[solved]Buggy Swires Height Field problem

mamairaja

01-01-2011 14:05:30

First I wish a very happy new year to betajaen, spacegaier and nxogre users

Today I have tried Some heightfield operation with nxogre. (code from wiki article http://www.ogre3d.org/tikiwiki/tiki-index.php?page=BloodyMess+Ogre::Terrain+Heightfield)
problem found when it comes to
NxOgre::HeightField *hf = mhf.end(name.c_str());

I am using Buggy Swires.

P.S. - please note that i have edited the original code
NxOgre::ManualHeightField *mhf = OGRE_NEW_T(NxOgre::ManualHeightField, Ogre::MEMCATEGORY_GENERAL)();
to stack mem
NxOgre::ManualHeightField mhf;
because it gave me a compile error

Could you please provide a clue?

Thank you

~mamairaja

mamairaja

01-01-2011 15:10:19

Curious to ask

NxOgre::ManualHeightField::end(...);
function requires Path& parameter, and I am providing string "OgreTerrain/1114696642".

What does this parameter means?

Here, do I need to do something with Nxogre resource system?

~mamairaja

betajaen

01-01-2011 15:28:20

Yep.

Don't give it anything, or tell it to cook to a file; "file://myheightfield.xm"

mamairaja

01-01-2011 19:06:41

I thank you for replying

Yep.
Don't give it anything, or tell it to cook to a file; "file://myheightfield.xm"
have a little problem with understanding.

Does this mean keep it blank all the time ( mhf.end() ) and even don't try anything like mhf.end("file://myheightfield.xm") ?

I have tried both ways but problem still comes.

BTW, from where i can find if these lines are printed? Took them from NxOgre::ManualHeightField::end

NxOgre_DebugPrint_HeightFields("ManualHeightField -- end");

NxOgre_DebugPrint_HeightFields("Attempting to cook to resource");

NxOgre_DebugPrint_HeightFields("Attempting to read heightfield.");

betajaen

01-01-2011 20:12:17

You should be able to do both.

As for those debug lines, from NxOgreManualHeightField.cpp of course.

warkanum

02-01-2011 13:30:40

I have the exact same problem since I switched to Buggy Swires. If you find a solution, please post.

My code where crash happens:



NxOgre::ManualHeightField mhf;

	mhf.begin(size, size);

	Ogre::Real normMin = -32768.0f;

	Ogre::Real normMax = 32767.0f;



	for(int x = 0; x < size; ++x)

	{

		NxOgre::Enums::HeightFieldTesselation tess = NxOgre::Enums::HeightFieldTesselation_NW_SE;

		for(int z = size-1; z >= 0; --z)

		{

			Ogre::Real height = data[(size * z) + x];

			short sample = (short)(((height - minHeight) / (maxHeight - minHeight)) * (normMax - normMin) + normMin);

			mhf.sample(sample, 0, 0, tess);

			if(tess == NxOgre::Enums::HeightFieldTesselation_NE_SW)

				tess = NxOgre::Enums::HeightFieldTesselation_NW_SE;

			else

				tess = NxOgre::Enums::HeightFieldTesselation_NE_SW;

		}

	}



CRASH HERE:

NxOgre::HeightField *hf = mhf.end("ogre://Terrain/physheightfieldcache.xm");

OR

NxOgre::HeightField *hf = mhf.end("");

By the way, betajaen, I really like new nxogre and character controllers. Nice job and thanks.

betajaen

02-01-2011 14:02:04

Alright. I'll see if I can replicate the problem.

betajaen

02-01-2011 20:27:00

Okay, it seems to be the Memory Resource class. For now, switch to use file protocol, and it should work then.

i.e.

#include "NxOgre.h"



void main()

{



 NxOgre::World* world = NxOgre::World::createWorld();



  NxOgre::ManualHeightField hf;

  hf.begin(8,8);

  for (size_t i=0;i < 8;i++)

  {

   for (size_t j=0;j < 8;j++)

   {

    hf.sample(i + (j*8));

   }

  }

  NxOgre::HeightField* h = hf.end(true, "file://test.xm");



  NxOgre::World::destroyWorld();



}

I'll try and investigate the exact cause.

[Edit]

Other things like the ManualMesh will have a similar crash if you use "memory://", just switch to "file://whatever" and it'll be good to go.

warkanum

03-01-2011 10:52:11

Thanks,



NxOgre::HeightField *hf = mhf.end(true, "file://test.xm");

is working for me.

betajaen

03-01-2011 11:19:05

Once you have the heightfield cooked to a file, there is no need to cook it again. You can just load it in now instead.

warkanum

03-01-2011 19:31:47

Indeed, I find cooking it and using it that way much better. I think the problem is solved, well for the forum thread anyway. Did you find the cause of that memory problem?

betajaen

03-01-2011 19:38:51

No, I haven't investigated it yet.

It's a heap corruption, so it's likely to be in the MemoryResource::write function. I'm likely to replace the internals with a buffer, instead of the "home brew" I have there. My Buffer class is more reliable and tested.

mamairaja

07-01-2011 11:54:49

Oh...
My own question answered without any notification to me.
Thanks betajaen

yes it works

thanx

betajaen

08-01-2011 18:36:39

Think I fixed "memory://"

It's on the latest BuggySwires (Lemonade), if anyone is willing to try it out.

mamairaja

09-01-2011 08:00:23

Yes sir,
works great.

One more thing new nexogre seems to be require all physx prefixed dlls (eg - PhysXCooking.dll) in the same debug/release folder

otherwise
mWorld = NxOgre::World::createWorld();

gives runtime error

i am using physx 2.8.4.4

Thank you for the help

[edit]probably it's not nxogre requirement, it's physx requirement[/edit]