Midgar Remake

  • Thread starter Thread starter Sapphire
  • Start date Start date
Status
Not open for further replies.
The main issue you will encounter is scope. We have thought about most of the issues you have mentioned, such as sector purpose zoning, *had a nice graphic we made, can't find it atm*.
Were you thinking of this:

MidgarPlateDemographics02.png


For anyone wondering, it's based off the Hoyt model modified to account for Midgar's layout. (In hindsight it doesn't make a whole lot of sense for industry to be concentrated on sector 7...)

Even at that we were limited in what was achieved, which was in part due to actually getting work done, due to limited numbers of actual developers. This would be a major hurdle for any project to overcome, and our Lead Dev Tempus was basically a one man army. Getting all the ideas concreted is one thing, but realising them is a whole other ball game. Just because there are ideas, expect yourself to get your hands dirty.
Yeah... looking back a lot of my development time was spent learning stuff on the side outside of school / work. Often times I'd think "ok, I'll just do this" and then realise that I need to learn a bunch of stuff before it can be done, haha.

...and our Lead Dev Tempus was basically a one man army
Not yet... I still need to learn to animate! (I've just finished learning how to rig - more or less)
 
If anyone remembers NWN's Aurora engine they had a very nice system for generating 'brushes' of city that you could add 'things' too. You will notice many games use 'fabricated' or built dwellings that the developer themselves construct. This is very labor intensive and in contemplation a bit of a waste.

There are ways around this. If you are familiar with a game called nethack it generates random levels for more than half the game. With specific features included for some levels. These features are STILL randomly placed. There are fixed levels etc.

What's my point? It's all a matter of perspective. I suggest using the computer for doing DULL boring stuff and you add things where it's needed. You don't need to create an enite world by hand. You do need to shape it. A lot of terrain generators have been used for moding games like oblivion. Without the use of a team of 200. Midgaar is huge city, so does it make sense to create EVERY street corner in it? As far as I can tell NO. How many dwellings are going to be similiar? (lots) There are going to be commonalities of archetecture with unique buildings thrown in.  Simply put do the statistics.

You have a city of say 5 - 15 million (midgaar) lets say 68% live below the plates and the 32% above. The above side is likely where most people work. The pizza plate design makes things sort of easier. However it is likely streets are ring structured (concentric rings). How big is each section? Just go through the scale etc.

Midgaar has already defined areas (sector 7 and 8 wall market) etc. Those can't be changed the rest is just noise (although you might want to look at Dirge of Cerberus) so the noise is what you are talking about. I would distribute housing toward the edge industrial then business as to grew toward the center. Business areas would be high class places to live and or work, mixed with food and shopping districts as well as military posts and government offices. It all has to fit in the center.

It doesn't matter 'what order' and you don't need to individually place each dwelling. I suggest brushing them in (as you aren't making this for walking in for now).

If you want it to be fleshed out, that adds a bit more complexity. Anyone use Farcry 2? It has an interesting engine for doing much the same things you are talking about. Tools exist tools can be created. If you want to peice together a large city I would suggest looking into standardizing dwellings sets street patterns etc. Pick and choose areas that need to be custom laid out (IE parts of sector 5 and 7 center of midgaar etc). Then use some of blenders features to brush in the rest (it will still be 3d). Granted it's a bit of work. However it will take you roughly 10 times the number of hours to create an entire city than creating a few choice sections and using the content you created those choice sections with over for the rest of the city in a 3d brush.

I've made entire cities using this technique using game modding software and tweaked the rest to make important areas.

It would be impractical to do otherwise. Of course this requires some programming skill, some knowledge of blender etc. I would estimate it would be about 3 months before you could start painting sections of the city. It also requires some very specific decisions and adhering to a standard format.

I don't know how blender handles thousands of similar objects. However that is the point at which you will be. You have silly details as in plan XY goes here (repeat 10000 times) this is why using automated generation tools that are parametric is actually a easier route in the end. Even if it is harder to handle to start with.

Something to consider in any case.
 
Notice I did not bother to use Crisis Core FF7 as an example. That was a dissapointmet for me when it came to how the different Secotrs of Midgar looked.
It's because they didn't use the original art assets from FFVII.
 
To continue the discussion about Midgar as a sandbox, I wanted to show the layout I defined with good old SimCity 3000.

http://imageshack.us/a/img16/7889/midgarsimcityplate.png

The map below shows the layout of residential, commercial, industrial, and "Government" buildings (the last includes, in Midgar's case, all the places run by Shinra and public locations such as parks, schools, etc.)


Obviously many buildings would exhibit similarities, especially on the Plate. In my opinion, the Slums are another story, where each Sector has a distinctive atmosphere. In the original game, Jessie even mentions that the Sectors used to be different villages which eventually merged into the big metropolis, before being renamed as "Sectors". Besides, one can quickly realize how the slums of Sectors 5, 6, 7 have their own specific "architectural" styles. Still within each Sector of the Slums, there would be "repeated" buildings. I am still ignorant about the methods for modelling populations of objects such as houses and how they can be distributed on a layout - and this aspect is probably software-dependant.

One other aspect I wanted to point out is the matter of scale. Even though Midgar is meant to be the gigantic dystopian metropolis, where a huge part of the worldwide population is concentrated, it is actually somewhat small. I doubt the Plate is more than 1 km in diametre, which would cap the city area to roughly 1.5 square km (combining the Plate and in the Slums). Let's compare it to a densely populated area of similar size in real life - say Monaco - and you can estimate Midgar's population to roughly 35000. It's actually slightly smaller than the size of the cities of the first Assassin's Creed. Still, it is considerably larger than any other town in this world (and a considerable piece of work to put together). 

As for the zoneing, I don't think the Hoyt model would hold in that case, because of the clear separation above/below the Plate, which creates clear separations such as "rich/poor", "fancy commercial zones/dodgy markets", "high tech industries/heavy polluting industries". For reference, I would also believe Sector 7 (on the Plate) to be a somewhat wealthy residential area (if one refers to the On the Way to a Smile novels).
 
@Tempus - thanks for finding that for me. I could not find it anywhere.

Well there are two trains of thought, one that you procedurally generate the city, which is fine if you weren't planning on making the city explorable, *is that even a word...*, in which case you would just model it out in low res, copy paste.

In the second case, which is something that Mayo Master is getting at is that you need to have a small enough scale. There are various renditions of midgar, scale wise, and I am not here to debate which source material is more correct nor argue with the purists when it comes to FF7. I am saying that in any game design needs to clearly set the boundaries.

@Cyberman - Most oblivion stuff is rarely done procedurally, at least not that I know of. If you are on about terrain placement via a heightmap, yes. But alot of the objects like trees, boulder etc are statically placed. There is runtime foliage alright, but that is just generated on top of the terrain.

For the Mako Dawn project we used the Fo3 engine, because we were fimiliar with the toolset and it had already transitioned a few times previous. Tempus'  revival of the project was unexpected;I thought it was going to remain in hibernation for longer. I am currently work on updating the format exporter for newer Blender versions, just in case I decide to restart the project myself, but I will probably move it to the Creation Engine.
 
Last edited:
Priced at either $500 or $4000. Hmm, I wonder who here would purchase this software?
Well if anyone was able to buy it that be a great addition lol.

But yes realistically it would be fun to see what people might be able to get out of it from the trial.
They do have a 30 day trial lol. Maybe i should have mentioned that.
Looks simple enough get started just probably, a lot of  tweaking to get things looking right.
 
@Cyberman - Most oblivion stuff is rarely done procedurally, at least not that I know of. If you are on about terrain placement via a heightmap, yes. But alot of the objects like trees, boulder etc are statically placed. There is runtime foliage alright, but that is just generated on top of the terrain.

For the Mako Dawn project we used the Fo3 engine, because we were fimiliar with the toolset and it had already transitioned a few times previous. Tempus'  revival of the project was unexpected;I thought it was going to remain in hibernation for longer. I am currently work on updating the format exporter for newer Blender versions, just in case I decide to restart the project myself, but I will probably move it to the Creation Engine.
I refered to NWN's Aurora toolset for doing that. It also is dependent on the target engine it ends in. I remember a 64x64 city scape was REALLY slow in NWN on my 192M machine circuit 2002 (LOL).

I'll try and keep this short. It's a standard set of algorythms. However it's not easy too use. PCB place and route is very similar. These are often used to place objects on 2 sides of a surface and sometimes on top of each other. In addition you have fixed placements. The big problem is the network that connects things (net list). For something like Midgaar that would be roads rails and walk ways. There are rules for that. The top of the Pizza is easier because it is a planned area. The slums are more difficult because they would be formed from the terrain they are on.

The aspects that have to be considered are: standardized top plate dwellings (not likely too have housing but condo style stuff), businesses etc. Along with transportation car, bus, and walkways. The rails are already fixed as far as I can tell (they handle most of the transport between upper and lower midgaar.
The bottom plate stuff is just a few sectors I believe 7 6 5 I don't remember where wall market is but 5 is where Areth house is right?

So what I am getting at is each 'area' to be generated needs it's own 'scale' standards (IE poles doors dogs cats spacing of housing etc). The pizza is planned (so it will be all standardized) the beneath sections shouldn't be too bad really either since they are kind of predefined.

Anyhow thats how the '3d brush' systems used in the Aurora Dev system in NWN worked. Creating new brush sets so to speak (see sim city scape suppled for example) requires a bit of knowledge, of what the end tool is. So simply put if anyone wanted to use said system it has to be planned. That requires coordination etc.

Hard to do on the internet I've noticed :D

Cyb
 
hmmm...
I leave the country for just a few days, and suddenly there is a whole different discussion going on here.... ;)
snadboxed environment, is interesting thought....
however using the way overpriced professional Esri software package..... is a bit too much I think.
There are other solutions for that, for 3ds max for instance there is GhostTown, which looks good and is able to do the same within max. For blender there are also plugins to do that.
However....Will it still be Midgar? Will it still be FF VII?
One could easily go wild on details, such as water system, transports, food supply chain, etc etc. before you know it, we are wiring up all the electricity of Midgar, is there cable TV in Midgar?  ::)

But... since the assets of FFVII are not logical created in that matter...

Personally, I try to get as close as possible to the original, although I do create assets in a reusable way, making segments, and piecing them together, like the way skyrim, fallout, and many other game assets are created. Looking at the details of the plates, a lot is reused, scaled up, down rotated and placed in a different spot. All these items can be called brushes for that matter.

For me a sandbox Midgar, would be nice, but not the primary goal.
 8-)
 
Hey Spoox, what do you think about my renders? Would you like some more realistic lighting in your images?
 
Well, if you can make some 3d field scene screens with the quality of the renders you showed, I think we'd all be ecstatic  :)
As you can see in the first sticky thread, we're in no shortage of job to be done, so I would invite you to pick a scene none else is doing. We'll be looking forward to having your modelling contributions!
 
Well I'm already working on some of my own mods so I don't have much time to dedicate on modeling. However, rendering already existing scenes wouldn't be too complicated for me. Once the settings are right, I could just walk my dog or something while my computer is rendering.
 
Oh. I didn't figure you were offering your services as "rendering farmer". Might be particularly useful for the people involved in remodelling Midgar.
 
Last edited:
Quick post, check this out. I have been working on it for quite some time now.
Pass it on to people! We are trying to get a lot of recognition on it.
 
Rev-amping the reactors again. No more floating geometry! Except the english 05/xx/xx/xx etc.... I'll fix that later.

2hzk93a.jpg


Midgar is simply too large for a sandbox creation, although.... Noen, Killerx20 and I have successfully gotten our Midgars in the CryEngine 3, yes... It was very fun to walk around :P.
 
Rev-amping the reactors again. No more floating geometry! Except the english 05/xx/xx/xx etc.... I'll fix that later.

2hzk93a.jpg


Midgar is simply too large for a sandbox creation, although.... Noen, Killerx20 and I have successfully gotten our Midgars in the CryEngine 3, yes... It was very fun to walk around :P.
It may be large, but if it was fully recreated in sandbox it would be beyond epic. Let me reprhase that.. It would be beyond EPICCCCC!!!
BTW nice job on the reactors, you have come a long way.
 
A pretty render.......

Reactors are now fully updated! Not one bit of floating mesh and all labeled and numbered correctly.

mbq2kg.jpg
 
Status
Not open for further replies.
Back
Top