Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[PatrikLundell]

GoblinCookie doesn't use the term "tree farm" to describe the farming of trees, but some kind of tree placement method apparently familiar to everyone who knows anything about the issue, and thus not warranting any further description.

There are various methods to adjust the tree density to match one appropriate for the biome. A simple "maximum 0.x trees per tile" has the effect that cutting down trees in parts of the biome will regrow them randomly on all free tiles, i.e. only a modest reduction of growth in the untouched area. This means you can "move" all the trees into one or more zones over time. It also does not provide much of a growth slow down as the biome starts to fill up (the tiles occupied by trees will of course not support saplings, so there is a slight reduction). Finally, it doesn't provide any method to encourage spacing between trees. Although I still believe my original proposal is better, it would still be better than the current situation.
I had hoped the proposal would generate a discussion about how to address the issue, but unfortunately it's mostly bogged down into claims and counter claims about how it currently behaves, including a discussion of whether "treesplosions" are real.

The embark I'm currently playing is probably a jungle (I've acquired a copy of an embark). At the 1 year mark I clear cut the embark, and at year 8 it's still possible to reach the trade depot (through other means than following the tree growth protection stockpiles surrounding my building), but I'd expect most other routes to be impassable around the 15-20 year mark (although the FPS has probably killed the embark before that). I've also seen no indication of any kill zones, with lots of L shaped groups of three trees, and other formations. No box of 4 trees yet, though.

[Button]

A simple "maximum 0.x trees per tile" has the effect that cutting down trees in parts of the biome will regrow them randomly on all free tiles, i.e. only a modest reduction of growth in the untouched area. This means you can "move" all the trees into one or more zones over time.

I think you're misreading the suggestion (Either that, or I'm misreading the suggestion and suggesting something else ). An "embark tile" is one square on the embark screen, which corresponds to a 16x16 48x48 area after embark. So if you choose a 2x2 embark, in a biome with a cap of 4 per embark tile, your embark could support a maximum of 16 trees, with each quadrant of the embark supporting a maximum of 4 trees. Clear-cutting the northwest quadrant would only reduce the tree count below the cap in that quadrant, so trees would only regrow in that quadrant.

Edited to correct the number of mini-tiles per embark tile.

[NullForceOmega]

Embark tiles are 48x48, which are further divided into 16x16 'sub-tiles'.

[Dozebôm Lolumzalìs]

GoblinCookie: Why do you insist that savannas are supposed to look like they already do? I was going to write a huge refutation of your arguments, but I saw it all boiled down to this:

A savanna is supposed to have very few trees, and definitely less than a forest.

Done. Either disprove this statement (ha) or show that a forest and a savanna are significantly different over a few decades. This statement doesn't refute all of your argument (the "devs didn't want all the tree simulation load to occur in one 'instant'" argument, for instance) but it does most. We are looking not at intention, not at what was wanted, but at what is actually happening.

But if you continue to say that a savanna looks perfectly fine with one tree every few tiles, I'm out. I can't argue with somebody like that.

[PatrikLundell]

@Button: Yes, I misunderstood the description, but the end effect is essentially the same. I thought the description meant per in embark tile of the corresponding biome, i.e. for instance 1000 tiles of biome X and 1000 of biome Y, and the remainder being non tree supporting for a 3*3 embark. Biome X would then have a tree cap of e.g. 0.01 (=> 10 trees) while biome Y would have e.g. 0.05.

My arguments remain the same for embark tiles but on a lower scale, but if you reduce it to sub tiles then the areas within you can shift trees get very limited. It also results in a rather step wise tree density control at low densities, Still, it would work better than the current situation.

@NullForceOmega: Are the sub tiles really controlling biome boundaries, i.e. are boundaries straight lines? My impression is that biome boundaries are more nuanced and curving, so sub tiles might not take part in the biome division?

[GoblinCookie]

GoblinCookie doesn't use the term "tree farm" to describe the farming of trees, but some kind of tree placement method apparently familiar to everyone who knows anything about the issue, and thus not warranting any further description.

I mean the solution by which you use a low-memory/processing power script to place a whole set of trees in an arrangement that is basically identical and

There are various methods to adjust the tree density to match one appropriate for the biome. A simple "maximum 0.x trees per tile" has the effect that cutting down trees in parts of the biome will regrow them randomly on all free tiles, i.e. only a modest reduction of growth in the untouched area. This means you can "move" all the trees into one or more zones over time. It also does not provide much of a growth slow down as the biome starts to fill up (the tiles occupied by trees will of course not support saplings, so there is a slight reduction). Finally, it doesn't provide any method to encourage spacing between trees. Although I still believe my original proposal is better, it would still be better than the current situation.
I had hoped the proposal would generate a discussion about how to address the issue, but unfortunately it's mostly bogged down into claims and counter claims about how it currently behaves, including a discussion of whether "treesplosions" are real.

The embark I'm currently playing is probably a jungle (I've acquired a copy of an embark). At the 1 year mark I clear cut the embark, and at year 8 it's still possible to reach the trade depot (through other means than following the tree growth protection stockpiles surrounding my building), but I'd expect most other routes to be impassable around the 15-20 year mark (although the FPS has probably killed the embark before that). I've also seen no indication of any kill zones, with lots of L shaped groups of three trees, and other formations. No box of 4 trees yet, though.

You clear cut the embark, hence your tree density problem.  If you cut all the trees down at the same time the saplings pop up everywhere at once, meaning that the trees end up very close together since it is only the grown-up trees that control the growth of new trees within the forested zone of the map.  Cutting down all the trees is likely to result in an ultimate increase in the density of the forest since you basically eliminate virtually any spacing as you need there to be large grown-up trees in order to keep new trees from going up.

GoblinCookie: Why do you insist that savannas are supposed to look like they already do? I was going to write a huge refutation of your arguments, but I saw it all boiled down to this:

A savanna is supposed to have very few trees, and definitely less than a forest.

Done. Either disprove this statement (ha) or show that a forest and a savanna are significantly different over a few decades. This statement doesn't refute all of your argument (the "devs didn't want all the tree simulation load to occur in one 'instant'" argument, for instance) but it does most. We are looking not at intention, not at what was wanted, but at what is actually happening.

But if you continue to say that a savanna looks perfectly fine with one tree every few tiles, I'm out. I can't argue with somebody like that.

There is plenty of evidence to support the conclusion that the savanna ends up with less trees than a forest, Dirst's 4yr savanna map has far fewer trees on it than my 7yr forest map does; while there are many more trees than there were to start with his map has quite impressive treeless zones while mine does not have treeless zones of anything like the same size.  Maybe another 3 years will make the difference but I doubt it. 

It is not a question of what they wanted, it is a question of what was possible.  The devs appear to have gone for an intial tree script that places the initial trees on the map in a semi-randomised pattern and a seperate growth mechanic that then grows new trees from scratch and grows the originally placed trees so that they are larger than they originally were.  The question of why they did not simply use the same mechanics to grow the trees as to place them is fairly easily answered; too much memory/processing power was needed to do the calculations needed to world-gen trees using such a model. 

Basically though, this discussion is quite fruitless since there is so little evidence on offer.  We should all go and play the game for 20yrs and then come back with screenshots of what a forest and a savanna look like after all that time.  Without more evidence I go with what is rational however, it is not reasonable to believe that the devs would have gone through so much work developing 3D trees and then placed no limitations on tree-growth depending upon the biome, so I believe that such limitations will exist.  It is also quite reasonable to believe that the tree-splosion idea is hearsay resulting from people not understanding why there is a discrepancy between the two tree placement models in play rather than actually being grounded in solid evidence. 

[TheBiggerFish]

Well, you're arguing for it, please do actually gather the evidence...

[PatrikLundell]

Quote from: PatrikLundell on November 28, 2015, 12:34:07 pm

    There are various methods to adjust the tree density to match one appropriate for the biome. A simple "maximum 0.x trees per tile" has the effect that cutting down trees in parts of the biome will regrow them randomly on all free tiles, i.e. only a modest reduction of growth in the untouched area. This means you can "move" all the trees into one or more zones over time. It also does not provide much of a growth slow down as the biome starts to fill up (the tiles occupied by trees will of course not support saplings, so there is a slight reduction). Finally, it doesn't provide any method to encourage spacing between trees. Although I still believe my original proposal is better, it would still be better than the current situation.
    I had hoped the proposal would generate a discussion about how to address the issue, but unfortunately it's mostly bogged down into claims and counter claims about how it currently behaves, including a discussion of whether "treesplosions" are real.

    The embark I'm currently playing is probably a jungle (I've acquired a copy of an embark). At the 1 year mark I clear cut the embark, and at year 8 it's still possible to reach the trade depot (through other means than following the tree growth protection stockpiles surrounding my building), but I'd expect most other routes to be impassable around the 15-20 year mark (although the FPS has probably killed the embark before that). I've also seen no indication of any kill zones, with lots of L shaped groups of three trees, and other formations. No box of 4 trees yet, though.


You clear cut the embark, hence your tree density problem.  If you cut all the trees down at the same time the saplings pop up everywhere at once, meaning that the trees end up very close together since it is only the grown-up trees that control the growth of new trees within the forested zone of the map.  Cutting down all the trees is likely to result in an ultimate increase in the density of the forest since you basically eliminate virtually any spacing as you need there to be large grown-up trees in order to keep new trees from going up.

Firstly, saplings do not pop up at once, the appear gradually over time (and probably completely independent of whether there are trees or not, except, of course, on the tree tiles themselves), so they should mature gradually as well. Secondly, the number of L shapes have gradually increased over the years, with the later ones appearing significantly after the original tree and the pairing tree, and I haven't cut down any trees since the original race, except to set up no grow zones around my fortress and the map edge, and the trees cut down at that time were those actually in those zones.

I've got one of these as well
TTT
  T
  .
  T where the missing tile contained a herb that's since died, but no sapling yet.

Edit: The embark is not a jungle, but tropical savanna or shrubland, based on the animals present.

[Dozebôm Lolumzalìs]

You believe? This is based on belief now? Let's, I don't know, ASK THE ONE PERSON WHO CAN KNOW???

[cochramd]

There is plenty of evidence to support the conclusion that the savanna ends up with less trees than a forest, Dirst's 4yr savanna map has far fewer trees on it than my 7yr forest map does; while there are many more trees than there were to start with his map has quite impressive treeless zones while mine does not have treeless zones of anything like the same size.  Maybe another 3 years will make the difference but I doubt it.

What your 7 year forest map looks like compared to someone else's 4 year savannah map is completely irrelevant to this discussion. The issue at hand is that after a few years a map of a tree-sparse biome will have become very tree-dense. Vattic has already provided evidence that this phenomena occurs; I have not seen any evidence suggesting the contrary.

Without more evidence I go with what is rational however, it is not reasonable to believe that the devs would have gone through so much work developing 3D trees and then placed no limitations on tree-growth depending upon the biome

Why not? The devs are human and capable of making mistakes, and Dwarf Fortress isn't even complete yet. Who are you to say that this isn't something Toady either overlooked or knows about and intends to fix later? You have zero proof of the dev's intent.

It is also quite reasonable to believe that the tree-splosion idea is hearsay resulting from people not understanding why there is a discrepancy between the two tree placement models in play rather than actually being grounded in solid evidence.

Except we already have solid evidence.

Edit: Decided to let a fort run with the fps uncapped to get some quick comparison shots. The embark is a "sparse" savannah.

Spoiler: year 5 (embark) (click to show/hide)

Spoiler: year 8 (summer) (click to show/hide)

Spoiler: simplified comparison (click to show/hide)

[GoblinCookie]

What your 7 year forest map looks like compared to someone else's 4 year savannah map is completely irrelevant to this discussion. The issue at hand is that after a few years a map of a tree-sparse biome will have become very tree-dense. Vattic has already provided evidence that this phenomena occurs; I have not seen any evidence suggesting the contrary.

Since the trees that are placed originally themselves bear no resemblence to what the trees ought to look like having had hundreds of years to grow, it is pretty much irrelavant that the total number of trees that we end up with on a savanna is greater than we started off with, since the trees are also a lot bigger than they started as well.  Fact is that the original trees are not grown using the model that is used after embark, they are placed using a relatively crude script. 

Since the total number of trees that existed on embark is *not* the number of trees that ought to be there given the mechanics actually in play in the game, it is far harder to prove there is a problem that simply pointing a finger at the number of trees we end up with at any given biome and pointing out that there were more that we started out with.  In order to prove there is a problem we either have to prove there is actually no limit at all on tree growth or that the limit is exactly the same on all biomes. 

Why not? The devs are human and capable of making mistakes, and Dwarf Fortress isn't even complete yet. Who are you to say that this isn't something Toady either overlooked or knows about and intends to fix later? You have zero proof of the dev's intent.

Neither do you.  Nor do I need any proof, since the conclusion follows from reasonable premises; it is reasonable to assume that the devs do not want uncontrolled tree growth on all biomes, hence if they are developing a model for tree growth they will have implemented some limitations on said tree growth.  You can of course prove me wrong by actually demonstrating that this is not the case, but the devs are stupid is not a reasonable premise to draw conclusions from. 

Except we already have solid evidence.

Edit: Decided to let a fort run with the fps uncapped to get some quick comparison shots. The embark is a "sparse" savannah.

Spoiler: year 5 (embark) (click to show/hide)

Spoiler: year 8 (summer) (click to show/hide)

Spoiler: simplified comparison (click to show/hide)

We have solid evidence that the number of trees we end up with is greater than the number of trees we start off with; not nobody is disagreeing with that.  Comparison between my screenshot and Vattic's screenshot shows that the trees are overall far scarcer in his 'final state' than my 'final state'.  There are very clearly observed clusters on the savanna map, areas where trees simply refuse to grow, the clusters on the forested map however are a lot smaller and more disputable.  Since the comparison reveals that there is a quite distinct difference between the overall tree coverage in both biomes, it does not follow that the tree growth model is indeed growing a smaller number of trees in savannas than in forests.  The drawbacks are that we do not know what a savanna looks like after 7 years but only 3.

Firstly, saplings do not pop up at once, the appear gradually over time (and probably completely independent of whether there are trees or not, except, of course, on the tree tiles themselves), so they should mature gradually as well. Secondly, the number of L shapes have gradually increased over the years, with the later ones appearing significantly after the original tree and the pairing tree, and I haven't cut down any trees since the original race, except to set up no grow zones around my fortress and the map edge, and the trees cut down at that time were those actually in those zones.

I've got one of these as well
TTT
  T
  .
  T where the missing tile contained a herb that's since died, but no sapling yet.

Edit: The embark is not a jungle, but tropical savanna or shrubland, based on the animals present.

Saplings do pop up virtually at once.  This is because as you said earlier they take 3 years to mature and Vattic's screenshot is only 3 years after embark with few or no saplings already present, meaning all those saplings popped up pretty much instantly upon embark; the amount of time it takes saplings to appear is basically negligable.

Animals often wander out of their home biome a few squares away; also you might have clipped the edge of a jungle when you placed your embark which would mean you would see jungle creatures on your map despite it mostly being savanna.  I would also like to see a screenshot of your embark and a clarification of how many years have passed since embark as well.

[PatrikLundell]

Two screenshots zipped up below. The first one is topside shortly into the 14:th year. The second one is the trade depot access map. The reason the lower right quadrant is cut off is because of two cage traps placed to catch giant tortoises (tropical savanna and shrubland only), which happened to end up at choke points (one being in my tree free zone). I've also seen giraffes, rhinos, mongooses, eagles, cavies and a bunch of other animals, with nothing indicating jungle.

http://www4.zippyshare.com/v/997GSMMW/file.html

Quote from: PatrikLundell on November 29, 2015, 01:17:20 pm

    Firstly, saplings do not pop up at once, the appear gradually over time (and probably completely independent of whether there are trees or not, except, of course, on the tree tiles themselves), so they should mature gradually as well. Secondly, the number of L shapes have gradually increased over the years, with the later ones appearing significantly after the original tree and the pairing tree, and I haven't cut down any trees since the original race, except to set up no grow zones around my fortress and the map edge, and the trees cut down at that time were those actually in those zones.

    I've got one of these as well
    TTT
      T
      .
      T where the missing tile contained a herb that's since died, but no sapling yet.

    Edit: The embark is not a jungle, but tropical savanna or shrubland, based on the animals present.


Saplings do pop up virtually at once.  This is because as you said earlier they take 3 years to mature and Vattic's screenshot is only 3 years after embark with few or no saplings already present, meaning all those saplings popped up pretty much instantly upon embark; the amount of time it takes saplings to appear is basically negligable.

Saplings appear one at a time (I guess there might be ticks where two may show up) over an extended period of time. They do not lie dormant and wait for you to clear the trees only to spring up at the first opportunity (some real life forest fire dependent plants do that, but they're neither trees, nor, presumably, modeled by DF, as forest fires aren't really modeled). Start a tree farm and see how saplings appear gradually over the years, rather than a sudden boom just after you've dug out the area. You can do the same thing above ground if the soil is two levels deep: just channel out an area and watch it get populated over time, or just remove all plants with a dirt road after clear cutting.
Also, saplings are already present at embark. And yes, I didn't quote the wiki page completely regarding the 3 year time period since it wasn't relevant to your 7 year one, but the wiki modifies the statement to say some mature early, so it might be that there is a variability in the maturation time with a "normal" time of about 3 years.

[cochramd]

Neither do you.  Nor do I need any proof, since the conclusion follows from reasonable premises;

If an Olympic gymnast lands on their face and snaps their neck while attempting a triple somersault, is it reasonable to assume that they did it on purpose? Just because someone is competent doesn't mean it's impossible for them to make a mistake. Competent people make mistakes all the time, including Toady and the other devs. If they didn't, there would be no bugs to report. And even if the devs were incapable of making mistakes, Dwarf Fortress is still incomplete. This could very well be something the devs know about and intend to fix later.

I'm not even going to pay lip service to everything else you've posted, as you clearly refuse to address the issue at hand. Instead, I'm going to start up a forest embark when I get home and compare it to Vattic's savannah embark. I predict that the savannah embark after 3 years will have a tree density comparable to the forest embark on day 1, and that after 3 years the forest embark will have a higher tree density than the savannah embark.

[Dozebôm Lolumzalìs]

Thumbs up for expressing what both of us think with a quick analogy, cochramd!

GoblinCookie: Okay. Fine. I will reduce my statement to one (not as easily definable) point:

There are too many trees.

Savannas should not have that many trees.

And you know what? I'm going to go ask Toady. That's what somebody did on the religion thread, and it seemed to work somewhat.

[cochramd]

And here's the forest embark:

Spoiler (click to show/hide)

I have PMed Vattic to ask him exactly what he did in his savannah embark, and when he gets back to me I will do exactly what he did in the name of science. I can tell that right now, at the moment of embark, it's approximately as tree-dense as Vattic's savannah embark after 3 years:

Spoiler: year 8 (summer) (click to show/hide)