1.5.0 changes

[Cham]

The comet clipping bug is solved ! Thanks Chris.

Now two more things about the comets rendering :

While very close to the comet nucleus, the front part of the "atmosphere" (or tail) just disappear in front of the central object while it's clearly visible at the rear. It should be visible on the front too. See the picture below :



The rendering is very ugly on the front, because of that clipping. This behavior isn't new with Celestia 1.5.0. Comets were always drawn this way since a long time ago. I think it should be corrected.

Also, the tail resolution is very poor. This is very understandable, since rising up the resolution could decrease significantly the frame rate performance (I tried several settings before, at the code level, and the impact on the frame rate can be very strong). But what about the parts I'm indicating with the red arrow, below ? Maybe I'm asking too much, but the polygons are way too much visible, IMHO (especially while moving around the comet). Is it possible to change that, without rising up the tail resolution ?



Now, about the star glare effect, there's a problem. I used the @ key to change the rendering code, and here's what I get :

The sun at a medium distance, without the @ key pressed. No glare effect :


Now the same view, after pressing the @ key. Huge effect, unrealistic and even violent to the eyes !

[ElChristou]

The problem is solved in the new star code, which will be the default very soon. To enable it now, press the @ key. The old star code is still required for cards that don't support point sprites. The manual toggle between old and new star rendering will be going away, and Celestia will automatically select the mode based on your graphics hardware's capabilities.

Is the new star code for OpenGL 2.0 only? with my GeForce4MX I'm unable to swith using @, I don't know if it's my card limitation, if it's the mapping of my Powerbook or else...

[chris]

The problem is solved in the new star code, which will be the default very soon. To enable it now, press the @ key. The old star code is still required for cards that don't support point sprites. The manual toggle between old and new star rendering will be going away, and Celestia will automatically select the mode based on your graphics hardware's capabilities.

Is the new star code for OpenGL 2.0 only? with my GeForce4MX I'm unable to swith using @, I don't know if it's my card limitation, if it's the mapping of my Powerbook or else...

The new star code should work with any graphics card that supports vertex shaders. The GeForce4MX doesn't support them in hardware, but it does have software emulation for vertex shaders that isn't too bad. Can you post a list of the of the OpenGL extensions supported by your card?

--Chris

[chris]

The comet clipping bug is solved ! Thanks Chris.

Now two more things about the comets rendering :

While very close to the comet nucleus, the front part of the "atmosphere" (or tail) just disappear in front of the central object while it's clearly visible at the rear. It should be visible on the front too. See the picture below :

The comet code needs a complete rewrite . . . I've never been that happy with it, and comet tails could now be made to look much better with OpenGL 2.0. I won't be able to do anything until after 1.5.0 though.

Also, the tail resolution is very poor. This is very understandable, since rising up the resolution could decrease significantly the frame rate performance (I tried several settings before, at the code level, and the impact on the frame rate can be very strong). But what about the parts I'm indicating with the red arrow, below ? Maybe I'm asking too much, but the polygons are way too much visible, IMHO (especially while moving around the comet). Is it possible to change that, without rising up the tail resolution ?

It shouldn't be that detrimental for performance to boost up the number of polygons in comet tails, especially if the number of polygons is made dependent on the size on screen of the comet.

Now, about the star glare effect, there's a problem. I used the @ key to change the rendering code, and here's what I get :

The sun at a medium distance, without the @ key pressed. No glare effect :


Now the same view, after pressing the @ key. Huge effect, unrealistic and even violent to the eyes !

Yes, there are still problems with stars in the 'scaled discs' mode.

--Chris

[Cham]

Chris,

I'm NOT using the 'scaled discs' mode at all ! Those pictures are in "dots" mode.

Actually, I'm getting this strong problem in all three star rendering modes.

[dirkpitt]

The new star code should work with any graphics card that supports vertex shaders. The GeForce4MX doesn't support them in hardware, but it does have software emulation for vertex shaders that isn't too bad. Can you post a list of the of the OpenGL extensions supported by your card?

ElChristou is running OS X 10.3.x, and the graphic drivers included with that OS version do not seem to properly support point sprites. On my machine (ATI 9700) for example, if I boot up OS X 10.3.9, point sprite sizes are erratic, appearing randomly as either 1-pixel dots or giant 10~20-pixel diameter discs. Explicitly setting the point size seems to put a brake on this behaviour, but results in all star glares having the same radius.

[ElChristou]

The new star code should work with any graphics card that supports vertex shaders. The GeForce4MX doesn't support them in hardware, but it does have software emulation for vertex shaders that isn't too bad. Can you post a list of the of the OpenGL extensions supported by your card?

Here:

Extensions:
GL_ARB_vertex_buffer_object: Support?©e
GL_ARB_texture_compression: Support?©e
GL_EXT_texture_compression_s3tc: Support?©e
GL_ARB_texture_border_clamp: -
GL_EXT_texture_edge_clamp: -
GL_EXT_texture_cube_map: -
GL_SGIS_generate_mipmap: Support?©e
GL_EXT_rescale_normal: Support?©e
GL_ARB_point_sprite: -

GL_ARB_vertex_program: Support?©e
GL_NV_vertex_program: -
GL_NV_fragment_program: -
GL_NV_register_combiners: Support?©e
GL_ARB_texture_env_dot3: Support?©e
GL_ARB_texture_env_combine: Support?©e
GL_EXT_texture_env_combine: -
GL_ARB_shading_language_100: -
GL_ARB_shader_objects: Support?©e
GL_ARB_vertex_shader: Support?©e
GL_ARB_fragment_shader: Support?©e

[chris]

Extensions:
GL_ARB_vertex_buffer_object: Support?©e
GL_ARB_texture_compression: Support?©e
GL_EXT_texture_compression_s3tc: Support?©e
GL_ARB_texture_border_clamp: -
GL_EXT_texture_edge_clamp: -
GL_EXT_texture_cube_map: -
GL_SGIS_generate_mipmap: Support?©e
GL_EXT_rescale_normal: Support?©e
GL_ARB_point_sprite: -

The problem is that point sprites are missing. How annoying . . . Don't worry though--I'll fix labels in the old star rendering path.

--Chris

[ElChristou]

Extensions:
GL_ARB_vertex_buffer_object: Support?©e
GL_ARB_texture_compression: Support?©e
GL_EXT_texture_compression_s3tc: Support?©e
GL_ARB_texture_border_clamp: -
GL_EXT_texture_edge_clamp: -
GL_EXT_texture_cube_map: -
GL_SGIS_generate_mipmap: Support?©e
GL_EXT_rescale_normal: Support?©e
GL_ARB_point_sprite: -

The problem is that point sprites are missing. How annoying . . . Don't worry though--I'll fix labels in the old star rendering path.

--Chris

One more reason to change my hardware...

[symaski62]

http://www.nvidia.fr/object/winxp_2k_93.71_fr.html

GeForce2 MX
GeForce2 MX 100
GeForce2 MX 200
GeForce2 MX 400
GeForce4 MX 420
GeForce4 MX 440
GeForce4 MX 440-SE
GeForce4 MX 440 with AGP8X
GeForce4 MX440SE with AGP8X
GeForce4 MX 460
GeForce2 MX Integrated graphics
GeForce4 MX Integrated graphics
GeForce4 MX 4000

The new star code should work with any graphics card that supports vertex shaders. The GeForce4MX doesn't support them in hardware, but it does have software emulation for vertex shaders that isn't too bad. Can you post a list of the of the OpenGL extensions supported by your card?

[dirkpitt]

Extensions:
GL_ARB_vertex_buffer_object: Support?©e
GL_ARB_texture_compression: Support?©e
GL_EXT_texture_compression_s3tc: Support?©e
GL_ARB_texture_border_clamp: -
GL_EXT_texture_edge_clamp: -
GL_EXT_texture_cube_map: -
GL_SGIS_generate_mipmap: Support?©e
GL_EXT_rescale_normal: Support?©e
GL_ARB_point_sprite: -

The problem is that point sprites are missing. How annoying . . . Don't worry though--I'll fix labels in the old star rendering path.

--Chris

One more reason to change my hardware...

Not so fast... here's my extension list on OS X 10.3.9. Notice that GL_ARB_point_sprite is present, but I still get incorrect rendering of point sprites. This is probably another driver bug that will never get fixed, but... (full extension list extracted with OpenGL Extensions Viewer

System Info
Mac OS X 10.3.9 (7W98)

Vendor
ATI Technologies Inc.
1.5 ATI-1.3.42

Renderer
ATI Radeon 9600 XT OpenGL Engine

Extensions
GL_APPLE_client_storage
GL_APPLE_element_array
GL_APPLE_fence
GL_APPLE_float_pixels
GL_APPLE_flush_render
GL_APPLE_packed_pixels
GL_APPLE_pixel_buffer
GL_APPLE_specular_vector
GL_APPLE_texture_range
GL_APPLE_transform_hint
GL_APPLE_vertex_array_object
GL_APPLE_vertex_array_range
GL_APPLE_vertex_program_evaluators
GL_APPLE_ycbcr_422
GL_ARB_depth_texture
GL_ARB_fragment_program
GL_ARB_fragment_shader
GL_ARB_imaging
GL_ARB_multisample
GL_ARB_multitexture
GL_ARB_occlusion_query
GL_ARB_point_parameters
GL_ARB_point_sprite
GL_ARB_shader_objects
GL_ARB_shadow
GL_ARB_shadow_ambient
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_crossbar
GL_ARB_texture_env_dot3
GL_ARB_texture_mirrored_repeat
GL_ARB_transpose_matrix
GL_ARB_vertex_blend
GL_ARB_vertex_buffer_object
GL_ARB_vertex_program
GL_ARB_vertex_shader
GL_ARB_window_pos
GL_ATI_blend_equation_separate
GL_ATI_blend_weighted_minmax
GL_ATI_separate_stencil
GL_ATI_text_fragment_shader
GL_ATI_texture_env_combine3
GL_ATI_texture_mirror_once
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_clip_volume_hint
GL_EXT_compiled_vertex_array
GL_EXT_draw_range_elements
GL_EXT_fog_coord
GL_EXT_multi_draw_arrays
GL_EXT_rescale_normal
GL_EXT_secondary_color
GL_EXT_separate_specular_color
GL_EXT_shadow_funcs
GL_EXT_stencil_wrap
GL_EXT_texture_compression_s3tc
GL_EXT_texture_env_add
GL_EXT_texture_filter_anisotropic
GL_EXT_texture_lod_bias
GL_EXT_texture_mirror_clamp
GL_EXT_texture_rectangle
GL_IBM_rasterpos_clip
GL_NV_blend_square
GL_NV_light_max_exponent
GL_NV_texgen_reflection
GL_SGI_color_matrix
GL_SGIS_generate_mipmap
GL_SGIS_texture_edge_clamp
GL_SGIS_texture_lod


Core features
v1.1 (100 % - 7/7)
v1.2 (100 % - 8/8)
v1.3 (100 % - 9/9)
v1.4 (100 % - 15/15)
v1.5 (100 % - 3/3)
v2.0 (66 % - 6/9)

OpenGL driver version check (Current: 1.5 ATI-1.3.42, Latest known: 1.5 ATI-1.3.36):
Latest version of display drivers found
According the database, you are running the latest display drivers for your video card.

[symaski62]

System Info
Mac OS X 10.3.9 (7W98)

Vendor
ATI Technologies Inc.
1.5 ATI-1.3.42

Renderer
ATI Radeon 9600 XT OpenGL Engine

yes !

http://ati.amd.com/support/driver.html

Mac OS X 10.3.x => Radeon

[dirkpitt]

System Info
Mac OS X 10.3.9 (7W98)

Vendor
ATI Technologies Inc.
1.5 ATI-1.3.42

Renderer
ATI Radeon 9600 XT OpenGL Engine

yes !

http://ati.amd.com/support/driver.html

Mac OS X 10.3.x => Radeon

Thanks for the tip, but Mac graphics drivers are not in general upgradeable by themselves.
Almost all driver upgrades for OS X are only possible by upgrading the OS itself. For example, point sprites are usable in OS X 10.4.x. I don't know if ElChristou's problem will be fixed by upgrading since we have different hardware.

[ElChristou]

Thanks for the tip, but Mac graphics drivers are not in general upgradeable by themselves.
Almost all driver upgrades for OS X are only possible by upgrading the OS itself. For example, point sprites are usable in OS X 10.4.x. I don't know if ElChristou's problem will be fixed by upgrading since we have different hardware.

Nop, definitively (using your analyse tool) the Geforce4MX cannot use point sprite...

I must change my hardware, but Chris is developing (if I'm not wrong) on the very last board from nvidia, so I'm wondering if I will be able to use all next features on the ATI Mobility Radeon X1600 of the actual MacBookPro... If it's not the case I don't know what to do... I need a laptop...

[chris]

Thanks for the tip, but Mac graphics drivers are not in general upgradeable by themselves.
Almost all driver upgrades for OS X are only possible by upgrading the OS itself. For example, point sprites are usable in OS X 10.4.x. I don't know if ElChristou's problem will be fixed by upgrading since we have different hardware.

Nop, definitively (using your analyse tool) the Geforce4MX cannot use point sprite...

I must change my hardware, but Chris is developing (if I'm not wrong) on the very last board from nvidia, so I'm wondering if I will be able to use all next features on the ATI Mobility Radeon X1600 of the actual MacBookPro... If it's not the case I don't know what to do... I need a laptop...

You should be fine with a Radeon X1600. It will be some time before there are features that are only useable on DX10 class hardware like a GeForce 8800--there's a lot of other Celestia work that needs to get done first

--Chris

[Cham]

Chris,

there's still a problem with a clipping plane on the foreground. Many planet orbits get clipped in various star systems :



The previous CVS compilation I made a month ago didn't had this problem. Now, it's back !

[chris]

About 7 : I think the white dots are related to FSAA too. The dots are very subtle but systematic. This is also a new bug with 1.5.0, and never occured with any other versions of Celestia (prior to 1.5.0).

Only Dirkpitt confirm the #7 bug ? Are both of us the only Mac users with an ATI card and FSAA enabled here ?

I'm seeing the bright dots when I have antialiasing enabled on my GeForce 8800. I think I understand why they're appearing. When multisampling is enabled, it's possible that one of the AA sampling points lies outside of the triangle. Centroid sampling was introduced to eliminate the artifacts that can result from sampling outside the triangle by guaranteeing that a sample lying inside the triangle will always be picked. This slide illustrates the idea:

http://pc.watch.impress.co.jp/docs/2004 ... igai12.jpg

Centroid sampling is supported in version 1.20 of the GL shading language, but I hope to find a solution that doesn't require it.

Edited: Hmm . . . I'm seeing similar artifacts in 1.4.1 when antialiasing is enabled. The dots appear even without OpenGL 2.0.

--Chris

[Cham]

Chris,

I never saw those white dots with Celestia 1.4.1. Antialiasing was perfect with that version.

[Sui Ota]

After addition of orbital movement to the Sol, when I select a location on an object (the Earth, for example) and press G key, the viewpoint is taken to strange place... It's very far place from the selected location!

[selden]

No matter what city name I type (London, Moscow, Tokyo), the viewpoint is positioned over the South Atlantic with the Earth at a distance of about 700,000 km. The distance to the selected city is shown as 50 km.

Unfortunately, the fields of the "GoTo Object" popup menu also aren't being filled in. either.