Jump to content

  • Log In with Google      Sign In   
  • Create Account


Member Since 26 Feb 2007
Offline Last Active Yesterday, 11:46 PM

#5220980 Is it realistic to expect to make money in Unity Asset Store/UE4 Marketplace?

Posted by on 02 April 2015 - 01:04 PM

I think, really, that its like any other business -- except you have the "location, location, location!" part already answered.


Identify a need that isn't being adequately met, execute it fully to requisite standards (its not good enough to just be the best of bad options), price it attractively -- get good value for your own time invested, but remove all question in the potential customer's mind whether they could do it better, or cheaper, or both. I forget the name of it, but there's a control-mapping component that's very popular, and I think Unity actually acquired/invested in them. Here at Microsoft, we recently bought the company that created (and successfully sold) the UnityVS plugin (which integrates Unity with Visual Studio for scripting) so that we can give it away for free to help people make more and better games for Windows platforms using Unity. Those are examples of lucrative, core needs that someone decided to meet and made a good living at it.


Now, as a seller without a professional pedigree, part of identifying opportunities that aren't being met is also being honest about which ones you can fully and successfully execute on. Do only the things you can do well, don't pass yourself off as more capable than you really are. That's how bad reputations are born, and nothing kills person-to-person business like a bad reputation.


Another avenue, if you work on your own projects, is to monetize the components and assets you might have made for your own work, perhaps some time after you release your project so that there aren't a bunch of games that look like yours or have your systems.

#5220522 XBOX ONE still Big Endian ?

Posted by on 31 March 2015 - 11:55 AM

The x86 architecture has been little endian since inception ( afaik ) so it stands to reason that a SoC or device with a x86 architecture will be little endian..PowerPC off which the XBox 360 and the Cell Processor in the PS3 were based is big-endian..


Yes. In more general terms, the chips that are little endian are almost exclusively those that have a legacy in 8-bit chips, like the x86 familly. The reason for this is that 8-bit chips didn't have to think about endianness as it relates to sequential addresses of data types that spanned more than one byte. When the 8bit intel 8008/8080 led to the 16bit 8086, a programmer porting code from one to the other would still expect the low-byte to be at the low address (hence, little-endian) -- keep in mind that most software at the time was written in assembler, so you couldn't simply recompile the program.


Chips that came around in the post-8bit era and didn't have that 8bit legacy are almost exclusively big endian or bi endian. I think in the past there were some silicon efficiencies that made big endian more attractive, but those have surely vanished by now. Neither is really better, they're just different.

#5220260 From scratch vs Unity

Posted by on 30 March 2015 - 12:48 PM

Any engine worth it's salt allows you to implement new ways of drawing or organizing things... There's no reason you couldn't implement a minecraft voxel -> mesh generator inside Unity.


Someone did a video-blog-style presentation on this on YouTube. I think the guy spent a week or so.


There's essentially two reasons to disqualify Unity or other engine middle-ware:

  • Your project is complex and has performance or other requirements that require specific tuning and makes middleware engines a bad fit.
  • Your project is so simple or unusual that you'll spend more time fighting/working-around middleware than you would rolling your own.



I don't see the cost factor / licensing as a big issue. Its an annoyance I guess, but you are getting tons of value for the relative pittance you are paying for it. And its not just the engine, its the toolchain, support, pedigree and community too. Roll your own engine and you have to roll your own tools, provide your own support (which can be good and bad), don't have a history (e.g. you'll rediscover all the platform bugs Unity fixed years ago), and don't have a community of users to help you figure things out -- what's more, you don't have a community of people you can hire from who already know your in-house engine and tools. When you buy into Unity or other engine middleware, you're essentially time-sharing a very large, dedicated engine-team for a bargain-basement price (the difference, of course, is that so is everyone else, and so you don't get to dictate direction and don't get an engine that's specifically tuned to your needs and non-needs).

#5220255 The Atomic Man: Are lockless data structures REALLY worth learning about?

Posted by on 30 March 2015 - 12:32 PM

Just wanted to quickly give a +1 to what Hodgman mentioned in the second half of his post: you can often get better performance *and* have less bugs by removing the need for mutable shared resources.




If you're at the point now of not wanting or needing to deal with it, the actionable item you can do now with relatively little pain is to reduce any reliance you have on mutable shared resources where you can, and take note of where you haven't. This is good practice even in single-threaded code. Almost any time you have mutable shared state (at least one mutator and at least one other reader) you are implicitly serializing the code that references it, the more visible the state is, the more likely it becomes that a larger portion of your code is serialized in this way.


Also, as said above, you can also create "lock-free" code out of policy -- by designing your system such that operations that would otherwise have to be synchronized simply don't tread on the shared state at the same time. Double-buffering is a form of this, but more generally you can declare, for example, that all updates (mutations) of a certain kind are complete at a given point, so only (unsynchronized) reads will occur thereafter. The cheapest lock is the one you never have to take.


Its also always a balance between fine and course-grained locking, and the requirements that might be imposed by your data structures. For example, do you lock the whole structure and do a lot of work before unlocking it, or do you lock only part of the data-structure that's relevant to each unity of work? That depends on your use patterns, and whether the structure is tolerant of fine-grained updates (e.g. std::vector is not for any operation that could cause it to grow or otherwise re-alloc if someone else is holding an iterator), and how expensive the necessary lock is.

#5219155 The coin-op market: Where is it thriving, and what companies support it?

Posted by on 25 March 2015 - 02:44 PM

Probably the biggest difference today is that integrated graphics are now powerful enough that you can run even pretty graphically rich titles at 1080p on modest settings. If I weren't pushing the envelope too much, right now I'd wait for someone to release a small-form-factor PC (like a Gigabyte Brix) based on the coming AMD Carrizo APU (should be any week now), drop a 64-128GB mSATA or M.2 drive in it, with 16GB of fast DDR3 (fast stuff because the GPU benefits measurably) and call it good. You'd have something nearly as powerful as not terribly far behind the current-generation consoles, maybe half as powerful as Xbox One. If you needed a little more grunt, Zotac makes some small computers that have laptop-style discrete GPUs inside, those can easily match or surpass the current consoles.


Lots of the newer-style arcade machines are running on DVI/HDMI - style LCDs now, so you don't need to worry about the wonky refresh rates that the old arcade tubes used, and demanded specifically-tuned GPUs to drive them. If you opt for an authentic modern arcade LCD, this'll probably be the most expensive component, probably between $600 and $1200 depending on the size.


With a little more horse-power, or maybe for basic 2D games, you could probably even use one of those inexpensive Korean 34" 4k displays (about $400), though they'll only do 30hz at 4k and 60hz at 1080p.


Also in that time, ITX form factor has really been adopted widely, and you can build a very powerful system that way. My current gaming rig is an ITX boad with an i7 4770 CPU (water cooled even), 16GB RAM, and a Radeon 290x. The case is basically the same size as a 24-pack of soda cans.


On the complete opposite end, if your needs a really minimal, a 1080p display and a Raspeberry Pi 2 wouldn't be bad. Its got GPU that's somewhere between the original Xbox and the Xbox 360 (its capable of pushing games akin to the late Xbox/early Xbox 360, at 720p) and a quad-core 900Mhz ARM CPU with SIMD, and 1GB RAM for 35 bucks.

#5219148 Why didn't somebody tell me?

Posted by on 25 March 2015 - 02:19 PM

I once had a restaurant server ask if we'd like a "cara fe" of water for the table.

#5219145 I'm good at programming, is it enough?

Posted by on 25 March 2015 - 02:05 PM

I'll leave the attitude issue alone, but I'll also reinforce that you really do need to know how to work with others -- and I don't mean getting along with them, I mean working in a group setting collaboratively.


Firstly, you need be conversant in programming, which simply making some games doesn't prove on its own. It means when your mentor or lead says "Here's what I need you to do. I'd like you to use the Visitor Pattern." or "We don't use Singletons here." Those statements actually mean something to you, and aren't just giving you keywords to go look up. Extra research is fine when implementing things, of course, but you need to be able to talk about those kinds of things around a boardroom table in front of the studio head and not look like a fool.


Secondly, programming in a team is very different than programming solo. You need to know lots of little skills -- You need to know how to file a well-structured bug against another programmer for instance, how to use source-control software in general (Hopefully you are using one yourself already) but even with those basic skills, working in a repository with multiple people making changes and merging them together is much different than you making and merging changes all on your own. You also need to be able to deal with schedules, make good time estimates, and keep to them as well as you can -- when working solo and it takes longer to do something than you anticipated it sucks enough, but when you do that on a team you might be holding them all up too (protip: don't be the guy burning hundreds or thousands of your employer's money by holding everyone else up).


Thirdly is baseline skill set, knowledge base, and good coding style. These are the table-stakes of getting in, and if you don't have them, no portfolio will get you in no matter how good the results are. This basically means that you can talk intelligently about code and about design problems, that you have a good overview of contemporary and appropriate technologies, approaches, design patterns, and idioms -- that you can talk about them and recognize when to use them or not use them, and that you have a natural grasp of what good code looks and feels like -- and not just the code you write, but also the interfaces your code provides for others to use, and that should enable their code to look and feel good too. This also includes relevant areas of mathematics -- mostly linear algebra, and a smattering of quaternions, geometry, statistics, calculus.



If you've never collaborated on a project before, I suggest that's a good place to start. You can attempt to spin up your own project, but its probably easier to find a project that's looking for help and offer your services to them. Depending on what your skill-level actually is, you'll want to find a project that's at, or somewhat above your skill level if you can -- if your skills aren't entirely developed yet, you might find that some projects with a high-caliber team might not want your help, even for free. Just take that as something to aspire to, and a clear sign that you're probably not ready for an industry position anyways. Something at your level will let you develop the kinds of skills I talked about above, something a bit above your current level will do that too, as well as challenging you to grow your solo skills.

#5219137 8 bit sprite animation

Posted by on 25 March 2015 - 01:41 PM

If by "especially 8bit style" you mean low-resolution, low-color as on the NES or Master System then you definitely do not do that using a 3D package. Drawing each frame by hand is pretty much necessary, since a lot of the tricks artists used to make things look good are not something that any 3D graphics package I know of can do (e.g. use dithering intelligently, or apply colors artistically to achieve the intended effect. Also, palette-swaps (by which I mean actually designing them) are another thing that no package I'm aware of does. A computer probably could be programmed to do these things reasonably well, but no one has, AFAIK).


On the 16 bits, with a larger color palette to use and more colors per sprite you started to see things like Donkey Kong Country that used pre-rendered sprites, created in a 3D program and then exported with texture and lighting applied. This had a charm of its own, but it wasn't the same charm as hand-drawn sprites.



For creating hand-drawn sprites, my preferred program is Cosmigo ProMotion, which is a sort of spiritual successor to Deluxe Paint (which was popular with game artists all the way through the GBA and DS, despite being a DOS-based program). Many of those who moved off of Deluxe Paint earlier moved to ProMotion from what I understand.


ProMotion costs $80 bucks, but its money well-spent IMO. Graphics Gale is only $20, IIRC, but I don't like it as much.

#5218875 Pixel Art Sprites - An Overused Art Style?

Posted by on 24 March 2015 - 01:22 PM

Some of the early 3D games that are more stylized than realistic (like FF7) age pretty well, actually. The image above looks like an up-res screen from the PC version or maybe an emulator. It looks pretty good as is, and doesn't appear all that different than WOW, say. A higher resolution and better texture filtering can go a long ways all on their own. But it didn't look that good on the PS1, not by a long shot. That's acutally one of the key differences between upgrading a 2D game vs. a 3D one -- you can't really up-res a sprite game without changing it -- scaling algorithms like Scale2x can look good, but they can introduce some noise; 3D games can be scaled to any resolution and you're never less-sharp for it.


Early 3D games that tried to be realistic tend to be ones that don't age as well. Look at any of the early 3D sports games, for instance, or military shooters on PS1 or PS2.

#5218871 Is this a viable entry strategy?

Posted by on 24 March 2015 - 01:11 PM

Honestly, start building skills that are directly related and demonstrable by building a portfolio of games. Go learn Unity 5, or Unreal Engine 4, or Cocoas-2D, or GameKit, or MonoGame, or anything really -- as long as you can use it to develop and showcase your skill at making games, you'll be moving in the right direction.


That said, design positions are a rarefied role in an already rarefied field in an already rarefied industry, and one which is extremely competetive with no shortage of young brainiacs willing to work 50-60 hours weeks for relative pennies on the dollar. I don't mean to be discouraging, but that's the reality of the situation. You may not posses the health, or the youth, or the willingness to give up a significant amount of personal time to be competitive, and you might have life-circumstances (e.g. people to support, debt, etc) that would be serviceable with a typical entry-level games industry wage, or much less as an IT monkey.


The upside of being alive today, though, is that there are avenues for dedicated, talented people to make their own way and break into viable markets like Steam or Xbox Live, or the Playstation Store. That's no cake-walk either, and far from guaranteed (or even likely) success, but its a shot you can make something of if you put in the work and have the vision.

#5218868 Pixel Art Sprites - An Overused Art Style?

Posted by on 24 March 2015 - 12:54 PM

Its not really true that 2D art inherently ages better than 3D, its just that the stuff we remember fondly (or at all) was that which came from the heyday of 2D technology -- The SNES, GBA/DS, Genesis, and Neo-Geo or other arcade platforms. No one really thinks Atari or Intellivision 2D sprite art cuts the mustard anymore. In the same vein, when we think of "retro 3D" we think of the Playstation/Saturn/N64 and early PC titles -- which are the Ataris and Intellivisions of 3D graphics in the mainstream. We only just made it to comparably advanced 3D to the SNES' 2D with the last generation I'd say -- maybe even the current generation.


I haven't seen many 3D games that embrace an early-3D asthetic -- there was a Kickstarter recently for a modern throwback to Quake-era FPSes (with art and design tropes to match), and it actually looked really good. I can count the number of retro 3D games I've seen that don't just look cheap on one hand with room to spare -- though, I've not spent much time looking.



But 2D art is definately lower-fidelity than 3D, and that makes it significantly easier to produce. That's not to say that great 2D sprites are easy, but they're just more forgiving of imperfections, which means there's usually fewer iterations (note: this applies less to very large and high-color sprites, but I don't think we're meaning to talk about that here).

#5218592 what do you use to document your c++ code?

Posted by on 23 March 2015 - 02:33 PM

For C++, Doxygen is still king of the hill as far as reference documentation goes, and I believe the last time I looked into it they now support the Microsoft-style triple-slash XML comments in addition to their own annotations. If you're already familiar with those, stick with them.


I do wish there were a tool based on Clang's front-end though -- Doxygen maintains their own parser, AFAIK, and that seems redundant now, and I would assume that it also loses some deeper context that could be important for documentation. Clang would not.


Doxygen is also nice since it generates static pages. This means that you can distribute the docs in a .zip file, or serve them up from a very basic (and innexpensive) web server, or even serve them up right from a github repo (this feature is called Github Pages) if you're open source and on GitHub (EDIT: actually, I believe you can have a public GitHub Pages site attached to a private Repo, if you want to keep the source hidden, but distribute DLLS / Headers on a release package.



But since programmer documentation is what I do for a living, I can also say that reference is good, but its not enough. People want good reference, but they also assume its there and take it for granted. What they usually ask for is code samples, example projects, and help understanding the general patterns of using your API. They usually don't care about overview-type topics unless they're really high quality, an overview that's just average is usually seen as just getting in the way.

#5217759 How 3D engines were made in the 80's and 90's without using any graph...

Posted by on 19 March 2015 - 04:22 PM

Hey ya.


With this Link. one hell of a good book.


I actually never had that one, but Tricks of the 3D game programming Gurus is newer and also covers software rasterization; its really the successor to LaMothe's Black Art. Neither is especially state of the art software rasterization though -- you'd want to look at Nicolas Capen's 3D rasterizer (that went on to become SwiftShader), or the Larabee software rasterizer that Michael Abrash was working on -- not sure if code was ever released, but there were a couple whitepapers.


Also, required reading is Michael Abrash's Graphics Programming Black Book, I always wanted this one and finally stumbled across a copy last year. In particular, I believe this one covers how they did the small-triangle optimization in Quake, and also covers how they optimized for the Pentium by making sure both instruction pipes were working together (pentium was the first commodity super-scaler CPU -- it had two pipelines, the 'U' pipeline could do all integer instructions, and a second 'V' pipeline could simultaneously do simple integer instructions (like add/subtract, bitwise operations, etc) as long as there was no data dependency on what was in the U-pipe already) -- that sort of thinking is still relevant today, although processors are now so wide and compilers are so good that you probably don't have to think about it.

#5217689 Primary/Secondary/?

Posted by on 19 March 2015 - 12:38 PM

As an addendum to my alternate UI proposal, I might propose that any "unspent" research points remaining in the middle might represent and influence the pace of "ambient" research -- so if you had two points to spend, you could put them both into construction and all other research would progress slowly or not at all -- or you could leave them both "unspent" which would increase the pace of ambient research in all categories.


In that system, if we assume that the "active" pace costs 1 research point, then having 5 research points unspent is effectively the same as having them spent 1 to each category -- but the advantage would be that if you got a 6th point, with 5 unspent and one allocated to construction, then all categories would behave as base-line "active" with base-line "aggressive" in construction -- and the twist would be that active maybe costs 1 point, but aggressive research costs 3 (which would emulate diminishing returns of one category getting too far ahead of other research, which is often the case in the real world). So in effect, the "unspent" points would become a sort of base-stat, that could be sacrificed to gain advantage in one or a few categories, but your pace is always increasing so you can put your research discoveries on a curve to balance them.


Obviously this is a different design than what you have, and its balanced differently (e.g. in this system, with a discovery curve, you couldn't plow everything into 1 category and master that while being downright rudimentary in others, but that seems possible in your system.) so take it or leave it smile.png I'm imagining all kinds of good uses for what I just described, so I'm definitely keeping it for myself -- but feel free to use it too if you like.

#5217683 2D vs 3D

Posted by on 19 March 2015 - 12:16 PM

I plan on writing a blog post about how 2D games being easier is a misconception. I do agree in some areas 2D is easier, however 2D does have its own set of problem that are just simply easier to solve in 3D. Especially design and gameplay wise, the challenges you face with 2D are just purley different than the challenges in 3D


Its true of course that 2D has its own unique challenges that 3D does not have. But in general 3D has most of challenges that exist in 2D, and they're all complicated by an extra dimension and 3 extra degrees of freedom -- plus 3D games have their own set of unique challenges and it pretty significantly dwarfs the set of challenges in 2D, in my experience. Usually, you only run up against significant difficulties in 2D when you are emulating 3D in some way -- isometric engines with large and tall objects are famously less straight-forward in 2D than achieving the same effect in full 3D, and old-school arcade racers like Pole Position are odd ducks, even if they aren't all that complicated.


Design is surely different as you point out, but the OP is asking about graphics programming, not design, so I'm not sure that's relevant to the discussion. I'll be interested in reading your blog though, it sounds like an interesting topic and maybe you'll sway me.