I saw a post in the Alien Flux 1.3 thread stating that using OpenGL instead of Direct3D can be bad for sales.

I'm in the same position of princec - I do OpenGL (I don't have an OpenGL game yet, though). I used my other OpenGL stuff in a few (windows) machines with no problem.

Many of you make 3D games or at least games requiring a 3D accelerator. What approach have you taken? Do you all agree in that Direct3D is the way to go for Windows deployment?

BTW, why is Small Rockets hardly mentioned here? Really nice games...

Small Rockets really does seem more commericial, although they sell downloadable games.
They develop for console as well and have quite a large team.

Still worth a mention though...

We use OpenGL, mostly because we make our games for both Mac and Windows.

If a game sells poorly, it's usually caused by a number of different things.
While OpenGL might reduce your market (on Windows), it's only one factor.

I detect a large amount of FUD with respect to OpenGL and DirectX. OpenGL appears to have been used entirely successfully for a large number of commercial games, ergo, it is definitely a viable and large market. It is also the only way to do it on the Mac and Linux, and if you're looking at expanding your business, there's a lot of hungry people in the Mac and Linux world who want entertaining.

Having said that, DirectX support is the only thing that appears to work properly on Matrox cards, and the S3's support of GL is also flaky. But then again - not many people have Matrox and S3 cards, precisely because they're not very good.

Perhaps I should write a tiny stub application that users can use to test on their systems before committing to a download?

Cas :)

There were some threads about why OpenGL isn't as good to use as Direct3D when developing for Windows. You can search for them but here's one http://www.dexterity.com/forums/showthread.php?s=&threadid=412&

I've been a big OpenGL advocate in the past, but I'm coming around to D3D, for a couple reasons:

1) Linux sales don't matter, and Mac sales barely matter. Cross-platform in general is barely worth the effort.

2) OpenGL doesn't exist in a vacuum. You need support code for graphics (GDI or SDL or whatever), you still need sound, music, input and network support. DirectX offers all of those in a nice package, versus a bunch of unrelated libraries

3) OpenGL is lagging behind. That doesn't apply much to us as indies, since we're not typically doing cutting edge graphics, but DX does a better job of keeping up with the times.

These are just my own opinions of course.

If you've already got an app running with a GL back end I'm pretty sure it's really not that big a deal to support Direct3D as well. I think this is probably one of those jobs that sounds more difficult than it really is. Of course I'm assuming that the code you need to port is relatively sane when I say that.

When Alien Flux was released I was visiting my parents and I downloaded it. My fathers machine didn't have gl support. So I didn't play it... I was too lazy to bother installing new drivers. Major hassle.

You keep making the argument that some commercial games use GL therefor it shouldn't be an impediment to your sales. I think that argument is quite flawed. There's a big difference between buying quake3 or something for 80$ and then getting it home and learning you need to download a driver, vs downloading some little app off of download.com and learning you need to install a driver just to try the demo. In that case you don't have any expectations, and you'll just move onto another demo. So what if you don't play it? You've never even really heard of it. Why should you put in any effort to try it? In the Quake case you're pretty sure you've got Id software's latest title and you just payed 80$ for it, so you put in the effort and download a driver. Or often there's a driver on the CD.

I think you really need to look at your particular scenario and not so much look at major commercial projects as a guide for how to market or where to go next. In your case what you have is an excellent retro style arcade shooter. Your game is being sold as a "try before you buy" title. Your game is being exposed to audiences via the Internet. What your arguments are akin to is something like me saying.. well my latest demo is 120mb but so what? The half life 2 demo was 120mb ... that didn't stop THEM from selling. That's just not a reasonable way to approach sales and marketing. Your game is great, but it's also not quake3 or half life2 or blizzard's latest release. The retail market is a whole different ballgame, and if you ever manage to sell in that market then it may be reasonable to try and do the same sorts of things that those other companies do, but until then you need to do what works for the the various scenarios you're faced with. And to my mind that means exposing your game to the largest number of people possible. Making it as easy as possible, not only for them to play the game, but also to download and install it.

Originally posted by Siebharinn
Mac sales barely matter.
That's simply not true!

Mac sales barely matter
Tell that to PomPom.

The argument against OpenGL seems to be very much similar to the one about why one shouldn't target 1GHz systems when you're writing puzzle games.

What we've got here is a marketing problem rather than a technology problem. Rather than trying to sell, say, Alien Flux to your grandma and everyone in the whole world, it would be far more sensible to have it targeted towards the people that are a) very likely to be able to run it and b) very likely to want to buy it. In other words, to compensate for losing maybe 50% of my Windows customers through GL incompatibility (I'd be extremely surprised if over 50% of my potential customers couldn't run it), I'd need to double my conversion rate by making sure the right kind of people download it in the first place - which as we all know, is a marketing exercise.

Adding Direct3D support is unlikely to gain me many customers for this exact reason. The people who are likely to enjoy AF are also the kind of people that like to play video games in the first place. This tends to imply but does not guarantee that they have a GL-capable video card and otherwise reasonable system specs and that they are very much more likely to have OpenGL drivers for their systems or the willingness to go and get them.

Adding Direct3D support is likely to bring a few more customers, for sure - but you overestimate the power of my tiny brain to learn a completely alien and new API from scratch, and make a Java interface to it, and then abstract all my rendering calls to one or the other API, and then go through a huge amount of testing. I suspect in reality we are talking about several months of development which might net me 10% more sales. I think my time is better spent perfecting GL support and writing more titles on the technology front, and improving my conversion incentives and marketing on the business front, rather than trying to throw another bit of technology at the problem.

In fact if there's any technology I need sorted it's the Mac port, because from what I hear from PomPom they've had stunning conversion rates from Mac sales. In fact I seem to recall they had stunningly good sales on the Mac - about 20x more sales relative to the number of Macs than might be expected.

There's more to the marketing side than is immediately obvious too. Consider:

- shipping the latest drivers for all the major card / OS combinations on a CD and allowing the customer to order a CD with everything on, saving them a load of downloading

- putting links right next to your demo to the latest drivers for each GL card ventor

- guerilla techniques: "You'll need OpenGL drivers to run Doom 3! Save some time later and download them now!"

- Offering a free copy of the game for anyone who can't get it to work in exchange for a detailed system spec breakdown. Ok, grandma won't be interested but it's a great way to get system compatibility feedback from a lot of people for the "price" of a few free copies

Which are a few things I'm thinking about doing.


Cas :)

I'm still not convinced (about continuing with OpenGL).

princec, your argument makes some assumptions (I'd be extremely surprised if over 50% of my potential customers couldn't run it) which may or may not be true - and if they aren't, you're in trouble :(

I don't think porting a game to Direct3D could be that hard (but you're using Java, hehe :D). What troubles me is support and overall experience with the API. You and me could do the port easily, but the few versions would be more hacks than true ports (at least I'd end up implementing things in D3D the first way it does work, not necesarily the best way). It would take some time to get some D3D expertise.

And my bonus problem : I don't touch Windows in the building process :) If I switched to D3D I'll have to :(

I tried both openGL and Direct3D, openGL sells a little bit more for us ( but ddraw 2D is still way ahead )

Sadly, people are just not ready for 3d hardware accelerated games or it has to be a "gamer" game ( who has a top of the line pc... )

pat.

Well, after checking out the OpenGL vs. D3D thread linked to in this thread, I have a couple of things to say:

----

Myth: Windows XP does not support OpenGL out of the box.

Fact: I installed XP at home and, without installing any additional drivers, ran my OpenGL apps without any problems. In fact, it was faster in XP with the default drivers than it had been in '98 running with the latest drivers!

----

Myth: Even a GeForce card will not run OpenGL apps in XP.

Fact: The situation I described above was with a GeForce 2 GTS.

----

Myth: No versions of Windows have had built-in support for OpenGL.

Fact: Load up any Windows OS since '95 and you'll find that they come packaged with OpenGL screensavers! In fact, AFAIK, all the default Windows 3D screensavers (flower box, 3D text, 3D pipes, etc.) are all OpenGL. If OpenGL was so terribly supported for the Windows platform, why did Microsoft choose that as the API for the default screensavers?

----

Myth: Matrox video cards don't do OpenGL.

Fact: Matrox cards DO do OpenGL. I'm using one right now! Yes, their OpenGL implementation is a bit lacking, but I wouldn't let that stop me from using OpenGL. If you have access to a Matrox card, you can test your app on it and you may have to find a work-around or two. But overall, it's not that bad. And remember, we're talking about a SINGLE brand of video cards that aren't all that popular. If you're gunning for the mass-market, most people buy their computers from Dell, Best Buy, Bob's computers, etc.. How many of those do you think OEM Matrox cards with their machines? My guess is not all that many. So you shouldn't worry too much about the mass-market being riddled with Matrox cards.

----

Another comment I'd like to make is that alot of people seem to harp on this idea that you have to go and download OpenGL drivers for your card to do OpenGL. Well, I haven't had to explicitly download OpenGL drivers for a video card in years. But guess what? Unless your app uses an old-as-the-hills version of DX, your end-user WILL likely have to go download DX if they don't already have it installed. And if they're not an active gamer, it is very likely they won't.

Honestly, I'm surprised nobody has pointed out this fact that you need to download DX. If your potential customers are running Win 95, you can't count on them having anything newer than DX version 2 (!!!!). If you only want to consider customers running 98 or later, then you'd better not use anything newer than DX 5 unless you want them to have to download DX.

Considering OpenGL should work out-of-the-box on most setups (if perhaps at sub-optimal performance with old drivers -- which goes for DX too) so that most users won't have to download additional software, I'd say the tables are heavily weighted in OpenGL's favor.

This is all, however, assuming you don't use any OpenGL vendor-specific extensions. If you just use tried-and-true core API functions from v1.1, you shouldn't run into problems.

OpenGL availability has to be just as good as Direct3D. If someone can play Quake 3 on their computer, they have OpenGL.

That said, the drawback of OpenGL right now is performance. It takes a lot of manipulation of extensions to get opengl to work as well as d3d. OGL wasn't created with support for vertex buffers or index buffers, both of which are necessary for fast triangle throughput. OpenGL, the spec, now has vertex buffers, but a lot of drivers don't support the vb objects yet.

Of course OGL is a must have for cross platform though.

There was a ogl driver update system that came out about the same time quake3 came out called glupdate, but I think it died.

Fact: I installed XP at home and, without installing any additional drivers, ran my OpenGL apps without any problems. In fact, it was faster in XP with the default drivers than it had been in '98 running with the latest drivers!
I installed XP Pro and had no OpenGL out of the box. I've had a number of people with XP report the same experience to me. After I told them to upgrade their video drivers, everything worked (and no, I'm not using any special extensions that came out within the last 5 years...just plain OpenGL 1.1)

Fact: The situation I described above was with a GeForce 2 GTS.
Same as me. OpenGL didn't work out of the box


Fact: Load up any Windows OS since '95 and you'll find that they come packaged with OpenGL screensavers!
Not too many games are going to run very good with software emulation. So let me qualify my statements here and in that old thread. When I say "no OpenGL support", I really mean "no hardware OpenGL support". Yes, every version of windows comes with software OpenGL support, but unless you are making 3D pong you usually aren't going to find it's performance acceptable. Of course its possible that a game made for first generation 3D card might run in software mode fine on a 3GHz computer (I haven't tried it, but I suppose it's possible these days).

Fact: Matrox cards DO do OpenGL.
Depends on the card. Newer cards? Yes. Older cards (Mystique)? Only under NT4 (or maybe not at all...I can't remember). Similarly with other cards. If you know you are aiming for the higher end market, you can count on stuff like that. However, if you want compatibility with older systems, D3D is the only choice for many older cards. As I said in the linked thread, I've never seen a card that doesn't support D3D on every Win OS.

Unless your app uses an old-as-the-hills version of DX, your end-user WILL likely have to go download DX if they don't already have it installed.
The problem is that virtually every game uses DirectX in some manner. On the other hand, by comparison, only a handful of games support OpenGL. If the user has played ANY 3D game in the last couple of years, there is a very high chance they installed DirectX7 somewhere along the way.


Honestly, I know I probably come off sounding like a Direct3D shrill, but that's not the case. I really am a big fan of OpenGL. I've spent a lot of time as a major contributor in the OpenGL messages boards. I've helped tons of people working with OpenGL. I've written some well known tutorials on OpenGL per-pixel lighting. Three years ago I even tried to get a job with NVidia and didn't get it because their manager couldn't justify hiring more people for the OpenGL team and I had no interest working on the Direct3D team. I hope this makes clear I'm not the #1 D3D fanboy, but there are some facts I need to face. OpenGL support has been an uphill battle. iD Software took a stand and said "support OpenGL", and pretty much everyone gradually fell in line EXCEPT Microsoft. After all this time, they never really convinced Microsoft to take OpenGL any more seriously. In fact, Microsoft sat on the OpenGL ARB board for many years, but just recently the pulled out of that. If that doesn't convince you they have no good future intentions for OpenGL, I don't know what would. I personally would like to make a stand and do the same thing iD is doing and say "I demand OpenGL". However, I'm not in much of a position to make those demands, and my better business sense tells me my time would better be spent elsewhere. Thus, I've convinced myself that making a transition to Direct3D is in my best business interests. Depending on your target market, your mileage may vary, but I feel my opinions probably apply to many indie developers.

Same as me. OpenGL didn't work out of the box


Hmmm... Well, I'm at a loss to explain this. I've set up several systems with XP, and each and every one of them has run my OpenGL apps without a hitch and without installing any additional drivers. And with very good performance as well.

However, wouldn't you say that most people running XP who are novices and are not likely to have downloaded the latest drivers themselves are usually the same people who have XP simply because it came with thier computer when they bought it? And if that's the case, it should have come with sufficiently new video drivers. Does this sound like a reasonable hypothesis?

We choose GL, and as Cas stated, were glad we did.
Mac sales have been a real suprise for us. The fact that there are way more PC users, then Mac, means nothing. We've found that your avarage Mac user is MUCH more likely to be the kind of person to whip out the credit card and download our game.
Its so unbalanced that we could confidently say that half our sales go to Mac users.
Im not sure if the type of game makes any difference, but from what we've gathered, the Mac crowd sure like a good shoot-em-up!
So, ignore Open Gl, and you ignore Mac users. We would have sold way less games if we were DX/Pc only.

An almost useless discussion. But there ARE some factlets.

1) Mac sales for indie devs CAN matter. GG are making considerable sales on Mac versions of thier games (more than windows if I recall)

2) Its not THAT painful to do both OGL and DX versions of a game, but both offer different amounts of opportunity.

I'm used to OGL, so I'll ship with OGL support out of the box. Most of my customers will be experienced gamers, so I'm not scared of requiring that (for the right project, we're doing a 2D game for low end machines as well).

I dont think its useful to talk about older cards too much, because pre-dx3+ the whole thing was a nightmare. If we're going back THAT far, then you'd need a damn glide version!

I'd say if you can, support both dx and ogl, if not, support whichever you feel like. I doubt using either would seriously hurt sales on any game that actually USES 3D in any way seriously.

Fact is, if youre doing 3D, then for the user to even RUN the game it requires them to have drivers of some sort installed. Anyone who cant get dx or OGL running isnt likely to have the equipment required to run the game anyway!

lets face is, which modern driver set doesnt come with dx and ogl support?

Pointless excersize over though, do what thou wilt shall be the whole of the law!

.Z.

Originally posted by BongPig
Im not sure if the type of game makes any difference, but from what we've gathered, the Mac crowd sure like a good shoot-em-up!
...And they sure like our 3D (OpenGL) falling blocks game as well! :D

Just like Pom-Pom and Garage Games, we have more sales on Mac OS X than we have on Windows.

Let's face it, most indie efforts are poorly executed clones. In the Windows market, those clones are indistinguishable from all of the other clones. On the Mac market, with less software in general, the clones stand out a little more. I'm not suprised at the fact that some Mac software sells better than the Windows version. I wouldn't be excited about the Mac market, I'd wonder why it isn't selling on Windows. Probably because there are a hundred more just like it already.

If you took the effort needed to port a project to another platform (and it always takes effort) and applied that towards making your Windows version better, you'd see a *huge* difference between what you could get in the Mac market and the increase you'd see in the Windows market.

In short, make your software better instead of portable. If you can do both, then good for you. Most people can't, or at least don't seem willing to. I'm using Torque for my game, so I get cross-platform pretty much for free. I won't be happy until it sells well on Windows, only then would I consider a Mac port.

Originally posted by Punchey
However, wouldn't you say that most people running XP who are novices and are not likely to have downloaded the latest drivers themselves are usually the same people who have XP simply because it came with thier computer when they bought it? And if that's the case, it should have come with sufficiently new video drivers. Does this sound like a reasonable hypothesis?
It may sound like a reasonable hypothesis, but I think it's kind of irrelevent to the more important question of how will users KNOW they need to install video drivers. My wife could easily install Windows XP (Microsoft can't make it much simpler) or the latest Nvidia drivers (we have a cable modem...30 seconds downloading, double click, and reboot). The problem is, installing video drivers isn't something she knows about. So she installs XP, plays some Direct3D games, and low and behold they work. All of her games work. She's happy. She downloads a demo for a game. It doesn't work. As I think I said in the other thread, the first thing she would think of is not "hmmm...what's wrong with my computer". It would probably be like "hmmm...everything else works except this game....must be a piece of junk". That's what I would be more concerned about.

I find it interesting that you say hardware OpenGL worked right out of the box. That's contrary to my experience and several other people's experiences. I wish I knew what the difference was. However, my experience tells me that it's not going to work out of the box for everyone, so that's one of the points I'd choose to focus on.

My main point is just that I think a person who bothers to install their own OS but doesn't bother to get the latest drivers after installing it is going to be relatively few and far between.

Incedentally, on the topic of cards that don't support OpenGL well, I was thinking more about it and realized... if you're really USING 3D at all, your game is going to need a somewhat recent card anyway. Like a TNT2 or so. If you're running some odd-ball S3 or Trident, you won't be getting decent 3D performance of ANY kind anyway - regardless of the API used. So as far as card compatibility is concerned, I think you're not eliminating many more customers by choosing OpenGL (if any) than you are by simply choosing to do a 3D game in the first place.

Originally posted by Siebharinn
Let's face it, most indie efforts are poorly executed clones. In the Windows market, those clones are indistinguishable from all of the other clones.

Wait a minute... most retail games are expensive poorly executed clones. I thought that was part of the indie credo.

Ok, maybe its not necessary to support DX, but if you are only gonna support OpenGL, then it should work on all OpenGL compatible systems. As I stated in my first reply, I could not get the game to go past the menu screen without crashing, even after installing the most recent OpenGL drivers for my vid card. My card IS openGL compliant, and was intended for OpenGL support. It may not be the greatest OpenGL compatible card, but i've played more technically advanced OGL games than AF on it with little to no problems. This seems like a *HUGE* problem to me because I can't tell you how many slightly outdated ATI cards there are floating around.

Originally posted by Punchey
My main point is just that I think a person who bothers to install their own OS but doesn't bother to get the latest drivers after installing it is going to be relatively few and far between.
Of course this topic is mere speculation, so we can't say anything conclusive, but I would probably disagree with you. Remeber that tons of Windows upgrades are sold retail off the shelf. Personally, I've never upgraded (I prefer to wipe and start from scratch), but I suspect that when you do, you have to reinstall drivers for your video card. I may be totally offbase on this one, but that's what I would expect, especially since video cards often have different drivers for each OS or OS group. To be safe, I would expect windows to just go with it's defaults and leave it up to you to reinstall fully functional drivers.

Originally posted by Siebharinn

If you took the effort needed to port a project to another platform (and it always takes effort)

That's not necessarily true. While I haven't tried porting anything to MAC yet (like most people I don't have one), I was able to compile my code on Linux with relatively few changes (mostly VC++ 6 bugs related to ANSI compliance like a FOR loop variable being accessible out of scope). If you're using SDL and/or other cross platform libraries it may require almost no effort at all besides setting up/figuring out the new compiler.

This is just slightly off topic:
http://www.macworld.co.uk/news/top_news_item.cfm?NewsID=4522

Apparently Mac support for Directx... Ordinarily I wouldn't put this here as the cost for such a solution would be big, but their sales page makes it look like they are very flexible. It looks like it might be affordable even to indies, but I have no idea. Just thought people might be interested.

Wait a minute... most retail games are expensive poorly executed clones. I thought that was part of the indie credo.


If we were talking about retail games, you'd have a point. Since we're not...



That's not necessarily true. While I haven't tried porting anything to MAC yet (like most people I don't have one), I was able to compile my code on Linux with relatively few changes (mostly VC++ 6 bugs related to ANSI compliance like a FOR loop variable being accessible out of scope). If you're using SDL and/or other cross platform libraries it may require almost no effort at all besides setting up/figuring out the new compiler.


As I'm sure you're well aware, there is more to cross-platform than simply recompiling without errors.

Originally posted by Siebharinn

As I'm sure you're well aware, there is more to cross-platform than simply recompiling without errors.

Ok, it also ran fine. :)

Originally posted by Siebharinn
Let's face it, most indie efforts are poorly executed clones.

I'm using Torque for my game, so I get cross-platform pretty much for free. I won't be happy until it sells well on Windows, only then would I consider a Mac port.
Out of curiousity, how many copies/month of your future well executed totally unique game do you plan on selling? ;)

How many copies/month on Windows would it have to be before you consider a Mac port (which you say you have more or less for free, anyway)?

I just throw in my 5 Euro cents (we don't have smaller coins here in Finland, sorry :)):

1. Alien Flux does not work on my ATI FireGL 8800 workstation OpenGL card that is under one year old and is designed to run OpenGL applications. DirectX programs work, even if the DirectX drivers are just "add-ons". I would really love to play Alien Flux but am unable to and quite frustrated about it.. :)

2. I started working on SSM with SDL/OpenGL on Linux and soon found out that the porting issue was more than just moot. You still have to debug a separate version for both platforms. I had barely 5 years of OpenGL experience so I am unable to say how a more experienced programmer would have solved the issue.

3. Learning DirectX without any prior contact took about a week or two. The difference is not very big. Writing Java wrappers is no different than for OpenGL projects. YMMV but one week of exploratory programming should easily tell how difficult it is.

4. Writing more than one backend for your 3D rendering code should not be something surprising. I am about to start the integration of a FastGraph backend to support software rendering and older cards in addition to the DirectX 8 rendering engine I currently employ. I will probably add an OpenGL backend for the Mac market at some point.

5. My most commong support request is wrong driver, wrong version of DirectX or some crazy monitor problem that requires tweaking of screenmode frequencies. I don't know how many never bother to contact me, but quite a few.

Out of curiousity, how many copies/month of your future well executed totally unique game do you plan on selling?


Quite a lot, otherwise what's the point? If it doesn't sell a lot, I'll figure out why and adjust that. I'll keep adjusting until it sells at the level I want it to sell at.

Out of curiosity, how many copies/month of your current not-very-unique Tetris clone do you sell?




How many copies/month on Windows would it have to be before you consider a Mac port (which you say you have more or less for free, anyway)?

I hadn't really thought of it in terms of units per month. I have some goals for what I want out of the Windows market, and I would probably consider Mac when I was at or near those targets.

It's not that I dislike Mac, or have something against Mac users. It's a matter of economics. I'm going to put my effort into where I see the best potential return. I don't have a Mac, so I either need to buy one or work out a deal with someone who already has one. Not just compiling, but installers, testing and after sales support. A new Mac is going to be around $3000, probably a little more. If I sell my game for $30, that means I have to make at least 100 sales on the Mac side just to break even. If I work out a deal with someone, then my upfront costs are lower, but the net is lower too. Either way, it makes more sense (to me) to try and get more Windows sales, because the cost/effort to increase sales is much less.

Originally posted by Siebharinn
Out of curiosity, how many copies/month of your current not-very-unique Tetris clone do you sell?
I'm very pleased with both Mac and Windows sales.

Whether our latest game is unique or not, is a matter of personal taste.
You may call it a clone if you want to, I don't mind. However, if you knew your game history you would probably call it a Columns clone... :p

I'm not a Mac envangelist, I like both platforms. I just wanted to add that the Mac is a viable platform for indies to consider.

Anyway, good luck with your first game.