture, feeding the capture to a processor, which then tries to gather depth information to create a "z-buffer" for the real world. Why do you need this? For occlusion of course. If you have a ball in augmented reality and it rolls behind the couch, your device will have to know that the 3D position of the ball isn't visible anymore because it is being occluded by the couch. But, maybe a portion of the ball is still visible? So you'd have to do some fancy logic against each pixel in the ball mesh to see if its depth buffer test is less than or greater than the real world depth buffer. Then you've also got to map out the shape of the objects in the real world so that you can have collisions with it. Your ball should bounce off of the couch rather than phasing right through it, and in order for that to happen, you have to do some time processing the environment around you in order to generate triangulated collision meshes. 


This is computationally expensive stuff, so the "magic" of the hololens is that you're really wearing a stand alone computer on your head which has enough processing power and battery life to do this decently for a reasonable amount of time. It's really quite incredible if you think about how much computational power you'd need to do this, and they've shrunk it down to the size of a head band. You also have a microphone, so you can issue voice commands to the system which is pretty smart at interpreting them (I've never tried that feature).


Anyways, trying to compare hololens to oculus rift is like comparing boats vs cars and complaining that your boat can't drive on the highway, or your car doesn't float. They are fundamentally very different...

In Topic: Space for Unity refugees?

11 July 2016 - 02:42 PM

So just to fill in guys that aren't from the Unity board..


They used to run XenForo and it was really solid. They wanted to interconnect more site features so they licensed Lithium. They made the Xenforo forum read-only and fired up the Lithium version a few days later after migrating the databases.


Well unfortunately the migration was a complete disaster. Swathes of posts disappeared, no PM's migrated, no thread subscriptions migrated, the default layout is unanimously abhorred by the entire community, and lots of other problems/complaints. Essentially they went live with what appears to be a completely untested system after the first pass at data migration.


No one likes it, no one is happy and they're currently trying to communicate with Lithium but its all across the pond so time zones bork any kind of rapid scheduling and its taking forever to get solutions on what is involved with rollback etc while leadership is in SF, the forum team is in Coopenhagen and Lithium guys are somewhere else I think.


So while that basically leaves most users trying to avoid the forums now, I wouldn't expect loads of people to shift over to a generic forum with a single section for Unity because it will never compare to the quality, convenience, integration and 'officialness' of the legacy XenForo Unity forums. For myself, I just got a color inverter so I could read the text but all I post there for now is to update my asset threads.

This is fascinating to me. I love learning from the mistakes of others.

I would consider this to be a significant deployment and roll out failure due to insufficient testing. With the way you describe it, it sounds like their sysadmin just did a database migration, didn't check to make sure everything migrated perfectly, assumed it was all good to go, turned one system on and the other off, and felt the blowback when people lost data and capabilities. 

It sounds like the cross system compatibility capabilities they were looking for sounded like a solid business case which benefits them and customers, but they botched the execution of it. Now, they face a public support crisis and the project is tainted, so if they try again in the near future, people will bring their bad memories of the last failure and be opposed to the change. Fascinating. This is why you can't screw up your deployment and integration step. There's always going to be a subsequent user training step in the SDLC which generates user resistance, but deploying an 80% solution is going to multiply your end user resentment, no matter the validity of the business case.

The interesting question to follow up with is, "How could this have been done better?"
"Move fast and break things" is the wrong policy approach here. I think if they could migrate the data from one database to another, there is no reason they couldn't run two instances of the forum software. Let the new instance be publicly accessible. Let people vet it before you go live with it. Mirror the data from the production database to the staging database on a 24 hour cycle. Let your engineers figure out how to build the interconnected services. When everything is working smoothly, make the staged software the production version, and then set the current production version to a hot spare which gets its data synched from the new production instance. If there's an unexpected catastrophe, you have an immediate go back plan. I would set the deployment and integration cycle for this forum software roll out at something gradual like 6 months.

When it comes to user training, people like familiarity. Your forum UI is like a brand, and when you change a brand, you don't want to change it more than 10% from what it was previously. If you want people to adapt to your new systems, adapt the new systems to look and behave like the systems they're familiar with, and then gradually and incrementally, add additional capabilities.

Anyways, all of this is a heavy mix of policy and technical work working together. Surely a company as big as Unity, with the capabilities to build the worlds leading game engine, has the technical capabilities to handle this phased deployment of forum software.

In Topic: What music boosts your creativity

10 July 2016 - 04:26 PM

Epic Pirate Music

More pirate music

Fantasy Music

In Topic: Space for Unity refugees?

10 July 2016 - 04:18 PM

I've been to numerous events and listened to talks, and almost always, a unity employee is on a panel. They say the same thing: Their greatest asset is the community they've built. They stress this over and over, and how much their company values their community. They claim it is one of the things which set them apart from other engines. I have a hard time believing that Unity would suddenly decide to take its most prized possession and decide it's no longer important. Is it more likely that unity has put their company at risk by doing something which hurts their community? Or is it more likely that a few members of this million person community are unhappy about some new community guideline? Remember, Unity has millions of people using their software, and a portion of those are active unity community members, and of that percentage, there may be some who detract from the focus of the community (helping each other make games).


Anyways, I'm totally disconnected from the Unity community since I use the Unreal Engine. Maybe I'm totally wrong, but I'd need more substantial proof than just a few random internet posts by possibly disaffected people from that community.

Personally, I wouldn't mind if GameDev.net had a forum for Unreal and Unity, but my expectation is that I wouldn't get a lot of value out of the UE4 side. If I have a question about UE4, I post on the UE4 forum or answerhub. The chances are much higher that I'll get a timely and quality response, possibly from an epic employee. I see GameDev.net being more of a engine and language agnostic knowledge base, so if GameDev.net started to create forums for specific engines, there's a good chance that we'd have to create forums for CryEngine, LumberYard, Havok, etc, and I imagine a few of those would have very little activity. Anyways, that's my $0.02 and I don't run this site, so I leave this as a consideration for those who do and for the community here :)

In Topic: Free or low-cost resources that have helped you

08 July 2016 - 04:31 PM

I have realized that "Low Cost" is a relative term...


Can I get a usable asset on the asset store for $30? What if I can only use 25% of it? WHO CARES! GET IT!!! Why?
1) It's probably of a high enough quality because its on the asset store. I don't know about Unity, but UE4 assets are all vetted.

2) My time is worth money.

3) Paying an artist to create the same asset would cost about $2,000 (80 hours * $25/hour)


Even if someone puts a collection of assets up for $100, it's still "low cost" -- It's much cheaper than $2,000.


If the asset is 80% of what I need, that's also 80% of what I don't need to build. The artist can do the remaining 20%. Suppose the asset costed $100, and I pay my artist $25/hour, and the purchased asset is 80% of what's required, and my artist has to put in 2 days to tailor the asset to fit 100% of the requirements. The true cost of the asset was $100 + (16 * $25) = $500. This is still cheaper than hiring the artist to spend 80 hours to reproduce the same work ($2,000).

The way to look at the cost of an asset, is this:
Creating an asset from scratch costs $25/hour. For the asset price of $X, is $X + ($25 * TimeToFix) < $25 * EstimatedTimeToProduce? Then its more cost efficient to pay $X. Keep in mind that estimated time varies wildly, and $25 is also a variable which changes by required skill set value.

You only lose if TimeToFix >= EstimatedTimeToProduce.
So, "free" may make $X == 0, but that may also mean TimeToFix < 0. We'll call this "Free_TimeToFix". If you buy an asset, the "Pay_TimeToFix" value may be such that "Free_TimeToFix" > "Pay_TimeToFix", so spending money on an asset may actually be cheaper than getting a free one. And if you think "TimeToFix" == 0 for all assets you get from someone else, you are either delusional or have a very low quality bar.