Mountainbanners - 1059 Early Spring by Fedor

Map Description:

Mountain-Banners, after the seventh and final year of the great Bridge-building challenge. The points of interest, if seen in order starting with "0. The sheer...", give you a guided tour of the fortress and its construction.

Bridge-building challenge thread (page five, includes 3D images):

Story: "A Journey to Mountain-Banners"

savegame

There are 32 comments for this map series, last post 2010-01-23
This map has been favourited by 23 people.

Add a Comment

Comments

Submitted by: Markavian - 2008-01-13 to 1053 Early Spring

It has a diagonal entrance on the surface? Neat! Have you considered starting a quarry to obtain the necessary material?

Submitted by: Fedor - 2008-01-13 to 1053 Early Spring

Just the beginning, just the beginning...

Yes, there is a chalk quarry, but it's not shown (it's under the river).

Submitted by: Lacero - 2008-01-20 to 1054 Early Spring

The side/front views are all broken, which is a bit of a shame for something like a bridge :)

Hopefully markavian will work out what I did wrong and we can fix it.

Submitted by: Markavian - 2008-01-20 to 1054 Early Spring

Hey Lacero, I've fixed the viewer now. Thanks for pointing out the bug, I had noticed it before, but didn't realise what the cause was until your brought it up. The calculations for Front and Side views were reversed, and the width and depth values at initilisation were the wrong way round. Wholey my mistake, I hope this improves your map viewing experience now.

Submitted by: Lacero - 2008-01-21 to 1054 Early Spring

Thanks Markavian, very much appreciated :)

Submitted by: Markavian - 2008-01-29 to 1055 Early Spring

ITS A WHOPPER ALRIGHT, and it uploaded ok, you should be able to upload it yourself from this point onwards... just make sure you're logged in because the limits are different (less) for visitors.

And make sure you put a point of interest for the side view because that looks fantastic.

[Message edited on 2008/01/29 at 02:41 by Markavian]

Submitted by: Fedor - 2008-01-29 to 1055 Early Spring

You got it! One side view POI coming up!

Submitted by: SL - 2008-01-31 to 1055 Early Spring

What font and color scheme did you use for this map, by the way?

Submitted by: Fedor - 2008-01-31 to 1052 Early Spring

There are two players, each with a different font and color scheme. The 1055 early spring fort is shown using Herrbdog 16x16 tileset (Guybrush's updated version), with Dystopian Rhetoric graphics and brown color edited to be more reddish and less yellowish.

Submitted by: SL - 2008-02-01 to 1055 Early Spring

Did you change any other colors or edit the font at all? I'm trying to re-compress it with the in-development version of the compressor to see how well the filesize-reducing enhancements I've been working on work on it, but I'm currently getting "Identified 638, failed to identify 2614, found 541 duplicates." Not quite sure what the 541 duplicates are at the moment, will have to mess with the compressor's innards a bit to find out. The 2614 unidentified tiles are what concern me the most at the moment. I don't think DR's dwarf graphics have 2614 different tiles in them, so maybe something is wrong somewhere (My main fort gets "Identified 481, failed to identify 157").

(Even so, the filesize is down to 247 KB)

At the moment I'm using the Herrbdog_16x16_tileset.bmp font downloaded from the wiki, and the default colors also downloaded from the wiki, except I changed brown to RGB 160,96,48 (copied from the color of one of your bins).

Edit: You don't happen to have your color depth at 16bpp instead of 32bpp or something, do you?

[Message edited on 2008/02/01 at 04:41 by SL]

Edit #2: As best I can tell, some of the font characters seem to be different. The first one I've been examining - originally to see if the colors were correct, starting with white - was a white statue. But it appears that the statue tile itself in this font doesn't completely match the one in your image.

[Message edited on 2008/02/01 at 07:20 by SL]

Submitted by: Fedor - 2008-02-01 to 1052 Early Spring

Here's a link to my init.txt file and my copy of the graphics, just to help things along:http://rapidshare.com/files/88488248/df_settings_mountainbanners.zip

The map was taken with 32-bit color enabled. But, even with 32-bit color, my computer sometimes produces images that don't yield perfect tile-by-tile matches. I didn't get a size warning on this particular map, but I not infrequently do.

I talked with Markavian some months ago about this and asked if the scanner could accept a little color leeway when matching pixels instead of requiring a perfect match. Don't know if he saw fit to pass that idea on to you.

A user-customizable figure, defaulting to 50, for the sum of the squared R, G, B deltas may solve a lot of problems.

[Message edited on 2008/02/01 at 10:47 by Fedor]

Submitted by: SL - 2008-02-03 to 1055 Early Spring

He probably forwarded it to me, but I don't remember for sure. I hadn't ever seen a map where that kind of thing happened until now, though. So, now now that I've actually seen the weirdness in your map, and can run tests on it, it looks like a good idea to me. There actually is a factor like you described (except not user customizable in the version that's currently released) that's already built into the tile-identification code for the c# version of the compressor. That value is set to 12 by default right now. I had worried that having it too high would cause misidentification of tiles, resulting in tiles showing up with the wrong image, but that doesn't appear to be happening in the tests I'm doing on your map now. I needed values of around 150 or above to match all the matchable tiles in your map. (It's actually somewhere > 100 below 150).

To get additional debug output which shows how many tiles were identified and how many were not, you can run the compressor like this from cmd.exe with a pipe to a text file: DwarfFortressMapCompressor[press tab one or two times here to autocomplete the program's filename] > output.txt

Look in output.txt when it's done compressing and the output should be there.

If you're using a graphical tileset, there will always be some unidentified. The key hint as to whether you're getting all the tiles will be to run it again with a higher value, say, around 50 higher (When I release a version which lets you change this :P). If you get the same amount of failed-to-identify, you probably got all the tiles you could. For example, on my test map, there are 151 unidentified tiles. On yours, there are 110 (once we get the value high enough to identify everything else). I'm using the DQ graphics that veryinky modified to have more variation, so there are probably more different graphic tiles in mine than in yours, which would be why there are more unidentified ones in mine.

It's funny how your map has this nowhere-near-precise color, and actually seems to need a DeltaR˛ + DeltaG˛ + DeltaB˛ of above 100 (somewhere between 100 and 150), but has no vertical black bars, and yet my map only needs 12 but DF writes it out with vertical black bars on it. (Especially considering we have the same font size and the same window size and the same bit depth (32)...)

By the way, I tested it with the following values: 12, 25, 50, 75, 100, 150, 200, and 300. As I increased the value from 12 to 150, the number of identified tiles increased from 638 to 3142, failed to identify dropped from 2614 to 110, and duplicates increased from 541 to 2900. Above 150 these values remained the same. Unique tile images ratio remained the same (0.2566288%), except that it was calculated before tiles were identified and duplicates were eliminated. If we recalculate it, it drops significantly due to the few thousand duplicate tiles which were eliminated. (For comparison, my map has 0.04048388%, 481 identified, failed to identify 157, and 6 duplicates.)

I've also made the compressor also check (after the normal check to see if the unique tile ratio is so high that something must be terribly wrong) to see if the unique tile percentage exceeds 0.1%, and, if your color match sensitivity is less than 200, to say "This multi-level map may have color damage - Dwarf Fortress may have written out the images oddly, for instance. We may, however, be able to correct for it, by increasing the color match sensitivity. It'll make tile identification take longer, but the end result will be a smaller filesize, if it works. Would you like to increase the color match sensitivity to 200?" (It also lists your unique tile percent and so forth there). If you say yes, it'll also let you know how successful it was at correcting things once it has finished tile identification and pruning duplicates.

It's funny how your map has this nowhere-near-precise color, and actually seems to need that value to be above 100 (150+ does it), whereas mine only needs 12, and yet we have the same font size and window size. Actually, I wonder if it's because of your CPU, graphics card, or graphics drivers instead. Also, as a side note, the graphical tiles (the Dystopian Rhetoric art, I mean) don't appear to have been screwed up the way the other tiles were, so perhaps it's fuzzy math or something in whatever's coloring the font characters for each tile?


[Message edited on 2008/02/03 at 09:56 by SL]

Submitted by: Fedor - 2008-02-03 to 1055 Early Spring

SL, thank you very much for your careful study of this problem. I don't have anything intelligent to say about why my particular computer garbles the colors slightly, but adding additional robustness to the map compressor as you describe *should* work for cases like mine.

Minor comment: perhaps "... increase the color match sensitivity ..." should read "... increase the color match leeway ..." or "fudge factor", or perhaps the sentence should be reworded to read something like: "Would you like to match colors with a total squared RGB difference of [value]?".

Submitted by: Gaulgath - 2008-02-17 to 1056 Early Spring

That is one of the most splendid towers ever uploaded to the DFMA.

Submitted by: Freeman - 2008-02-26 to 1054 Early Spring

You make my day!

Submitted by: Keldor - 2008-02-27 to 1057 Early Spring

Hrmm... do you have the ini setting for interpolation set to LINEAR instead of NEAREST? That might be able to account for the differences between identical tiles. Otherwise, I can't think of ANY reason they shouldn't be identical... Very strange.

Submitted by: Fedor - 2008-03-02 to 1057 Early Spring

That would make sense, Keldor, but I did actually have it set to NEAREST.

Submitted by: Markavian - 2008-04-06 to 1058 Early Spring

This is really turning out to be something special. Love the rope-reed mosaic.

Submitted by: Lacero - 2008-04-06 to 1058 Early Spring

Fedor will have the final one up in a bit. We wanted the secret bits to stay secret until it was done.

That's why theres no PoI on this one :)

Submitted by: Fedor - 2008-04-06 to 1059 Early Spring

And it is done.

[Message edited on 2008/04/07 at 11:42 by Fedor]

Submitted by: Quift, - 2008-04-18 to 1059 Early Spring

Thats absolutely fabulous. Love the scale of the project, and the execution is brilliant with a great sense of eastetic detail. You really should turn siege back on to see how they and mega beasts fare...

Submitted by: Markavian - 2008-04-19 to 1059 Early Spring

(Sorry about the list of points of interest, they're sorted by height/depth as apposed to alphabetically.)

Submitted by: Fedor - 2008-04-30 to 1059 Early Spring

Thanks, quift!

I actually don't think this design is at all well-suited to handle sieges and megabeasts. The defensive plan consists of three strongpoints (side towers, valley fort) and one minor guarded entrance (plateau approach). The basic problem is that the dwarves have to walk so far to assemble, and the megabeasts or goblins can appear so close to the fortress (due to the smallness of the map). Imagine trying to defend a place that takes two full days - 48 hours non-stop running, no stops for food or drink - to get from one end to the other of!

[Message edited on 2008/04/30 at 11:49 by Fedor]

Submitted by: ShunterAlhena - 2008-05-01 to 1059 Early Spring

This is fabulous! All my fortresses taken together pale in comparison to the scale and architecture of this magnificent project. Hopefully none of my dwarves living in our little hole Cloisternotched will hear about this place - they think they have a nice fortress!

Submitted by: Markavian - 2008-05-01 to 1059 Early Spring

Fedor, what a wonderful fortress. The points of interest really bring it to life, and I am simply stunned for words at some of the features. The suspension chains made me yelp in suprise.

I have a favour to ask, you're not using the latest version of the map compressor, which means extra information about your fort isn't available. It would be a grand help if you could get the latest version, then future versions of the map viewer would be able to display this fort in glorious detail. At least that's the plan.

- Markavian

Submitted by: Fedor - 2008-05-02 to 1059 Early Spring

You got it, Markavian. Email with new version map sent.

Submitted by: Markavian - 2008-05-02 to 1059 Early Spring

Thank's Fedor, its uploaded, I kept the old file on the server as well. The POIs still work, and you can switch between view modes now, so I guess this is all set for the future... will keep you informed.

Submitted by: sneakeypete - 2008-12-17 to 1059 Early Spring

Ohh, methinks someones been editing the map description.
Its good though, more people ought to see this amazing fort.

Submitted by: Gairabad - 2008-12-27 to 1059 Early Spring

Epic.

Submitted by: Fedor - 2009-01-01 to 1059 Early Spring

Yeah, I edited the map description recently because of the change to the new forum, and therefore of the links.

Submitted by: Morty - 2009-01-15 to 1059 Early Spring

Marvelous. The task of building this from only 7 dwarves and untamed land can only be described as Epic (as someone already said up there).

Submitted by: Rolan7 - 2010-01-23 to 1059 Early Spring

Wow, what a work of art.
Wait, I was talking about the tapestry, what bridge are you talk... by Kun!
Thanks for the tour!

Viewer Controls

SHIFT + Key doubles keyboard scroll rate.

Problems?

Do you only see a blank space?

Don't have Flash?
You can download the compressed map file: 2008-04/fedor-Mountainbanners-region9-1059-13.fdf-map but you will need the .NET version of SL's DF Map Compressor to convert to the .PNG image format.

Login: