Monthly Archive for March, 2009

From Gametrailers

It looks like there’s a new service in the works called OnLive that aims to push graphics rendering to the server side. Allow me to explain why this pseudoutopian bullshit is not going to work in clear and simple terms.

First off, the whole thing relies on compressed streaming video. Can you imagine playing your games with the vsual quality of YouTube videos? Not appealing, to say the least. Aesthetics, though, are the least of our worries.

OnLive needs a 5 Mbit connection to deliver 720p video at playable framerates, and you will always have a reaction delay in the video equal to the length of a network roundtrip.

Allow me to math it up. A typical well-tuned game gets about 60 frames per second. How long does it take to get the depression of the mouse button to the CPU? Almost none. It’s a short hop over USB right into the hardware registers. It’s negligible, and yet gamers buy special gaming mice and keyboards left and right because even at these extreme optimization levels, the delays are noticable.

Now let’s take that button click and translate it to a packet, fire it over the network, have OnLive’s server handle it, render the frame, and pump the image back to us over our broadband connection. Even if we had a massive OC-48 trunk running directly into our home network, we’d experience a staggering performance hit.

Basically, we could have the fastest connection possible and our framerate would still be decimated. Sixty FPS? Hah. More like twelve. It’s not a question of processing power, it’s a simple function of the laws of nature. Light has a fixed travel speed, and that means a network roundtrip will never be faster than client-side rendering.

Needless to say, this will simply be unacceptable to most. So the question remains, is there a market for it at all?

Well, real-time server-side rendering has its place, and that place is embedded devices. Most 3G-capable phones are better off streaming video in than trying to render it because of their small screen resolutions and limited hardware capabilities, and complete incapacity for upgrades.

Netbooks make a little sense for the same reasons.

Laptops are iffy at best.

Desktops are right out.

There are exceptions to these. I imagine there might be some market for Mac and Linux gamers who don’t wish to dual-boot. The problem with this is, well, most Mac users aren’t gamers anyway, or already have consoles, and most Linux users either aren’t gamers or wouldn’t like OnLive anyway because of the DRM (I am here.).

This is an age of massive decentralization. A company like OnLive is simply thinking in the past, and I predict failure.

I’m facing a bit of a conundrum with this whole game-making thing, and it is goddamn fundamental in nature. I have spent a significant amount of time putting together a 2D isometric game engine in pygame, and I’m quite happy with it so far, but the daunting task of the artwork is making me consider scrapping that work and making a 3D engine built on Panda3D.

On the one hand, 2D graphics (as evidenced by the existence of this website) is something of which I am capable. Unfortunately, doing even simple animations for an isometric game engine like the one I have written is utterly mind-numbing work. I have to draw every frame of every animation in five different orientations[1]. And for layered sprites like weapons, armor, etc., each layer has to be individually animated in this fashion. This multiplies very quickly. Consider a mob like an ogre or something. I need to draw images for it. So if this mob has four four-frame animations, that means I have to draw sprites for just that one mob[2]! It’s just too much damn work to draw all those little pictures.

On the other hand, 3D graphics imposes its own problems. I’ll have to completely redevelop the game engine, seemingly easy enough using something like Panda3D as a basis, but any engine I pick requires learning all the intricacies thereof, a daunting task in itself. Beyond that, I’ll need to do 3D art (which I am terrible at) and still have to draw textures for everything.

So what do I do? Do I keep on with the 2D engine and grind through all those little sprites, or do I scrap it and learn 3D art and animation from the ground up? I’m thinking I should just stick with the 2D engine. I’ve already put an awful lot of work into it to just throw it away, after all. But is all that work going to result in a good enough game? Are enough people still interested in 2D games to justify the work?

Maybe I’m going about this all wrong by trying to go it alone, but am I really likely to find a group of people who would like to collaborate on something like this, which is to say rather difficult work requiring specialized skills, all for no pay?

I’ll need to find answers to some of these questions soon. I’m so sick of spinning my wheels on this project.

[1]: Taking North to be vertically toward the top of the screen, these orientations are North, Northeast, East, Southeast, and South. Northwest, West, and Southwest are made by simply mirroring Northeast, East, and Southeast respectively.

[2]: When you consider the different sentient races, their animations, visible equipment, and variations; mobs, and static objects, we’re easily talking about thousands of sprites, perhaps tens of thousands. I’m getting a cramp in my hand just thinking about it.

This concept has been redesigned. To view the original, hit the “read more” link.

Continue reading ‘Savant’