Change Region / Mesh Path at Runtime

fallen

badlogic ha scritto
To be honest, I'm not sure I can rule out a bug in spine-c. Hard to tell from the limited information in your previous post. I'm happy with whatever you can throw at me!

Ok, thanks for the response. That helps put me on the right track

I will try to see if I can get the memory leak to occur with simpler scenarios, starting small and building up to see where it starts for the first time. Then I should be able to hand over a sample.

Appreciate the support

tonysyata

My solution based on cocos2dx:

void
CCSpineNode::slotChangeDisplay(const char* slotName, const char* imgPath) {
  if (!_spSkeleton || !slotName || !imgPath)
    return;

  spSlot* slot = spSkeleton_findSlot(_spSkeleton, slotName);
  if (!slot)
    return;

  spAttachment* attach = slot->attachment;
  if (!attach)
    return;

  spAttachmentType type = attach->type;
  spRegionAttachment* attachReg = NULL;
  spMeshAttachment* attachMesh = NULL;
  std::string regionName = "";
  if (type == SP_ATTACHMENT_REGION) {
    attachReg = SUB_CAST(spRegionAttachment, attach);
    spAtlasRegion* reg = (spAtlasRegion*)attachReg->rendererObject;
    regionName = reg->name;
  }
  else if (type == SP_ATTACHMENT_MESH) {
    attachMesh = SUB_CAST(spMeshAttachment, attach);
    spAtlasRegion* reg = (spAtlasRegion*)attachMesh->rendererObject;
    regionName = reg->name;
  }
  else
    return;

  std::string name = "";
  std::string path = "";
  char buff[1024];
  memset(buff, 0, sizeof(buff));
  strcpy(buff, imgPath);

  int i = strlen(buff);
  while (buff[i] != '/') {
    buff[i] = '\0';
    i

---

;
  }
  path = buff;

  name = imgPath;
  name = name.substr(path.length(), name.length()-path.length());
  memset(buff, 0, sizeof(buff));
  for (int i = 0, c = name.length(); i < c; i++) {
    if (name[i] == '.')
      break;
    buff[i] = name[i];
  }
  name = buff;

  CCSpineManager* mgr = CCSpineManager::sharedSpineManager();
  spAtlas* atlas = mgr->addAtlasWithImage(name.c_str(), path.c_str(),
                                          regionName.c_str());
  if (!atlas)
    return;

  spAtlasRegion* atlasReg = spAtlas_findRegion(atlas, regionName.c_str());
  if (!atlasReg)
    return;

  if (type == SP_ATTACHMENT_REGION) {
    attachReg->rendererObject = atlasReg;
    attachReg->regionOffsetX = atlasReg->offsetX;
    attachReg->regionOffsetY = atlasReg->offsetY;
    attachReg->regionWidth = atlasReg->width;
    attachReg->regionHeight = atlasReg->height;
    attachReg->regionOriginalWidth = atlasReg->originalWidth;
    attachReg->regionOriginalHeight = atlasReg->originalHeight;
    spRegionAttachment_setUVs(attachReg,
                              atlasReg->u,
                              atlasReg->v,
                              atlasReg->u2,
                              atlasReg->v2,
                              atlasReg->rotate);
  }
  else if (type == SP_ATTACHMENT_MESH) {
    attachMesh->rendererObject = atlasReg;
    attachMesh->regionU = atlasReg->u;
    attachMesh->regionV = atlasReg->v;
    attachMesh->regionU2 = atlasReg->u2;
    attachMesh->regionV2 = atlasReg->v2;
    attachMesh->regionRotate = atlasReg->rotate;
    attachMesh->regionOffsetX = atlasReg->offsetX;
    attachMesh->regionOffsetY = atlasReg->offsetY;
    attachMesh->regionWidth = atlasReg->width;
    attachMesh->regionHeight = atlasReg->height;
    attachMesh->regionOriginalWidth = atlasReg->originalWidth;
    attachMesh->regionOriginalHeight = atlasReg->originalHeight;
    spMeshAttachment_updateUVs(attachMesh);
  }
}

addAtlasWithImage: create a atlas use memory data

you do not need to create a new attachment, just change the rendererObject and uv info of the attachment

fallen

@tonysyata - thanks for posting your solution! This looks like it will do great, as long as you aren't caching the Skeleton object. I first implemented something like this, but found that since we cache all of our skeleton objects, and have 8 characters on screen sharing the same skeleton, that making direct changes to the slot would mean that all 8 characters end up with the same changed images.

By any chance are you having issues with memory leaks in your animations? I am still trying to figure that out / create a reproducible case that isn't linked into our entire game.

tonysyata

@[cancellato]
I just cache the "spAltas" and "spSkeletonJson" objects (in my project, i use json fromat)
”spSkeletonData“ object can not share in my solution
And not found memory leak yet

fallen

badlogic ha scritto
Hrm, it appears that the timelines in the loaded animation isn't disposed correctly. Could you provide me with a minimal C(++) reproduction case for this?

Think we finally figured it out

we were running a few animations we called "Setup XXX" that only had key frames on the 0 timeline.

Once we expanded these setup animations to have key frames on both 0 and 1, all memory leaks ceased to exist.

This might be a bug in spine-c, but we're happy with the way we were able to fix it

« Pagina precedente