{
  "$type": "site.standard.document",
  "bskyPostRef": {
    "cid": "bafyreiee6o73xcoduis5zawvkfcqnodscahn57urqugfx3ld43d623voi4",
    "uri": "at://did:plc:dxjzgxe7cvirxkwfjr2tjspt/app.bsky.feed.post/3mlvdg2cfhq52"
  },
  "path": "/t/problem-between-tiles/49591#post_2",
  "publishedAt": "2026-05-15T02:27:04.000Z",
  "site": "https://hub.jmonkeyengine.org",
  "textContent": "Probably the vertexes of one tile edge corners don’t line up _EXACTLY_ with the vertexes of the other tile edge corners. Like, exactly down to the maximum decimal place.\n\nWith tiles, it’s nearly impossible to get this perfect without some kind of quantization down in the vertex shader. (And even then, I’m not 100% convinced.)\n\n(Aside: Mythruna used to generate it’s geometry with a 100% seamless world, no matter what, you would never see gaps between two abutting squares. With the modern Mythruna and the modern JME, something changed and I see these gaps quite frequently now. In principal, I’m generating meshes exactly the same now as I was then and so I’ve always suspected that something changed in the JME pipeline with respect to vertex transforms… but I cannot prove it. But I will say all of my vertexes should be on integer bounds in these cases.)\n\nOr there is some edge-to-edge phenomenon that can happen in glsl that even I’m not familiar with. (Wouldn’t put it past the universe to have some low level left-edge vs right-edge small “don’t overlap pixels” optimization that was turned on.)\n\nIn any case, when I have gaps they are SUPER regular. Your gaps look slightly irregular so there may be some additional discrepancies in the math before it even gets to JME.\n\nYou may be able to do some clever debug output to see if they match down to the smallest decimal place or not.",
  "title": "Problem between tiles"
}