View Full Version : HL2 and othe new engines.. an excuse for bad modelling?

08 August 2003, 10:03 PM
I've noticed in recent weeks/months a disturbing trend on CG and other game art fora. I see many many models starting to appear that are supposed to be 'realtime' models, low poly if you will, but that really have high counts, and no effort to optimise or design efficiently. The models are then flagged up as 'HL2' models or somesuch, because well, everyone knows that with these engines you can have 5000 polys a character, so why bother trying to use your polys optimally.

I do hope the trend doesnt become more prevalent. The higher polycounts of these engines dont mean you can abandon trying to make low poly models and just make every round thing out of a 24 sided cylinder! In fact i see these higher polycount engines as a great challenge, because given 5000 polys instead of 1000 you should be able to make something that REALLY hums by cleverly using those polys where they count.

High poly is high poly. For a tank model that means 100,000 polys and more. 5-6000 is nothing, and requires you to use the polys incredibly efficently if your results are to be spectacular (as they need to be if you are inserting them alongside the masterpieces that these modern games include)

Anyone else noticed this? or am I being a reactionary fool?


08 August 2003, 12:33 AM
No I notice this trend too, mainly on game sites in modeling forums bunch of kids. They don't understand the concept of optimization and it is quite ridiculous to see some character models for example that have a nurbed like body form almost too smooth and the excuse behind it is that "HL2 can support higher poly counts"


I even see som HL2 mods teams getting a very wrong impression of the capabilities of the source engine. For example I saw this one player model a mod team was working on, it was something like 15,000 triangles and they simply explained that...

" a year or two from now when our mod is released people will have stronger computers"

LOL, Nothing to worry about Baddog, let these kids find out their mistakes on their own. Somehow they think LOD technology means, they can totally uberize the poly count of a model since the engine can handle it.

Thats nice but now place 10 or more of those 15,000 poly players on screen at the same time on a 1.5ghz computer for example. They will get like 5 fps.

Just because technology is advancing, it does not mean computer users are keeping up to date. For holy sakes, some people still play HL1 on a 300 mhz computer and in software mode and this is 2003. An excuse that everyone will be having 5ghz, 3.5gigs ram, and 512mb video cards in 2006 doesn't mean the can just create super high poly models with the laziness to optimize as I'm sure some people in 2006 will only be maybe on a 1.5 ghz computer.

08 August 2003, 05:41 AM
That's because those posters are hip to what is already the standard. *cg-character artists, whatever the medium need to know how to do QUALITY hi rez if they every want to progress past the non-paying hobby stage, and move on to the gettin paid stage).

Whether or not they make correct use of the extra tris is irrelivent to the fact that they are *trying* (usually what follows is Learning, and Improving).

...seems to me like they're just honing their skills for the reality of realtime. (reality of realtime?)

08 August 2003, 05:44 AM
Originally posted by baddog
I've noticed in recent weeks/months a disturbing trend on CG

(also, not sure why a werd like "disturbing" would be used to preface this POV)

08 August 2003, 08:47 AM
I've read in a Q&A from the Valve crew that this Level of Detail Scaler won't actually take a high poly model and scale it truly dynamically depending on the distance from the player etc. Instead the modellers must make like 3 different models for up close, medium and far distance - and the engine will simply 'suddenly' change to a lower poly model when it moves a certain distance away. (or use lowest poly models all the time on low-end systems for example).

Edit: my point being that these kids will HAVE to optimise their models unless they want the ones with stupidly high poly counts shown _all_ the time.

08 August 2003, 11:52 AM
Originally posted by StormCyko
Edit: my point being that these kids will HAVE to optimise their models unless they want the ones with stupidly high poly counts shown _all_ the time.

I thought that went without saying. Did I miss something?

08 August 2003, 12:15 PM
@baddog you are so damn right, I sea so damn much wasted polygons, those guys could do more with 5000, i personalle prefer it a little more "edgy" but the more details, no one needs 2 subdivisions on planar polygons, the only thing that allows this "polywastage" is the lighting in the engine, if it looks better you can put more polygons into straight faces, but in most cases you don't have to...
so i totally agree :D

08 August 2003, 04:44 PM
It isn't just in gaming baddog, the trend is evident throughout the entire software industry. The computer is a vaccum, and like a gas content will expand to fill it.

Extra capabilities have always led to extra overhead. I started computing back in the 64K days, and guys did amazing things given the limited space and processor power. As processors, memory and storage increase the quality of content is very slow to keep up.

It can be blamed on sloppiness, but I think it is more human nature. When a limitation is removed we also lose the inventive, frugal, elegant solutions that working within the limitation required.

That's just my $.02


08 August 2003, 05:05 PM
thanks for all your comments.

I really do agree electrofirma, Ive seen it over the years in programming (I first programmed AND did game graphics in 48k total memory). Of course what this does is leaves a lovely hole in the business for someone who can extract every last ounce of possibility from the hardware *grin*


08 August 2003, 05:18 PM
Of course what this does is leaves a lovely hole in the business for someone who can extract every last ounce of possibility from the hardware.

No! It creates opportunity for those who are ready to start tapping the possibilties of the new hardware. There are still going to be those 'Oh, my God!' *jawdrop* titles. The run of the mill artists will still be creating yesterdays games on tomorrows hardware, while the real talent will be devoted to the things that just weren't possible before. :scream:

08 August 2003, 05:23 PM
ah that is exactly what I meant! a spece for those who can get the new hardware and instead of making a high polycount but inefficient tank that looks no better than the best of the previoud generation, instead will make a high polycount incredibly efficient and therefore amazing tank ;)

sorry if my wording wasnt clear


08 August 2003, 03:15 AM
baddog, could you hit me up with some links? The HL2 Engine and its modding capabilities has recently piqued my interest.

Thanks dood :)

CGTalk Moderation
01 January 2006, 01:00 AM
This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.