Blog Archives

« Mappers and Designers de_dust_pcg: Source »

On Source map conversions (Jun 6, 2005)

The past week has been quite interesting in respect to the conversion of various HL1/CS1 maps to the HL2/CSS Source engine equivalents. Not only have the first official screenshots of the Valve cs_assault conversion been released, but a good batch of screenshots for the Black Mesa: Source conversion were posted too.

The difference in approach is worth a look. Specifically, the Assault remake is a reinterpretation of the original map, with many changes aimed at making the map better and more realistic. The Black Mesa Source approach has however been to maintain the original gameplay exactly, whilst making the world look better and more detailed than it was in HL1. Both approaches have the obvious benefits and drawbacks.

As a matter of interest, one of the first CS:S maps I tried to create was a conversion of Assault. As my first CS:S map I was fairly proud, but I made one fatal mistake: I stayed too close to the original map. Not only was the layout and scale exactly the same, but the geometry and structure followed the same basic rules. I did of course make use of the Source engine to make it look better, but not even Source can rectify a fundamental error.

The result was basically a higher-resolution, more detailed version of Assault, but with all the problems and issues of the original, plus some very ugly (thinking about it) aesthetic decisions I made. Embarrassingly, I got feedback from Cliffe and MacMan, highlighting spectacularly the mistakes I had made. Again, this all came down to me sticking too much to the original design, and trying to fit a new engine and new artwork to an old design.

It doesn't work. Trust me.

That's why the screenshots are of particular interest. First, Cliffe and Valve have done a tremendous job of completely redoing Assault, not least because trying to mutate an existing design like that is hard work. It's a challenge keeping all the important and fun gameplay elements whilst fixing the issues and not being sucked into creating a brush-for-brush replica, or something entirely different altogether. What's important though is that this is precisely the correct time to improve the map through radical changes. These opportunities don't come often.

On the other hand, I do wonder about the BM:S approach. Creating a conversion of an existing game is an admirable (and enormous) task, but trying to create a replica of the gameplay might be the crucial error here. If anything, this is the best opportunity to improve on HL1, and create new versions of the same basic missions, but actually making better use of the engine and not restricting it to the same basic structure as its father. It would perhaps be preferable to maintain the same concepts and basic layouts, but breathe life into them through sensible thinking and expansion rather than a new dash of paint (even if it does add more detail and depth, that isn't enough).

I suppose my fear is that conversions like these are inefficient: it is the chance to do so much more, so why limit your creation to the exact specifications of the original? A spanner should be thrown into the works, not just for the sake of being different, but to achieve something new and exciting. Something even more fun and immense and spectacular than before.

I am of course excitedly looking forward to the release of both Assault and BM:S, but I do wonder...

« Mappers and Designers de_dust_pcg: Source »

user comments

Atrocity at 19:18 on Jun 6, 2005

Very nice article, I like the screenshots of assualt and they look damn good, and of course the SS of BMS look cool, but I think they make the worlds look rather empty. I think they should make it look like people are living there, or high active within the area.

Atrocity at 19:21 on Jun 6, 2005

On a second thought, recreating a CSS map would, and could be very fun......

OB at 13:09 on Jun 7, 2005

It has been very interesting to see the way in which many of the old maps have been converted to Source. Generally, it seems as though Valve have kept within the spirit of the old classics, whereas the Turtle Rock conversions have taken things into whole new areas.

For me, the latter approach is the most impressive. Looking at a map like cs_italy (by Valve I think), it feels as though the bare minimum has been done to make it source worthy. I also get the same impression with de_chateau, de_piranaesi, and the dust maps. It looks as though cs_assault might be going the same way.

When I look at the Turtle Rock maps, such as de_train, de_prodigy, de_tides and de_inferno, I feel as though the designers have excelled themselves. Inferno in particular is quite simply, the best conversion of a map I have ever seen.

True, some of the Valve made maps are more "theme" orientated, and therefore, might have the boundaries that other maps do not. However, I can't help but feel that for my personal taste, conversions by Turtle Rock would produce the most impressive results.

Tom at 17:32 on Jun 7, 2005

I've always been amazed at what Valve do with the old CS maps. If you've seen Raising The Bar, there's a section on how they did it for the CS:S launch maps - took shots of the old versions, then produced imaginative watercolours of the same scenes enriched, then tried to make maps of that. It's a mad method, but it captures the spirit perfectly while giving itself enough creative freedom to stun visually. de_prodigy is one of the best-looking multiplayer maps I've ever seen.

Next to that kind of polish, I admit Black Mesa Source looks sparse. But what bits would you reinvent? Are you talking just visual stuff, or making areas bigger?

dave at 18:20 on Jun 7, 2005

Tom: the approach you mention (painting over map screenshots) is quite common, although I only first came across it when working on Nightfire. It's suprisingly effective! I was working on a Casino map, when they took a screenshot, sent it to an art guy, and gave me the resulting image which included things like changes of style, lighting and haze and things. Very very useful.

My main issue with the BMS approach is that the HL1 maps were designed around the engine. They were kept simple and basic because trying to make them look any more detailed would drawn attention to the engine limitations. So, HL1 has a lot of very weird indoor structures which you'll never ever find in the real world.

However, Source works the other way - maps only look good if they are believable and grounded in the basic rules of realistic structures. HL2 was amazing because the whole game fitted expectations of a real, livable world and adhered to the same rules. Getting HL1 maps to look as good requires considerable re-engineering to make them seem realistic. It would be entirely possible to maintain identical missions and gameplay, but within a more realistic setting that works better with the engine. I think that would be the preferable approach if I wanted HL1 to feel as alive as HL2 did.

boubou at 20:12 on Jun 7, 2005

Am I the only one who feels the converted CS:S maps are too dark? I liked how the originals were much brighter.

Tom at 14:05 on Jun 8, 2005

Good answer. Are there layouts for an underground lab that would be recognisably realistic to a player, though? I'm not sure I'd know a real one if I saw it, since I never have. Whereas with the exception of the Combine structures, Half-Life 2 recreated environments I have seen in some form before. I guess a lot of games go for underground labs and space stations precisely because we don't know what they should look like.

Perhaps the goal with an exotic setting should be to make it look like those scenes do in films, since they are (usually) actual constructions of some sort, even if they're fabricated for the task.

I remember the first map I ever made was a space station, and it had some 30 degree pointy corners in it. Someone commented that these looked odd and you wouldn't see them in real life, to which I said "This isn't real life, it's a space station in Duke Nukem 3D." He was right, though - I squared them off and they looked better.