'Tiled' Tile-map Loader

[fallahn]

Not sure you're looking at the current version.

In m_AddTileToLayer()

        layer.vertexArrays[m_tileInfo[gid].TileSetId].append(v0);
        layer.vertexArrays[m_tileInfo[gid].TileSetId].append(v1);
        layer.vertexArrays[m_tileInfo[gid].TileSetId].append(v2);
        layer.vertexArrays[m_tileInfo[gid].TileSetId].append(v3);
 

In draw()

        states.texture = &m_tilesetTextures[i];
        rt.draw(layer.vertexArrays[i], states);
 

Individual sprites are stored for tiles during loading, but they are only ever drawn by objects which are of type Tile, and image layers which consist of a single sprite.

EDIT: On reflection object layers which use a lot of tiles could benefit from also using vertex arrays - but it may involve a considerable rewrite when taking into account transforms such as flipping / rotating objects which is a feature that should be added to Tiled soon. This would be much easier to handle using sprites.

[fallahn]

I don't know how many people use box2D in their projects (certainly no one has asked for this) but I thought I'd add this feature anyway. You can now create box2D physics bodies from tmx map objects - here's a demo vid:



I wrote a bit about it on my blog, and uploaded the class to the github page.

[blackwave]

Hi,

This project looks awesome. I'm trying to compile it on Linux and having some trouble. Specifically, I get the following error;

Code: [Select]

src/MapObject.cpp:220:37: error: non-const lvalue reference to type
      'Vector2<[...]>' cannot bind to a temporary of type 'Vector2<[...]>'
        return Helpers::Vectors::Normalize(end - start);
                                           ^~~~~~~~~~~
/home/cole/Desktop/sfml-tmxloader/include/Helpers.h:70:47: note: passing
      argument to parameter 'source' here
                static sf::Vector2f Normalize(sf::Vector2f& source)
                                                            ^
1 error generated.

Which is associated with this function:

Code: [Select]

sf::Vector2f MapObject::CollisionNormal(const sf::Vector2f& start, const sf::Vector2f& end) const
{
Segment trajectory(start, end);
for(auto& s : m_polySegs)
{
if(trajectory.Intersects(s))
{
sf::Vector2f v = s.End - s.Start;
sf::Vector2f n(v.y, -v.x);
//invert normal if pointing in wrong direction
float tAngle = Helpers::Vectors::GetAngle(end - start);
float nAngle = Helpers::Vectors::GetAngle(n);
if(nAngle - tAngle > 90.f) n =- n;

return Helpers::Vectors::Normalize(n);
}
}
return Helpers::Vectors::Normalize(end - start);
}

I have limited experience with C++, and I don't know what this code is supposed to do, but I was able to make the code compile by changing the second return line to:

Code: [Select]

return sf::Vector2f(0,0);

The examples seem to work with this code (although it's probably doing something horrendous somewhere; I don't know).

Is this problem just on my end? Am I doing something incorrectly? Thanks.

[fallahn]

Hi, a lot of the collision stuff is not thouroughly tested - I don't even have a linux machine right now - so this may well be a  bug. Try constructing a new vector from end-start and return that via normalize and let me know if it works. I'll make a note of it on the tracker.

[Sixoul]

How would I get an affect of walking in grass similar to Pokemon?



The character's legs disappear in the grass but their upper half is still visible. Is this possible with tiled and your map loader?

[MadMartin]

For that you have to have the sprite separated in two halfs (upper and lower body). Just check if the sprite is drawn onto a tile where semi-high things are (grass, snow, etc.) and only draw the upper body.

Not sure if this is possible with this map loader, but it is not that difficult in general.

[Sixoul]

That's what I was discussing with a friend and we came to the conclusion it would have to be a sprite split in half.
Thanks.

[cyberpala]

@fallahn

Hello I'm new here , started programming with C++/SFML. I want to try out your loader with the benchmark example. (VC2013) What have I do to compile the Benchmark example. (always get linker Errors)

The SFML intergration in VC2013 works.

Thanks

cyberpala

[fallahn]

What linker errors do you get? The map loader needs to be linked against zlib as well as sfml

[Ricky]

Hello,

Sorry to bother, I'm interested in the project and I'd like to learn how to use it however I can't even get it built!
It's probably something wrong on my side but here is the error cmake is giving me:

CMake Error at CMakeLists.txt:32 (install):
  install Library TARGETS given no DESTINATION!


CMake Error at cmake/Modules/FindSFML.cmake:307 (message):
  Could NOT find SFML (missing: SFML_SYSTEM_LIBRARY SFML_WINDOW_LIBRARY
  SFML_GRAPHICS_LIBRARY)
Call Stack (most recent call first):
  CMakeLists.txt:40 (find_package)

[eXpl0it3r]

CMake Error at cmake/Modules/FindSFML.cmake:307 (message):
  Could NOT find SFML (missing: SFML_SYSTEM_LIBRARY SFML_WINDOW_LIBRARY
  SFML_GRAPHICS_LIBRARY)
Call Stack (most recent call first):
  CMakeLists.txt:40 (find_package)

It doesn't find SFML. You need to set SFML_ROOT in CMake to the root dir of SFML.

[crypto]

Hey,

There seems to be a bug in

TileQuad::Ptr MapLoader::m_AddTileToLayer(MapLayer& layer, sf::Uint16 x, sf::Uint16 y, sf::Uint32 gid, const sf::Vector2f& offset)

Where you do

//applying half pixel trick avoids artifacting when scrolling
v0.texCoords = m_tileInfo[gid].Coords[0] + sf::Vector2f(0.5f, 0.5f);
v1.texCoords = m_tileInfo[gid].Coords[1] + sf::Vector2f(-0.5f, 0.5f);
v2.texCoords = m_tileInfo[gid].Coords[2] + sf::Vector2f(-0.5f, -0.5f);
v3.texCoords = m_tileInfo[gid].Coords[3] + sf::Vector2f(0.5f, -0.5f);

If I run it as is, all my tiles have a half-pixel weirdness to them, but if I remove the extra half pixels, it works perfectly. I tried with tilesets that have no margin/padding and some that have a margin/padding, it happens with both. I'm using SFML-master (today's snapshot) and MinGW 4.8.1.

Before / After

[fallahn]

What happens when you scroll your view? The half pixel fix only really applies when a view is scaled or scrolled. How do the examples look when you compile them? They work fine on every machine I've tested, but I haven't performed an exhaustive set of tests. If the examples also exhibit this behaviour for you then it may be a graphics driver issue

[crypto]

I tried Benchmark.cpp and the issue is still present. However, if I remove the half pixels and maximize the window, scrolling will sometimes vertical gaps, is that what you meant?

I'll look at it more when I get back from work. I'm running Win8 64 bits with a Radeon HD5850, latest drivers.

[fallahn]

Yes, the gaps when scrolling is what I mean. It's a well documented artifact using vertex arrays (there are several posts on the forum about it). I don't have access to a Win8 machine, but I do have my suspicions about the latest AMD driver. When I updated from 13.9 to 13.12 on 2 machines I found that 1 in 3 times (more or less) I couldn't even create an SFML window without it hanging on a white screen. Updating to the latest beta drivers fixed this. AMD are obviously working on updating their OpenGL implementation too, as updating the drivers broke a few of my shaders which, as it turned out, was due to my shader code not being as strictly adherent to GLSL as they should be (so in this case the AMD update is a good thing). I'm not sure what to suggest as a fix unfortunately.