649 réponses à ce sujet

[MerricksDad]

Well, Sen, your data for forcing placement is spot on. Unfortunately, the data for correct height is off, but only because my tiles are rotated in relation to your same tiles. Also, I haven't determined which lines in your list are duplicated yet, but at least one is an alias of a previous line, just at a rotation of probably 90 degrees. I'll get that info for you later.

 

So for anybody using Sen's data, something about the logic in the tileset takes into consideration one corner having importance over another. I suspect that is the top left corner, because it is first listed in most places you see the tile entry. I'll see if I can figure out what the exact logic is once I get my tiles set up and compare my data to Sen's. In the meantime, if you want your tiles to work perfectly with Sen's data, make your tiles in the same exact rotational alias as Sen's tiles.

 

What I mean by that is that if you have a tile with corners

 

12

03

 

then it has these potential aliases

 

23    30    01

10    21    32

 

I found this because I normally have the lowest corners in the tile follow a strict order of

 

lowest      low

high  highest

 

except in the case where they HAVE to be in another order, such as diagonals. In any case the top left of my tiles are ALWAYS the lowest point.

 

My apologies. The data is spot on in both ways. And I was so far in correct. I just found a digit incorrect in my own code while implementing your data. I will retry and see if that fixes the model position issue.

 

I found this out by creating an alias column on your spreadsheet, and there was no possibility that your data was wrong for the first part.

 

Edit:

 

After checking that your data works or not, I find that it only works on those tiles which are oriented the same direction. As expected, something about the corners, probably one being primary, or calculated differently when placed, changes what numbers are needed. Instead of working out the details on about 20 tiles with up to 15 variants each, I'm writing a small script to reorient all my tiles which don't match one of your tile orientations. Mine only goes to level 3, and my 4th level is down into water. So far, I only have lake water at level 0, so I had my script assume water was 0 when referencing your chart. That seemed to work perfectly.

 

As of right now, I can raise to level 1, but not level 2. Raising to level 2 raises an area, but then I can back-track and carve out to level 2, but cannot have a flat section of level 2. I can do the same with level 3. Again, no flat sections of level 3 can exist.

 

Also, I am not sure what is going on, but in the toolset, I have about 6 tiles which will not display. When I hover over their location, the filename has a period at the end. When I check the file, there is no extra period. Neither is there a period in the tile name in gmax, or in the reference to that tile in the SET file. Has anybody seen this error before in the toolset? My filenames are exactly 16 characters long, which I understand is the max file name reference size in all NWN content.

[MerricksDad]

Applying the rotational changes didn't seem to fix anything either, so I guess I'll start doing it by hand.

 

Here's what it looks like so far. You can see three levels of height and various transitions between those levels.

 

[YeoldeFog]

It looks like Jelly and seems to be very bouncy!

[MerricksDad]

@sen

 

Found the error. In the spreadsheet, the entry for tile 0200 is incorrect. It should be a height and corner adjustment of 0. You can verify that by checking the related tile in your SET file and the actual tile file in your override kit. That seems to be the only error anywhere, because it works perfectly now, and it is accurate to say that tile rotation in relation to your spreadsheet has absolutely NO effect on your data at all.

 

@YeoldeFog

 

I think I shall populate it with Jello Jigglers and Death Blancmange

 

 

Anyway, packing the blank set for viewing right now. And all this without a coffee yet. Time to have a coffee cake muffin and a double shot, so I can start my day.

 

Just FYI, the package has a smooth crosser, but the smooth variants are currently not labeled with the smooth crosser, so they show up as just variants to the matching corner heights. I'll fix em later once I perfect a script to detect their variants.

 

Each tile model comes with an internal dummy called tileData. Under that, 3 more dummies exist: corners, crossers, and offset. These three things represent the script-detected corner heights (not modified for Sen's data), if the edge region is a smooth crosser, and the offset provided by sen's data which was applied to the tile. Anybody who wants to access the files at a later time with data scanning programs, like mine, can detect more details about the tile model this way, and make adjustments accordingly.

[s e n]

@sen

 

Found the error. In the spreadsheet, the entry for tile 0200 is incorrect. It should be a height and corner adjustment of 0. You can verify that by checking the related tile in your SET file and the actual tile file in your override kit. That seems to be the only error anywhere, because it works perfectly now, and it is accurate to say that tile rotation in relation to your spreadsheet has absolutely NO effect on your data at all.

 

@YeoldeFog

 

I think I shall populate it with Jello Jigglers and Death Blancmange

 

 

Anyway, packing the blank set for viewing right now. And all this without a coffee yet. Time to have a coffee cake muffin and a double shot, so I can start my day.

 

Just FYI, the package has a smooth crosser, but the smooth variants are currently not labeled with the smooth crosser, so they show up as just variants to the matching corner heights. I'll fix em later once I perfect a script to detect their variants.

 

Each tile model comes with an internal dummy called tileData. Under that, 3 more dummies exist: corners, crossers, and offset. These three things represent the script-detected corner heights (not modified for Sen's data), if the edge region is a smooth crosser, and the offset provided by sen's data which was applied to the tile. Anybody who wants to access the files at a later time with data scanning programs, like mine, can detect more details about the tile model this way, and make adjustments accordingly.

I should correct the spreadsheet then If I have access to my vault account, dunno

[MerricksDad]

I'm packing the updated spreadsheet with my first release of this thing. Also, I think there may be another, but to find it, I need to make a script to test for me. 0200 was easy, because it was the most logical place for the issue I was having. This other one is one or two spots out from the raise/lower tool center, so I don't know which tile it is trying to show me is wrong. Will update when I find it.

 

The little script I made to detect the smooth crossers worked pretty good, and helps me find that other hole. It only happens when I try to make a high place, a medium place, and a span in between. So it has to be down in the 2-3 range somewhere, I think. If I drop down to the original 50 or so tiles, I should be able to locate it fairly simple, since I am not using the full set up to 4 elevations.

 

Edit:

 

Found all instances of other errors to be either my doing, or in my interpretation of my water as level 0. I then found one subset of regions which cannot be smoothed. Other than that, this is really slick in the toolset. Gotta go get my kid off the bus now. Be back to finish and upload.

[s e n]

it gives you red?

[MerricksDad]

It did, which should be an indicator of a missing tile, but the tile wasn't actually missing. Something in the -1 setting on all my 0120-type tiles needed to be reversed, and then I noticed the same issue on anything where a water corner was substituted in a 0-place on that tile type.

 

I'm up to like 590 tiles now 3 levels + ground level water, all with smooth modifiers that can be applied, and up to 2 additional variants of any given tile.

 

The cool part is that I kept the modifier stack list intact from the granitelands set, so I should be able to just release this blanks set, then apply some granite lands magic to ALL the tiles, pop in some water planes, and have a great 90% done prototype for granitelands to release maybe next week.

 

This may not seem fun to any of you, but I think it is very entertaining, especially to put together stuff other people came up with recently or so long ago.

 

 

Above is the proof shot I made for myself so I know for a fact 3 levels worth of non-softened tiles is done and that I don't need to mess with them anymore.

 

I cannot seem to make one for the water combo. Water breaks at or near a tile with ww20, forcing the 2 to change to a 1, causing other issues from that point forward.

[MerricksDad]

Well, bad news. You won't be getting this tonight. When those tiles fail to load into the toolset, they create a white hole. Any time the white hole gets made, the data for the cols and rows in the map file get corrupted, ruining the area on next load. All tiles adjacent to the messed up tile row are also scrambled.

 

I'll work on it tomorrow and see if I can figure it out. Something about it is causing the filenames to be 17 characters in the toolset. Not all, but many. Every tile that does not load gets a random character after it. I may have to see if the toolset has problems with 16 character names, or if the models themselves have issues. Usually it is the same models failing to load, but sometimes they do load. It's weird. There is no error prompt.

 

Also, I am unfamiliar with the details of MDA's tileset editor 5, but it cannot recognize my gog nwn install, faults on entry, and faults any time I try to click a tile entry, often crashing on the spot. So I am stuck editing everything manually.

[The Mad Poet]

I just want to say any of you guys who choose the obvious self-harm of making these tilesets have my undying adulation. Seriously. This is a hell of a lot of work and just to make one or two extra tiles has been a chore I've dropped. I just reskin these monsters. Thanks for all your hard works.

[MerricksDad]

Got the majority of the water working last night. Will test a little more today to make sure I don't get loading failures on the smooth options for water. If that continues, I may just drop the range of tiles with an issue and release as is.

[MerricksDad]

ok, going to pack it up and put it on the vault now. I renamed the files back to the way I had them before, as tmd05_#####. This current method DOES seem to stress the toolset for some reason. Normally resref data is allowable to length of 16 characters. That is apparently not true of tile names, as 16 characters seems to cause some over-read, and then all the data starts to get screwy in the toolset.

 

I looked through tiles one by one and I am 90% sure I have all the missing tiles accounted for, primed to the correct height/placement logic combo, and all that. But there is one tile logic instance which still fails. If you have a strip of water at ground level, and you have a strip of height 2 blocks placed along that water strip, and then you paint down the strip with water again, or otherwise change any given water tile adjacent to height 2, one of those height 2 corners will convert to height 1. I've poked and prodded and can find no reason. It also does not do it every time, which leads me to believe that it is either a failure in the toolset, or I have a tile which is wrong. Thing is there is only 3 tiles that can fit that description, and only one of those matches the area exactly (without crossers) and would be affected. I've tried poking around with the adjacent tile heights ww20 and ww02, and they seem equally correct.

 

You can also get a similar behavior with ww30, but not ww03.

 

There is nothing keeping you from going back and reusing the height tool to set the level to 2 or 3 again, as the tile does exist, and is set up in the SET file.

 

I was able to work out some non-placement bugs, and I feel confident that they were related only to the length of the file names. Depending on how many tiles I had, a different tile was the corrupted one, so I think it is a matter of no tile actually being corrupt, and more a matter of a memory access error inside the toolset. Let's hope so.

 

You should be able to find the "Placeholder Tileset" on the vault shortly.

 

http://neverwinterva...eholder-tileset

[T0r0]

Ooh! Can't wait to take this for a test drive. I'm sure this might help me plug in some missing tiles.

[MerricksDad]

I will say that reverting to the meaningless 5 tile digit is less informative than the other naming scheme, but at least you can easily find that exact tile by 5-digit code in the SET file. What I want to do is have a floating descriptor above the tile. That might be my next release. I just need to pick out a nice decal font and set up a top-down character placement script.

[MerricksDad]

I have successfully created a clone of the blank set with floating text over the corners. It reports on the tile the corner heights, the crosser, and the center offset, if it has one. I'm going to pack this and post it as an alternate on the vault, under the same entry as the previous.

 

[T0r0]

Genius !

[MerricksDad]

I am frightening myself. I just spent the last 2 hours working out a script to construct bases and general details for a set including this:

 

4 levels of base terrain (0 - 4)

2 levels of forested terrain (1-2)

2 levels of water (two levels below 0)

no crossers of any kind

no tile variations

no duplication at higher levels (ie 0123 is the same as 1234)

 

I am getting 1590 tiles required minimum

 

... can I do this? My other method of applying trees was going to be more tiles anyway. I think I can do this.

 

There is an option where I omit level 3 of the base terrain, and I could drop probably 500 tiles. I would still need a level 3 base set of 4 tiles to get me to that level initially, but then I can omit any combinations containing non base terrain and level 3. This would give me a smooth raise, a steep raise, and a high mountain, like I want, and I would be able to populate level 1 and 2 hills as forest, but would leave level 0 (meaning any flat places at any height) not covered by tileset trees, ever. That leaves only placeable tree groups for use on flat regions.

 

To build a chasm in such a setup, I would just raise two regions, 2 to 3 tiles apart. If I wanted tree cover below, 3 tiles apart with two +1 raises inside would do the trick.

 

The 1000-1500 tile setup allows me to drop water anywhere I want, and that can be either deep water, or enterable shallows.

 

I think this is worth doing.

 

The alternative is make a treeless tileset and just use a ton of placeable trees. I am not certain I would personally like that, although it would save a lot of work on my part.

 

Thinking out loud here...what if I separate each tile into 4 regions, create a set of 3 alternative chunks per region per total height and what it leads into, and then fashion tiles mechanically, like I did in '08-'10....Then I could use my tileData properties to patch together a tile out of 4 chunks per tile, chose randomly from the 3 alternates available per corner.

[Tarot Redhand]

Can you automate the creation in any way? Otherwise 1500 tiles sounds almost like masochism. I remember being almost ready to pull my hair out when I did about 250 placeables for one of the ccc's.

 

TR

[kalbaern]

Personally, I'm not opposed to a lack of trees as I think that while it is more work for a builder, using placeables allows us to create forests and forested slopes with a wider range of variety or environments.

[3RavensMore]

Personally, I'm not opposed to a lack of trees as I think that while it is more work for a builder, using placeables allows us to create forests and forested slopes with a wider range of variety or environments.

 

Second this one.  I often just cycle through the available tiles until I get a treeless one when I build.  Additional placeable trees are always nice to see as well.

[Tonden_Ockay]

I say do what you feel is best for the tileset. That said I my self love things to be part of the tileset vs placebles when ever possible. Even with my gaming PC placeables really slow down the toolset when building.

[MerricksDad]

I wouldn't be doing this if I wasn't using some form of automation. I'm too lazy and uncoordinated to use GMAX as a drawing tool. It doesn't even have a proper union/carve function because it fails in math all the time.

 

I started modeling on a program for Quake/Hexen called WorldCraft. My room mate and I purchased a copy at like 15 bucks or something back in 1998. The program itself was great, but we ultimately had to rebuild the compilers for light and vis, because the compilers could not handle our 3D workload. Early on, I was into detail a bit too much. Kevin was more into play-ability. Together we made a good team. We need to find a way to do more team projects like that for NWN.

 

 

ok, on the topic of trees: My original goal was to create a series for pine forest, another series for spruce forest, and then pretty much fill wetter cracked rises with aspen, and then release placeable trees as: pine, spruce, aspen, cottonwood, birch, and scrub oak. This would allow me to duplicate the black hills feel. But that method would require thousands of tiles with fixed crossers. At best, if I did it as corner terrain, that is still near 1000 tiles.

 

on rocks: I originally wanted to supply a rockless grass terrain, a schist terrain topped with pine needles, and a bare granite terrain. I then wanted to offer placeable granite boulders, schist boulders, schist shingles, deep mine debris, and decals of various crushed rock textures to cover up the bare granite landscape. I know now that doing granite and schist separately is an ungodly quantity of tiles, and could be easily done as two separate tilesets. But I also know that just offering large walk-blocking outcrops can serve the same purpose, and lets the builder fashion at least 3 varieties of tiles by hand.

 

It may be best to ultimately leave the tiles as naked as possible and release a massive package of placeables. When defined as static, that allows the builder to create a low-cost environment, and allows me to move forward with other tilesets and placeable kits.

 

I may recarve the granite structures in such a way that it leaves much more room for the area builder to place various shapes. This should allow the builder to choose a tile modifier that could represent rounded granite, layered granite, layered schist, or some other combination. If I take a clue from the construction method of Torchlight 2, I can build a variety of limestone which can also be applied to a cliff edge, and then we'd have the full combination of anything you might see in the black hills.

 

Let me see if I can whittle at the count a bit and come up with an easier functioning build.

[T0r0]

 

 

We need to find a way to do more team projects like that for NWN.

Amen !!

[MerricksDad]

removal of forest corner terrain, as well as keeping all 4 levels of height adjustment, yields only 543 tiles required.

 

I really need to keep a raised grass area, so I think grass 0 and grass 1 will retain a walkmesh grass. Grass 2 I will start to transition into a more grassless terrain with more ground debris and maybe some on-tile plant clumps. By Grass 3 there will be a sharp rise, and that area will be a great place to have purposely constructed schist and granite coverings, like paper doll clothes. By Grass4, there will be no option but to have the granite fully fleshed out on the tile, however, this also makes a great place to create placeable round spires and granite needles, especially on corner tiles. This method also has the benefit of making really simple (like childlike simple) walkmeshes.

 

What I can do then is make flat-terrain mixers for either grassed area, or bald area. That bald area can then be modified with decal placeables, and it still leaves me a grass flat to play with as prairie area. Flat terrain mixers are really easy to do if you keep them confined within another terrain type. Basically it leaves only 4-6 tiles required.

 

The shadows will begin to get fairly strong below a 3 level rise, so it might be best to not have any vegetation at the base, just piles of dirt and rubble. I'll leave those neutral in color, so that they can represent both schist and granite debris from paper doll rocks placed over the rise edge. In the black hills, where you have a 4 level rise, usually the bottom is water-cut, leaving only moss, lichen, and fine debris at the base, with boulders strewn about, or even mostly buried in finer debris.

 

It may be best to leave trees entirely in the realm of placeables. One main reason for that is because you don't all like the same texture colors I do, or the shapes. My detailed trees may not serve your purpose, if what you want is a more smooth looking area. While you can simply replace textures, you cannot as easily replace entire mesh areas. At a later date, I may release one with my personal selection of vegetation. Let's just call that an expansion pack. 1000 tiles for an expansion pack seems a bit much, but 1000 tiles just for two terrain corner types is a tileset is more ridiculous.

[MerricksDad]

I may need to take a pinch of time off this whole project. Before I went out to SD, I noticed a bruise form behind my right knee, for no explained reason. I imagined that I had just pinched it and didn't remember it, or that I had sat wrong in my chair. It took about 2-3 weeks to get rid of the mark walking around every day in SD, and I felt really good doing it.

 

When I got home, my joints started to set up a bit, and I noticed one day after taking my dog for a walk, that my toe bone in the flesh behind my third toe on that same toe would not move. I flexed my foot, and instead of fixing it, I think I tore the flesh straight off the bone above the bone.

 

A week went by, and my thighs at the inside of the knee, down to the muscles that attach to the knee began to feel strange. When I stand up now, they are kinda stiff, as if they are no longer made of muscle material. After I stretch them once, they are fine. If I sit down again, even for 2 minutes, the next time I stand they are damaged feeling again.

 

Two days ago I noticed a bruise on my left bicep, about the size of a mosquito bite, but under the skin rather than on top, like a bug bite. There is no pain or swelling.

 

Last night, before walking around a local park, I noticed the joint at my foot connection is sore, from the muscle just above the ankle down under the muscle which actually straps to the top of the foot.

 

Then while walking tonight, it feels like a shin splint half way up my left leg, and the pain in the left foot is worse. Tonight I also notice the pain or other sensation is the same as in my knee sides.

 

I must rethink my health at this time, and that may be a temporary death of this project. I certainly hope not, and would like to think even on raining days I would be here working on this. But I need to exercise, and I think a lot, or I am going to be in the hospital shortly.