Multiple hardware buffers

[sjcomp]

I am trying to use separate hardware buffers for my mesh and I have a problem with the second buffer containing uv coordinates. I used forum threads: http://www.ogre3d.org/phpBB2/viewtopic.php?t=2893 and
http://www.ogre3d.org/phpBB2/viewtopic.php?t=4371
as my starting point.
The problem is that texture is not displayed correctly. The texture is multiplied multiple times accross the triangle, while it should be stretched. Any suggestions about what I am doing wrong? Thanks a lot.
The code is as follows:

Code: Select all

		Ogre::MeshPtr mesh = Ogre::MeshManager::getSingleton().createManual("Mesh",Ogre::ResourceGroupManager::getSingleton().getWorldResourceGroupName());
		Ogre::SubMesh * sub = mesh->createSubMesh();
		sub->operationType = Ogre::RenderOperation::OT_TRIANGLE_LIST;
		sub->useSharedVertices = false;

		const unsigned int num_verts = 3;
		const unsigned int size_vert = 6;
		const unsigned int uv_size_vert = 2;
		const unsigned int total_verts_size = num_verts * size_vert;
		const unsigned int uv_total_verts_size = num_verts * uv_size_vert;

		Ogre::Real verts[total_verts_size] =
		{
			-100.0, -100.0, 0.0,  0.0, 0.0, 1.0,
			100.0, -100.0, 0.0,  0.0, 0.0, 1.0,
			0.0, 100.0, 0.0,  0.0, 0.0, 1.0,
		};
		Ogre::Real uv_verts[uv_total_verts_size] =
		{
			0.0, 0.0,
			1.0, 0.0,
			0.0, 1.0
		};
	    bool vertexShadowBuffer = true;
		Ogre::HardwareVertexBufferSharedPtr vbuf;
		vbuf = Ogre::HardwareBufferManager::getSingleton().createVertexBuffer
			(size_vert*sizeof(Ogre::Real),
			num_verts,
			Ogre::HardwareBuffer::HBU_STATIC_WRITE_ONLY,vertexShadowBuffer);

		Ogre::HardwareVertexBufferSharedPtr uv_vbuf = 
			Ogre::HardwareBufferManager::getSingleton().createVertexBuffer
				(uv_size_vert*sizeof(Ogre::Real),
				num_verts,
				Ogre::HardwareBuffer::HBU_STATIC_WRITE_ONLY,
				vertexShadowBuffer);

		Ogre::HardwareBuffer::Usage indexBufferUsage = Ogre::HardwareBuffer::HBU_STATIC_WRITE_ONLY;
	    
		Ogre::HardwareIndexBufferSharedPtr ibuf;
	    
		sub->indexData->indexCount = num_verts;
		sub->indexData->indexStart = 0;
		sub->indexData->indexBuffer = Ogre::HardwareBufferManager::getSingleton().
			createIndexBuffer(Ogre::HardwareIndexBuffer::IT_32BIT,
				sub->indexData->indexCount, indexBufferUsage);
		ibuf = sub->indexData->indexBuffer;
	    
		Ogre::Real *pReal = (Ogre::Real*) vbuf->lock(Ogre::HardwareBuffer::HBL_DISCARD);
		unsigned int* pIndexes = static_cast<unsigned int*>	(ibuf->lock(Ogre::HardwareBuffer::HBL_DISCARD) );

		for (int i = 0; i < total_verts_size; ++i)
		{
			*pReal++ = verts[i];
		}

		pReal = (Ogre::Real*)uv_vbuf->lock(Ogre::HardwareBuffer::HBL_DISCARD);
		// writing to a vbuf from uv_verts array
		for (int i = 0; i < uv_total_verts_size ; ++i)
		{
			*pReal++ = uv_verts[i];
		}

		*pIndexes++ = 0;
		*pIndexes++ = 1;
		*pIndexes++ = 2;

		ibuf->unlock();
		vbuf->unlock();
		uv_vbuf->unlock();

		sub->vertexData = new Ogre::VertexData();
		sub->vertexData->vertexStart = 0;
		sub->vertexData->vertexCount = num_verts;

		Ogre::VertexBufferBinding* bind = sub->vertexData->vertexBufferBinding;
		bind->setBinding(0, vbuf);
		bind->setBinding(1, vbuf);
	    
		size_t offset = 0;
		Ogre::VertexDeclaration* decl = sub->vertexData->vertexDeclaration;
		decl->addElement(0, offset, Ogre::VET_FLOAT3, Ogre::VES_POSITION);
		offset += Ogre::VertexElement::getTypeSize(Ogre::VET_FLOAT3);
		decl->addElement(0, offset, Ogre::VET_FLOAT3, Ogre::VES_NORMAL);
		decl->addElement(1, 0, Ogre::VET_FLOAT2, Ogre::VES_TEXTURE_COORDINATES, 0);

		mesh->_setBounds(Ogre::AxisAlignedBox(-100.0, -100.0, -100.0, 100.0, 100.0, 100.0)); 
		mesh->load();
		return mesh;

[genva]

Code: Select all

		bind->setBinding(0, vbuf);
		bind->setBinding(1, vbuf);

Should be

Code: Select all

		bind->setBinding(0, vbuf);
		bind->setBinding(1, uv_vbuf);

Another suggestion: Hardware vertext buffer floating-point value type is float, not Real. Just avoid potential issue when compile Ogre in double-precision mode, in this case Real is typedef'ed to double.

[sjcomp]

Code: Select all

bind->setBinding(1, uv_vbuf);

Thanks genva. Shame on me for such a miss.

Another suggestion: Hardware vertext buffer floating-point value type is float, not Real.

Does it mean that I should use float as in the following lines?

Code: Select all

float uv_verts[2] ={0.0f, 0.0f}; 
pFloat = (float*)uv_vbuf->lock(Ogre::HardwareBuffer::HBL_DISCARD);

instead of:

Code: Select all

Real uv_verts[2] ={0.0, 0.0}; 
pReal = (Real*)uv_vbuf->lock(Ogre::HardwareBuffer::HBL_DISCARD);

Example in wiki and all I have seen in forums were using Ogre::Real (wiki example). How can I make sure that I do not have dependancy on the compilation options of the Ogre?
Thanks a lot!

[genva]

Yes, use float access hardware vertex buffer float-point value always. The wiki is kinda out-of-date.

[sjcomp]

Thanks a lot, genva!