sea edges

[wilddog]

OK I'm getting old and can't remember.

Whilst editing the map I'm hitting the old problem that bottom left hand corner of the heights map are leaving the 'square' as land and so leaves little pointed squares leading out to the sea.

How do you get rid of this? I'm using heights 1,1,1 settings and that doesn't seem to work. Also I wasn't sure if the height HTG map had anything to do with that. Except that when I got rid of it it still generated an RWM map but still had the occasional issue I mentioned above. However, some heights were greater than 1,1,1 and didn't show the issue and some were 1,1,1 and still showed the issues. Any idea?

[Spurius]

Mmm. Just used the exact same technique to get rid of them, and still works the same here as in RTW, meaning no problems. I do throw away map.rwm every time I change a stretch....

[Muizer]

OK I'm getting old and can't remember.

Whilst editing the map I'm hitting the old problem that bottom left hand corner of the heights map are leaving the 'square' as land and so leaves little pointed squares leading out to the sea.

How do you get rid of this? I'm using heights 1,1,1 settings and that doesn't seem to work. Also I wasn't sure if the height HTG map had anything to do with that. Except that when I got rid of it it still generated an RWM map but still had the occasional issue I mentioned above. However, some heights were greater than 1,1,1 and didn't show the issue and some were 1,1,1 and still showed the issues. Any idea?

Have you tried 0,0,0 for tiles that stick out to the lower left and upper right?

[wilddog]

Muizer
Haven't tried 0,0,0 as that used to cause a height problem before (peak). In RTW you only needed to use the 1,1,1 size before and it got rid of the problem.

Spurius
Don't really know what to say. I do the same and get the issue. Even using the original map. Renaming the hgt map, deleting the rwm map and looking at it I can see a square in south Cornwall where it shows the problem and one in South Wales (not set to 1,1,1,) where it doesn't.

[Spurius]

Well, wasn't trying to make fun or anything, just could have slipped your mind, happens to me too. Puzzling, then - it DOES work for me, all the ones I did so far vamoosed neatly - can only guess there is still some difference in your overlay of heights/ground/regions compared to mine - maybe just minutely off in yours or whatever.

[Muizer]

Muizer
Haven't tried 0,0,0 as that used to cause a height problem before (peak). In RTW you only needed to use the 1,1,1 size before and it got rid of the problem.

All I can say for certain is that those tiles need to have a value "halfway" between the adjacent land tiles on the one hand and the sea tiles on the other. IIRC, on my maps, giving those tiles a value of (0,0,0) as being halfway between (1,1,1) and (0,0,255) seemed to work OK.

[Trajan]

You know that there's a mapping sub-forum right? Moved to Mapping.

[wilddog]

Muizer
I was doing some experimenting yesterday. Just a few quick ones. I think my main problem was editing the terrain text files for heights was showing up the problem more but I was still seeing it when using 111 or 000 for shoreline.

I Haven't spent too much time on it yet but don't really understand why I'm getting it. I need to decide some heights on shorelines so will probably come back to this once I really edit the coastline. However Is till see it on the original map also in a few locations.

[alpaca]

Muizer
I was doing some experimenting yesterday. Just a few quick ones. I think my main problem was editing the terrain text files for heights was showing up the problem more but I was still seeing it when using 111 or 000 for shoreline.

I Haven't spent too much time on it yet but don't really understand why I'm getting it. I need to decide some heights on shorelines so will probably come back to this once I really edit the coastline. However Is till see it on the original map also in a few locations.

Which are those?

[wilddog]

Alpaca
I presume you meant in the original map

See here at the bottom of Cornwall

[Spurius]

[posted this before, argh]

Have spitting image problem in my bigmap, same place - but the 1,1,1, technique works everytime for me - really don't get what's different for you, there. I find 7,7,7 or 10,10,10 there, pencil in 1,1,1 and kill map.rwm, restart and it's gone, like it should.

Must be some difference in how your maps overlay compared to mine then, only thing I can think of.

[wilddog]

Spurius
I find 2,2,2 there. However when I switched to 1,1,1 I still saw the square edge. However it might be I'm not using correct combinations of sea colour to offset??
Its just that the northern tip of Cornwall doesn't show the problem (or south western part of Wales) but they have different sea coloured pixels. I'd never paid much attention to that before and had been using a standard 0,0,153 for sea so maybe that's the problem?

At the moment not really paying too much attention to this as I'm mostly sorting out the rivers first.

[Spurius]

Think I maybe found the problem - if you overlay the regions and heights, do you ENLARGE the regions to heights size, or REDUCE the heights to regions format?

I always do the latter, since that is MUCH more reliable - taking off is more accurate than adding to, here...

[wilddog]

Spurius
I always reduce heights to region size. Though in the past I did that with paintshop pro rather than photoshop elements. Not sure why that should make any difference though as its the value in heights it uses isn't it?

[wilddog]

Spurius
I finally decided to have a look at this and see what was happening. Also downloaded your big map to see if it was having the same problems on the Cornwall bit and it was. Switching heights to 1,1,1 reduces the size of the problem square but doesn't remove it.

The simple solution for getting rid of it is to change the sea pixels to be a lower blue value eg 0,0,240 seemed pretty good although for the 2,2,2 issue 0,0,249 or below seemed to resolve it (Note I only did this for the problem squares not all - I don't know if there are any side effects to that or not).

Also spotted that the pixels where sea pushes into land ie the top right pixel being land and the lower left being sea it is best NOT to use heights 1,1,1 and increase to 2,2,2 - this will then stop the reverse happening i.e. an edge of sea appearing in the land.

Thumbnail below (sorry its quite big) shows Dee estuary left side corrected, right sides top shows indentation of sea (corrected afterwards as explained above). Mersey estuary on right side hasn't been corrected and so shows both problems.

[alpaca]

Whoops. Hmm can you try converting that map with hgt converter and see what happens? Beginning to become curious.

[Spurius]

Spurius
I finally decided to have a look at this and see what was happening. Also downloaded your big map to see if it was having the same problems on the Cornwall bit and it was. Switching heights to 1,1,1 reduces the size of the problem square but doesn't remove it.

The simple solution for getting rid of it is to change the sea pixels to be a lower blue value eg 0,0,240 seemed pretty good although for the 2,2,2 issue 0,0,249 or below seemed to resolve it (Note I only did this for the problem squares not all - I don't know if there are any side effects to that or not).

Also spotted that the pixels where sea pushes into land ie the top right pixel being land and the lower left being sea it is best NOT to use heights 1,1,1 and increase to 2,2,2 - this will then stop the reverse happening i.e. an edge of sea appearing in the land.

Thumbnail below (sorry its quite big) shows Dee estuary left side corrected, right sides top shows indentation of sea (corrected afterwards as explained above). Mersey estuary on right side hasn't been corrected and so shows both problems.

I found a final solution (well for now) that does the trick for me, I select all sea, expand that 1 pixel, and then select only that 1 pixel edge and replace it all with 1,1,1 that solves all triangles BUT gives the reverse problem of BLUE triangles on land in places where diagonal corners touching blue are also 1,1,1, and if I change those to 2,2,2 its perfect. So the diagonals touching 0,0,253 water(which it should be) from left bottom to top right should be 2,2,2, any other direction it's 1,1,1 in my last version.

[wilddog]

Alpaca
Same issue still exists on the standard map in a few places so you should be able to use your converter on that.

I haven't downloaded the converter yet and not really sure what you are trying to find out. I haven't added in may cliffs before so want to see what that does.

Spurius - I understand? You've resolved the sea into land as I concluded above but for the reverse if you are doing land into sea and using 1,1,1 and 0,0, 253 it doesn't work as far as I can tell. It just reduces the size of the land into sea triangle you see but it certainly doesn't resolve it as far as I can see. For my maps I also don't want to reduce the land to 1,1,1 in all cases so changing the sea for me is the better option.

[Spurius]

Well this is with a 1 pixel edge of 1 1 1 and a beachline also 1 pix wide in groundtypes, with the diagonal corners 2,2,2 as described in my previous post:



Beach is optional of course. But no brown or white triangles, anywhere.

(Is not the version currently in dl btw)

[Muizer]

I. So the diagonals touching 0,0,253 water(which it should be) from left bottom to top right should be 2,2,2, any other
direction it's 1,1,1 in my last version.

Indeed. I suspect 2,2,2 bordering 0,0,253; 1,1,1 bordering 0,0,254 and 0,0,0 bordering 0,0,255 (which is what I always use) will all give a clean coastline. I wouldn't be surprised if this went a long way further, meaning you could make real cliffs with 10,10,10 bordering 0,0,245.