Texture Compression Problems

[Faust]

Hi,

i moved our engine from jogl / joal to lwjgl a few weeks ago. Yesterday a tester detected errors when using dds textures. First i thought it has to be a hardware problem (graphics hardware does not support the texture compression format), but as i figured out today, the problem exists on every hardware.

I cannot find the mistake i am making and also double checked if something went wrong when the engine was ported.
The current system i am working with is a nvidia 7600 GS with the following information provided by the driver:

OpenGL Version : 2.1.2
OpenGL Vendor  : NVIDIA Corporation
OpenGL Renderer: GeForce 7600 GS/PCI/SSE2

Available Texture Compression:
COMPRESSED_RGB_S3TC_DXT1_EXT(33776)
COMPRESSED_RGBA_S3TC_DXT3_EXT(33778)
COMPRESSED_RGBA_S3TC_DXT5_EXT(33779) 

--- Extensions ---
GL_ARB_color_buffer_float
GL_ARB_depth_texture
GL_ARB_draw_buffers
GL_ARB_fragment_program
GL_ARB_fragment_program_shadow
GL_ARB_fragment_shader
GL_ARB_half_float_pixel
GL_ARB_imaging
GL_ARB_multisample
GL_ARB_multitexture
GL_ARB_occlusion_query
GL_ARB_pixel_buffer_object
GL_ARB_point_parameters
GL_ARB_point_sprite
GL_ARB_shader_objects
GL_ARB_shading_language_100
GL_ARB_shadow
GL_ARB_texture_border_clamp
GL_ARB_texture_compression
GL_ARB_texture_cube_map
GL_ARB_texture_env_add
GL_ARB_texture_env_combine
GL_ARB_texture_env_dot3
GL_ARB_texture_float
GL_ARB_texture_mirrored_repeat
GL_ARB_texture_non_power_of_two
GL_ARB_texture_rectangle
GL_ARB_transpose_matrix
GL_ARB_vertex_buffer_object
GL_ARB_vertex_program
GL_ARB_vertex_shader
GL_ARB_window_pos
GL_ATI_draw_buffers
GL_ATI_texture_float
GL_ATI_texture_mirror_once
GL_EXT_Cg_shader
GL_EXT_abgr
GL_EXT_bgra
GL_EXT_blend_color
GL_EXT_blend_equation_separate
GL_EXT_blend_func_separate
GL_EXT_blend_minmax
GL_EXT_blend_subtract
GL_EXT_compiled_vertex_array
GL_EXT_depth_bounds_test
GL_EXT_draw_range_elements
GL_EXT_fog_coord
GL_EXT_framebuffer_blit
GL_EXT_framebuffer_multisample
GL_EXT_framebuffer_object
GL_EXT_gpu_program_parameters
GL_EXT_multi_draw_arrays
GL_EXT_packed_depth_stencil
GL_EXT_packed_pixels
GL_EXT_pixel_buffer_object
GL_EXT_point_parameters
GL_EXT_rescale_normal
GL_EXT_secondary_color
GL_EXT_separate_specular_color
GL_EXT_shadow_funcs
GL_EXT_stencil_two_side
GL_EXT_stencil_wrap
GL_EXT_texture3D
GL_EXT_texture_compression_s3tc
GL_EXT_texture_cube_map
GL_EXT_texture_edge_clamp
GL_EXT_texture_env_add
GL_EXT_texture_env_combine
GL_EXT_texture_env_dot3
GL_EXT_texture_filter_anisotropic
GL_EXT_texture_lod_bias
GL_EXT_texture_mirror_clamp
GL_EXT_texture_object
GL_EXT_texture_sRGB
GL_EXT_timer_query
GL_EXT_vertex_array
GL_IBM_rasterpos_clip
GL_IBM_texture_mirrored_repeat
GL_KTX_buffer_region
GL_NV_blend_square
GL_NV_copy_depth_to_color
GL_NV_depth_clamp
GL_NV_fence
GL_NV_float_buffer
GL_NV_fog_distance
GL_NV_fragment_program
GL_NV_fragment_program_option
GL_NV_fragment_program2
GL_NV_framebuffer_multisample_coverage
GL_NV_half_float
GL_NV_light_max_exponent
GL_NV_multisample_filter_hint
GL_NV_occlusion_query
GL_NV_packed_depth_stencil
GL_NV_pixel_data_range
GL_NV_point_sprite
GL_NV_primitive_restart
GL_NV_register_combiners
GL_NV_register_combiners2
GL_NV_texgen_reflection
GL_NV_texture_compression_vtc
GL_NV_texture_env_combine4
GL_NV_texture_expand_normal
GL_NV_texture_rectangle
GL_NV_texture_shader
GL_NV_texture_shader2
GL_NV_texture_shader3
GL_NV_vertex_array_range
GL_NV_vertex_array_range2
GL_NV_vertex_program
GL_NV_vertex_program1_1
GL_NV_vertex_program2
GL_NV_vertex_program2_option
GL_NV_vertex_program3
GL_NVX_conditional_render
GL_S3_s3tc
GL_SGIS_generate_mipmap
GL_SGIS_texture_lod
GL_SGIX_depth_texture
GL_SGIX_shadow
GL_SUN_slice_accum
GL_WIN_swap_hint

The texture format is DDS_PixelFormat.FOURCC_DXT1 so i am using EXTTextureCompressionS3TC.GL_COMPRESSED_RGB_S3TC_DXT1_EXT as internal format for opengl. The "bb" byte buffer of the image data is in native order and direct. To upload the image data i use

GL11.glBindTexture(texType, id);
GL11.glTexParameteri(texType, GL11.GL_TEXTURE_MIN_FILTER, minFilter.getID());
GL11.glTexParameteri(texType, GL11.GL_TEXTURE_MAG_FILTER, magFilter.getID());
GL11.glTexParameteri(texType, GL11.GL_TEXTURE_WRAP_S, wrapMode_s.getID());
GL11.glTexParameteri(texType, GL11.GL_TEXTURE_WRAP_T, wrapMode_t.getID());
ARBTextureCompression.glCompressedTexImage2DARB(GL11.GL_TEXTURE_2D, i, cData.getFormat(), nWidth, nHeight, 0, bb);

nWidth and nHeight are both 256 and the buffers capacity, limit, position is correct.

When calling this, i get an GL11.GL_INVALID_OPERATION Error.
I read the methods docu http://www.opengl.org/sdk/docs/man/xhtml/glCompressedTexImage2D.xml but couldn't find out what the problem is. In fact, i worked the code did work before switching from jogl to lwjgl. By the way GL13.glCompressedTexImage2D results in the same error...

Any suggestions anyone? Would be great..

Faust.

[princec]

One gotcha in the port from JOGL->LWJGL is that LWJGL respects the position() and limit()s of the buffer you pass in. Have you remembered to flip() the buffer after filling it?

Cas

[Faust]

Hi, thanks for the suggestion. But yes, i tried it with flipped (position 0, limit within range 0 - capacity) and also with the rewinded buffer (position 0, limit == capacity). Does not change a thing...

[Faust]

Okay, i was currently creating a little applet for presentation purposes.

I took the out-of-date dds loader files from Christian Cohnen (http://209.85.129.132/search?q=cache:ZgRBsYG2kloJ:www.chriscohnen.de/cg/tutorials/DDStutorial/dds_tutorial.html+lwjgl+dds&cd=1&hl=de&ct=clnk&gl=de&client=firefox-a) and actually, now it is working...

i guess this is a good basis for debugging my software.

Thanks anyways.

[princec]

Are you sure the error is not from a call prior to the texture call?

Cas

[Faust]

Okay here comes the solution:

I use a ByteBuffer which contains all mipmap levels (instead of having ByteBuffers for each level). The jogl-call stops reading after width*height*blocksize whereas lwjgl reads from buffer position to buffer limit. If width*height*blocksize != buffer.limit() - buffer.position() in bytes then a invalid operation is generated.

I hate those kind of bugs... took me hours... and the solution is that simple. But when you persist that the code ran before and should run now you end up trapped in your own mind.