Guy Jeffries

View Original

About the Succession Maps

This post is a long and meandering story about the how’s and why’s of the Succession Map project. You have been warned.


Once Upon a Time…

Since the autumn of 2021 these maps have been cooking/rendering on my workstation overnight. I’m not sure how long a single map takes to cook from end to end as the majority of the cooking pushed the whole 500 maps forward stage by stage.

Like a supply chain, each cook got the maps to the next stage of production where they would be warehoused before the whole 500 got shipped to the next cook.

Cooking would pause every morning so work could proceed on the next phase of construction, racing to get as many features working before the relentless cook caught up. I was Gromit laying train track and the nightly cook was my personal penguin.

As you can imagine it becomes wearing to see 500 cooks crawl forward one by one relentlessly. Apart from the first few maps and random spot checks, 95% of these maps are as unknown to me as they are you. To save my sanity I injected some excitement into this project for the final push.

Taking the maps from 3d height fields to 2d images every map has been randomly assigned a batch number. Each batch is 10 maps, making 50 batches. In this way every map rendered to 2d images is a new discovery. It’s much more satisfying seeing new directories of maps appear out of sequence and the gaps between them being slowly filled.

When talking about cooking to produce data this is what professionals mean. As you can see we dress it up as techno-cooking but it relys on fire, a pot and a dry day.

Tools

The software tools used to produce these maps are Houdini 18.5 by SideFX and Gaea 1.2 by Quadspinner.

The landscape is shaped largely by Gaea and the process is automated by Houdini. Each map is the product of a mix of authored settings and a pair of seed numbers. I tried to use the seeds as much as possible, authoring parameters only when randomness would make a mess. Looking back I could have added more randomness, but like the wish-list, that will have to wait for version 2.0.

I’d also like to mention Stayawake. With all CPU cores at full capacity, memory filling to 40+gb then emptying to 8gb every 50 seconds, without a mouse wiggle Windows 10 wisely decides the workstation isn’t busy and should power save. Whether a lack of IT skills or Windows 10 being obstinate (like it is with updates) this little app was the only thing that stopped Windows dragging this project out to take a decade.

Here’s quick tour of the factory, mind you don’t step on any wires.

This is the 3d production part of the factory.

It starts by laying out starting data such as how many landscapes to make & seed numbers. This is also where the factory cleans itself ready for the next batch of work.

The first job was to make 500 base landscapes which actually look like sheets of untidy 3d static at this stage.

These bases are then split into 5 for weathering. Here the major forms for Primordial, Deadlands, 1st Age, Low Tide and Dry Mesa are eroded and carved. We now have landscapes, easy see, Everything else is just polish.

The newly finished Dry Mesa landscape goes through another round of weathering using the Primordial and 1st Age erosion to produce Reborn Mesa and 2nd Age.

These 7 landscape types now become 11 for adding water. Seas are filled, lakes and rivers are cut and snow is laid on the mountains.

The Ice Age gets 2 of these 11 as it needs enough snow to cover the land and another load of snow to not quite cover everything. Snow is always awkward to produce.

The Gaea water is currently everywhere so it gets some Houdini love to make it behave in a way that makes sense.

This is also where the volcanos are dropped in. Gaea then rains a tiny bit into the top of each one and I call this lava.

This is all of the 3d work finished and represents 90% of the cooking.

This is the second shed of the factory, as it were.

To start with this factory gets the same brief as the first and tidies itself up assuming it made a mess yesterday.

A quick decision on how to order the jobs and the 2d images are in production.

First all that useful 3d data is packaged into a monster set of usable masks. Until now file size has been small but these boys are over a gigabyte each. Ouch!

Here the mask data is passes to a side shed to prepare the 3d landscapes for their jouney into the 2d styles.

The full colour terrain gets some extra attention here. it’s cooked 4 times to create the full colour terrain. Its cooked 4 times as a single cook fills 160gb of memory and frankly thats just silly.

This tiny slice here is the magic where 3d becomes 2d.

Finally each of the 2d images are packaged into their final form with borders, copyright messages.

That little orange circle at the end is where the whole job gets passed off to make the teaser images.

I hope you enjoyed the tour. We’ll get into the indivual machines in a later post.

Early Development

The first versions of these maps were produced in 2020 during lockdown (in the UK) as an experiment to see how big an area I could make work in Gaea which looked like something worth cooking.

As I picked at this experiment between jobs the network in Gaea began to breakdown due to memory constraints. The biggest blow was that I could only cook a single landscape before needing to reset, it had become too complex. To continue I’d have needed to invest in yet more memory or a more expensive Gaea license so I could use it with Houdini.

Memory is boring and Houdini’s PDG looked exciting so I guess that was the decider. Armed with Houdini I could break down the massive Gaea network into sections so it could fit in memory. The bonus being I could make it even bigger!

This didn’t happen all at once. I had no reason apart from learning PDG to do this project so it went on the backburner for a few months. Making some Unreal landscapes for the Store was a possibility but after Unreal 5 dropped the timing seemed off.

As part of another project I discovered this collection of generators which led down a rabbit hole that ended in a plethora of quality apps that now exist for making maps for role playing games. I do love role playing games and pick up rules and campaign books fairly often but I don’t get to play all that much, being busy with projects like this one.

As part of the rabbit hole I got a hankering to make a role playing campaign, or at least start making one. I had a go with a couple of the map making apps but got frustrated pretty quickly. Going straight to map without a plan just made for a crappy map and planning was too much like my day job. I hunted about for pre-made maps and discovered that whenever I came across a good candidate someone had drawn roads and place names all over it. I couldn’t find a raw map to play with.

From this discovery the Succession project was born. I ran some tests on myself and my gaming peers and we found it really enjoyable making up world spanning campaign stories around the opportunities the geography offered (both alone and in groups). This type of creative process felt more like real history being made rather than a map being shaped around a story.

Early attempt at simply making a big place.

That RAM I just couldn’t get excited about

Fully 3d cliffs quickly went on the wish-list as the cooking overhead was more than my machine could cope with.

Is the Project a Success?

Archerland Dry Mesa epoch.

Making 500 maps was plucked out of the air really. The intention was to give a really good shakedown of the factory. Give it enough rope to throw up anomalies and also to see how long it took before the maps became blandly repetitive.

So far I’ve been pleasantly surprised by the variety in the maps. Even though from a distance they are all quite similar when you zoom in and start looking around the opportunities for world spanning stories are still stirring. The Dry Mesa map is a bit of a disappointment, I should have let that go a bit more. In context with its partners it works but on its own it’s becoming samey to me.

So far anomalies are not ground breaking. They subtly make the maps illogical but not in a way that makes them fail. I’m actually a bit disappointed by this as it shows I’ve been more conservative with the randomness than I intended.

Commercial success depends on whether these fill a niche in the fictional map market. And of course if I can communicate how they can be used and enjoyed.

If they prove popular enough to support continued development I’d love to do some more. Now the base system is built it can be bent to many different projects 2d and 3d.

If you haven’t done so already grab a copy of the free sample from DrivethruRPG and enjoy the map images in 4k.