Procedural generation, a misunderstood topic.

lysan

lysan
So we hear this from time to time. ProGen is boring, handcrafted is better. However we need to understand that the ProGen only is as good as the seed injected* to the code. *maybe a wrong expression

What is it that make the human eye catch the difference between a ProGen and a handcrafted object? Flaws! we look for flaws however not the kind of flaws we normally see, we look for patterns and imperfections.

Some of the most difficult objects to make in large numbers are organic objects.

[video=youtube;An-k7Mx3xVc]https://www.youtube.com/watch?v=An-k7Mx3xVc[/video]


When we see complex objects we normally say "my computer can't do this" however you're wrong, the beauty lays in the math and the equation used to create it.

[video=youtube;9HI8FerKr6Q]https://www.youtube.com/watch?v=9HI8FerKr6Q[/video]

That leads me to our little synthetic world, Elite Dangerous, the boundaries are not so much the hardware today, it's more directed towards how we or the designers created this world.

[video=youtube;Zkx1aKv2z8o]https://www.youtube.com/watch?v=Zkx1aKv2z8o[/video]

So with this last geeky video, you will see that math will make our world rich, if the math is applied in the creative way.

Someday I hope we will see the planets up close, not just with rocks but with vegetation and variety, or will we forever drift in space searching for life?

[video=youtube;zBDUglItruc]https://www.youtube.com/watch?v=zBDUglItruc[/video]
 

lysan

lysan
Lucius-Darcia said:
Very intresting and special +1
And i understand only half of that what the math-man says :D
Click to expand...

Well he is trying to explain how you can build wonderful things with math, however he is a geek and therefor a geek translator is needed.

In many ways we see lazy developers when we talk about ProGen, there are ways to use it to make huge areas populated with vegetation or other natural occurring forms. Where the handcrafted need to be utilized are man made structures, as they are very difficult to make unless they are build from modules.

In some of natures design you can see the mangelbroth equation, and that is quite funny as these vegetation are 100% natural.
 

Mikey2305

Mikey2305
Yes, i think FD are doing some wonderful and incredibly sophisticated things with their numbers. It both amuses and frustrates me to see this written off as 'lazy programming' often by people who have no idea what they are talking about. Mind you, neither do I!


@lysander... mandelbroth? Is that some kind of soup? ;-)
 
Last edited:

lysan

lysan
Mikey2305 said:
Yes, i think FD are doing some wonderful and incredibly sophisticated things with their numbers. It both amuses and frustrates me to see this written off as 'lazy programming' often by people who have no idea what they are talking about. Mind you, neither do I!


@lysander... mandelbroth? Is that some kind of soup? ;-)
Click to expand...

Never said FD was lazy, my point is and was that in the past many designers was lazy and used ProGen in a lazy way, I do recognize the good work done by FD. My whole point is that you CAN make vararity and complex game design by using math in a creative way.

mandelbrot, auto correct was not aware and keep changing it -_-

Phones and fat fingers and all. :D
 
Last edited:

Cheapchips

Cheapchips
I wonder what FDev are using for the space trees. I can't really tell from pics and vids if they are repeating, authored models or not. They certainly have enough variety at a glance.

They might not be using procgen at all for these, but it would be nice to think that this is the first steps towards in game flora.
 

lysan

lysan
Cheapchips said:
I wonder what FDev are using for the space trees. I can't really tell from pics and vids if they are repeating, authored models or not. They certainly have enough variety at a glance.

They might not be using procgen at all for these, but it would be nice to think that this is the first steps towards in game flora.
Click to expand...

Don't know as I haven't seen them up close, but FD should make it as in the first video, that would make more sense.
 

fennster

fennster
Looks to me as the space trees are not PG, from this screenshot anyhow....

repeat.jpg
 

[VR] Deggial

[VR] Deggial
Interesting post and I do (in general) agree with your love for PG! It is a powerful tool to inhabitate an almost unlimited space (be it "just" a planet or a whole galaxy)! Especially the second movie is a thing of wonder and beauty! :)

I also want to add this ineresting video to the discussion:

[video]https://www.ted.com/talks/michael_hansmeyer_building_unimaginable_shapes[/video]


However, I can undestand the caveats in respect to PG, too.

1) Self-similarity and the human brain

Nature follows procedural regeneration rules. As demonstrated in the first movie, an artificial scenery has to follow this approach, too, if it wants to look natural.

Unfortunately, the human brain is evolved to generalize. This is basically necessary to cope with the above fact; we simply wouldn't be able to handle an environment composed out of unlimited individuals. Hence it recognizes patterns and generates an abstract representation of them. "Cornifers" are recognized as such due to their typical shape. It doesn't matter, whether or not a single branch is missing or grows in a slightly different direction than the branches of the neighboring tree. These details are lost, all we see is "a conifer".

In consequence, even the most varied forrest might be percepted as many instances of "oh, just another tree!".

This happens even more, if there is no direct comparison available. Outposts are already very varied in ED. However, as the same base elements are used and there is no second outpost nearby, the variation in detail is lost and all we see is "another outpost".
(If one pays close attention, it is possible to recognize the differences, of course. I don't deny this!)

2) The difference between purpose and reason

There is no purpose in nature, only reason. As a natural scientist (a biologist), I find it incredibly fascinating to find out the reason why things are as they are. Why grow cactuses as they do? Why do some trees lose their leaves during winter and others don't? Which protection do have the later against water loss?
Knowing those reasons (or at least the fact that there are such reasons, even if I don't know them themselves) makes walking around in different biotopes so interesting.

This aspect is completely missing in procedurally generated computer sceneries.
Yes, there might be a diversity of stones, plants and animals. There migh be variation inbetween every instance of them. As mentioned above, all this is necesary to make the scene believable.
However, all of them were created by the will of a programmer/an artist. There is no inner reason, only artificial code that generates the outcome. Studying them will teach me something about the technical and aesthetical capabilities of the creator, but nothing more than this!

Dealing with purpose, on the other hand, is a completely different thing! There has to be a thinking mind behind purpose - a human mind (as far as we know for now). And learning about the purpose of things (real and virtual) can be an incredibly interesting thing to do! Archeological discoveries, studying the construction of an unknown machine, deconstructing a mythology ... all this is a worthwhile target for science.
However (again this word!), this kind of purpose can not be achieved by procedural generation. We need somebody to "invent" things, place them purposefully in the game world and link several correlations together.

TL;DR: In my opinion, both is needed: Hand-created elements, purposely placed in the universe as well as procedurally created and varied "fillers" (which are in no way less important, don't get me wrong here! Without the filling inbetween, everything just would fall apart!)



---

Jusr another note:

Pleeeease, for the love of RNGesus, replace "seed" in your OP with "algorithm".
Really, the seed is nothing! It is a mere number. 17746 is a seed. 8847629 is one, too. The seed in PG determines the distribution of patterns. How the patterns actually look, is entirely related to the underlying code.
This might be seen as persnicketiness. And maybe rightfully so. I still always cringe when I read It ...
 
Last edited:

2plex

2plex
Well, well... geeks unite!

The first 3 vids only reiterate dull, boring details for use in particularly single-minded brains.
The only one that is interesting in the OP's post is the 4th vid, because it involves a creative leap in content (aka. meta content).
Unless someone can clearly demonstrate that procedural generation itself is capable of that creative leap, I consider it a dull and boring mechanism, only useful when used as a tool by a creative artist.
And Elite Dangerous gameplay is a long, long way from that.
 
Last edited:
You must log in or register to reply here.
Back
Top Bottom