It is a great idea to have tile maps created from different-sized images! Currently I can take big PNG and cut it out by special script to set of images and create *.tsx file automatically which is recognizable by Tiled.
But I have one suggestion which you may be will implement in the near future: it will be great to disable autoplace tiles in tileset window and allow to place/move them manually (or pre-place them in *.tsx by script) to have same types of tiles grouped.
Quite an interesting suggestion! I guess it could be realized by either supporting the manual arrangement of images somehow, or by supporting your original image directly and you’d have to tell Tiled where your tiles are.
Do you like it that Tiled uses individual images here, or would you rather have it use your source image as-is? I think the latter would be much easier to implement.
“image” layer for my big image, which contains all my tiles;
“objects” layer where I put a rectangle over each specific tile (it is specifying tile boundaries);
“slopes” layer where I put a collision polygon.
I am running my custom script which gets this TMX as a parameter and making a set of PNG images (image per tile) and also TSX file, containing a list of all these images. The script puts all these files to subfolder.
I am connecting this TSX to my maps (via complex procedure of create/export empty set/reload Tiled) and I can see all tiles randomly shown in a list as I specified in my first message.
If it is possible to make TSX editor inside of Tiled, where I can specify tiles by rectangles and then using them to create maps - it would be fantastic.
But for the first step, it would be just good to have placements of tiles as they was placed on full original image. I could detect X,Y position of each tile by script and save it to TSX, so it will work.
Actually I don’t need for manual movement of tiles in set. An “original placement” support is just enough.
