BETA Release OGLAClient Beta 100511: Bug hunt continues

[Quick_Nick]

Intel GPU.
The problem is that it reports OGL 1.4 as supported, yet it does not support FBO somehow.
Will have to make feature testing more aware of unusual cases.

Could you please run it again with this module:
http://orbides.1gb.ru/orbf/ogla-100506-1.zip
(put into modules directory)
EDIT: as well as http://orbides.1gb.ru/orbf/glgrcln.zip , posting glgr.txt
And post the log after it crashes?


Quite likely the case. Intel GPUs are flirting with standards like crazy, besides being generally underpowered.

I had to disable Norton to get glgrcln.exe to work, but did get it to work.

glgr.txt

GPU supports OpenGL 1.2
GPU supports OpenGL 1.4
GPU supports OpenGL 2.0
glgruva=1, usevbo=1, vboav=1, glgr_stensh_aupd=0, gl_comp_sup=1, gl_shm4=0
glGetCompressedTexImageARB=$100224B0
glGetCompressedTexImage=$100224B0
Extensions:
GL_EXT_blend_minmax GL_EXT_blend_subtract GL_EXT_blend_color GL_EXT_abgr GL_EXT_texture3D GL_EXT_clip_volume_hint GL_EXT_compiled_vertex_array GL_EXT_cull_vertex GL_SGIS_texture_edge_clamp GL_SGIS_generate_mipmap GL_EXT_draw_range_elements GL_SGIS_texture_lod GL_EXT_rescale_normal GL_EXT_packed_pixels GL_EXT_separate_specular_color GL_ARB_multitexture GL_EXT_texture_env_combine GL_EXT_bgra GL_EXT_blend_func_separate GL_EXT_secondary_color GL_EXT_fog_coord GL_EXT_texture_env_add GL_ARB_texture_cube_map GL_ARB_transpose_matrix GL_ARB_texture_env_add GL_IBM_texture_mirrored_repeat GL_EXT_multi_draw_arrays GL_NV_blend_square GL_ARB_texture_compression GL_3DFX_texture_compression_FXT1 GL_EXT_texture_filter_anisotropic GL_ARB_texture_border_clamp GL_ARB_point_parameters GL_ARB_texture_env_combine GL_ARB_texture_env_dot3 GL_ARB_texture_env_crossbar GL_EXT_texture_compression_s3tc GL_ARB_shadow GL_ARB_window_pos GL_EXT_shadow_funcs GL_EXT_stencil_wrap GL_ARB_vertex_program GL_ARB_fragment_program GL_EXT_stencil_two_side GL_ARB_vertex_buffer_object GL_EXT_texture_lod_bias GL_NV_texgen_reflection GL_ARB_depth_texture GL_WIN_swap_hint

Tried to run a scenario again with the new dll, but crashed when loading.
I included the whole whole ogla log since it sometimes shows different errors depending on when the error occurs.

 

[tl8]

Sorry, that means what exactly?

Sorry, it only happens on a new OGLA install

8/05/2010 8:51:07 AM|    INIT| OGLA v100506 (GLGR)
8/05/2010 8:51:27 AM|    INIT| GPU supports OpenGL 1.2
8/05/2010 8:51:27 AM|    INIT| GPU supports OpenGL 1.4
8/05/2010 8:51:27 AM|    INIT| GPU supports OpenGL 2.0
8/05/2010 8:51:27 AM|    INIT| glgruva=1, usevbo=1, vboav=1, glgr_stensh_aupd=0, gvsync=0, gl_comp_sup=1, gl_shm4=1
8/05/2010 8:51:27 AM| OGLADBG| OGLA 100506 Debug.
Vessel program 0:
Vertex shader was successfully compiled to run on hardware.

Fragment shader was successfully compiled to run on hardware.
Fragment shader(s) linked, vertex shader(s) linked. 
 
Vessel program 1:
Vertex shader was successfully compiled to run on hardware.

Fragment shader was successfully compiled to run on hardware.
Fragment shader(s) linked, vertex shader(s) linked. 
 
Vessel program 2:
Vertex shader was successfully compiled to run on hardware.

Fragment shader was successfully compiled to run on hardware.
Fragment shader(s) linked, vertex shader(s) linked. 
 
Vessel program 3:
Vertex shader was successfully compiled to run on hardware.

Fragment shader was successfully compiled to run on hardware.
Fragment shader(s) linked, vertex shader(s) linked. 
 
Vessel program 4:
Vertex shader was successfully compiled to run on hardware.

Fragment shader was successfully compiled to run on hardware.
Fragment shader(s) linked, vertex shader(s) linked. 
 
Planet program 0:
Vertex shader was successfully compiled to run on hardware.

Fragment shader was successfully compiled to run on hardware.
Fragment shader(s) linked, vertex shader(s) linked. 
 
Planet program 1:
Vertex shader was successfully compiled to run on hardware.

Fragment shader was successfully compiled to run on hardware.
Fragment shader(s) linked, vertex shader(s) linked. 
 
Planet program 2:
Vertex shader was successfully compiled to run on hardware.

Fragment shader was successfully compiled to run on hardware.
Fragment shader(s) linked, vertex shader(s) linked. 
 
Planet program 3:
Vertex shader was successfully compiled to run on hardware.

Fragment shader was successfully compiled to run on hardware.
Fragment shader(s) linked, vertex shader(s) linked. 
 
Planet program 4:
Vertex shader was successfully compiled to run on hardware.

Fragment shader was successfully compiled to run on hardware.
Fragment shader(s) linked, vertex shader(s) linked. 
 
Planet program 5:
Vertex shader was successfully compiled to run on hardware.

Fragment shader was successfully compiled to run on hardware.
Fragment shader(s) linked, vertex shader(s) linked. 
 
Planet program 6:
Vertex shader was successfully compiled to run on hardware.

Fragment shader was successfully compiled to run on hardware.
Fragment shader(s) linked, vertex shader(s) linked. 
 
Planet program 7:
Vertex shader was successfully compiled to run on hardware.

Fragment shader was successfully compiled to run on hardware.
Fragment shader(s) linked, vertex shader(s) linked. 
 
Planet program 8:
Vertex shader was successfully compiled to run on hardware.

Fragment shader was successfully compiled to run on hardware.
Fragment shader(s) linked, vertex shader(s) linked. 
 
Planet program 9:
Vertex shader was successfully compiled to run on hardware.

Fragment shader was successfully compiled to run on hardware.
Fragment shader(s) linked, vertex shader(s) linked. 
 
Haze program 0:
Vertex shader was successfully compiled to run on hardware.

Fragment shader was successfully compiled to run on hardware.
Fragment shader(s) linked, vertex shader(s) linked. 
 
Scatter program:
Vertex shader was successfully compiled to run on hardware.

Fragment shader was successfully compiled to run on hardware.
Fragment shader(s) linked, vertex shader(s) linked. 
 
Haze program 0:
Vertex shader was successfully compiled to run on hardware.

Fragment shader was successfully compiled to run on hardware.
Fragment shader(s) linked, vertex shader(s) linked. 
 
Scatter program:
Vertex shader was successfully compiled to run on hardware.

Fragment shader was successfully compiled to run on hardware.
Fragment shader(s) linked, vertex shader(s) linked. 
 
8/05/2010 8:51:30 AM|   PLNTR| Loading Textures/Saturn_ring.tex OK
8/05/2010 8:51:30 AM|   PLNTR| Loading Textures/Uranus_ring.tex OK
8/05/2010-8:51:31 AM:[getheihmap]:Notice: File "heightmaps\Earth-lv12-870-188-21-15.hei" not found or not loadable.
8/05/2010-8:51:31 AM:[getheihmap]:Notice: File "heightmaps\earth-lv13-429-340-9-8.hei" not found or not loadable.
8/05/2010-8:51:31 AM:[getcmap   ]:Notice: File "textures\Earth-lv13-483-104-11-9.bmp" not found or not loadable.
8/05/2010-8:51:31 AM:[getcmap   ]:Notice: File "textures\earth-lv8-119-94-3-3.bmp" not found or not loadable.
8/05/2010 8:51:32 AM|    PLNT| Loading Textures/Earth.tex OK
8/05/2010 8:51:32 AM|    PLNT| Loading Textures/Earth_cloud.tex OK
8/05/2010 8:51:32 AM|    PLNT| Loading Textures/Earth_lmask.tex OK
8/05/2010 8:51:32 AM|    PLNT| Loading Textures/Moon.tex OK
8/05/2010 8:51:37 AM|    INIT| OGLA v100506 (GLGR)

 

[anemazoso]

So I had not realized it but my work laptop has an ATI card and after seeing this thread I thought I would check it out and give some feedback.

So I tested both scenarios, switched from external to internal views as well as all cockpit views and did not experiance any problems.

One issue I did have is there was no atmosphere visible. Not sure if that is a related issue.

Laptop specs:

AMD Athalon 64 X2 dual core TK-42 1.6 GHz.
3GB RAM
64 bit Win7
ATI Radeon HD 3200, 256MB

And yes I am getting a low FSP on this machine but oh well

 

[Fixerger]

Was busy these days. What you need from Ati users? It runs without any crashes on my R9600. But it still has some problems... like:

+ Triangles problem? http://ipicture.ru/upload/100508/Ef2R1tfj99.jpg ;
+ Fonts readable... but still very poor;
+ MFD has some problems with drawings... empty lines... for example in MAP mfd or Orbit MFD...;
+ no shadows at all;
+ very bad performance with terrain (near 16 fps, that card can't held it anymore).

 

[Quick_Nick]

Strange, it should work at least as fast as DX7 client on old hardware (unless you want to see a load of advanced graphics, in which case it's logical).

Well, since it's an old computer, it's also a HORRIBLE graphics card. GeForce 6200 overclocked. In Orbiter 2006, it causes some strange graphical bugs. So it's not entirely surprising that the results it produces are unusual.

EDIT: Also, as soon as I can I'll get and post info for the duel-core (laptop) that DOES run things properly. I have a feeling it's not running everything right, but it does at least run. (some serious bugs when enabling shadows though)
Oh, and I love how I happened to get some 3D glasses in the mail recently. The DG looked decent in 3D, but the buildings near the VAB looked awesome! (place the ground camera at the VAB)

 

[Artlav]

OGLAClient 100508:
orbides.1gb.ru/orbf/oglaclient-100508_beta-100503.zip
Requires Orbiter beta 100503.
Raytraced haze tables: http://orbides.1gb.ru/orbf/ogla_earth_scatter.zip

Minor update, nothing big fixed, something might be added.

Changes:
*Feature detection cleanup, does not hang with Atlantis on old GPUs, should not crash on inconsistent ones.
*Continuing cleanup of code, might cause new bugs or swat others
*Major rebuild of 2D and GDI code, could fix and make bugs
*Added stereo eye distance control, and cleaned up stereo rendering

I had to disable Norton to get glgrcln.exe to work, but did get it to work.

Why use an "antivirus" anyway?

Tried to run a scenario again with the new dll, but crashed when loading.
I included the whole whole ogla log since it sometimes shows different errors depending on when the error occurs.

So, that GPU returns something for an extension it clearly shows as unsupported.

Please try current version, it should work, but without MFDs.
Second, try updating your video drivers (after trying), and see if MFDs would appear.

Oh, and I love how I happened to get some 3D glasses in the mail recently. The DG looked decent in 3D, but the buildings near the VAB looked awesome! (place the ground camera at the VAB)

Then, i've added an option for the eye distance (in meters) - default of 0.05 is good for DG cockpit, more - for building effects.
Also, the whole feature is now cleaned up and centred.

No but, I'm not using v1.25 but the latest code revision in the SSU SVN repository.

Is it any specific scenario that CTDs or just a special one.

Atlantis cockpit, crashes every odd time, only with OGLA.
Crashes every time with debugger attached, showing something inside MG_Atlantis.dll
I'm using SSU 1.25 from OH.
Last thing in orbiter.log:

(PanelO6::RegisterVC) Begin registration.
PanelR2: Getting signals
PanelR2: Getting signals
PanelR2: Getting signals
PanelF7::OnVCRedrawEvent called

Now, about that panel - could you tell me what exactly is being done to it?
It does not seem to be loaded with the mesh, but set into it latter.
I'm assuming it's c3.dds, there is a lot of blitting into it.

Or, better still, where can i get the sources, seeing SSU have GPL mentioned in it?

So I tested both scenarios, switched from external to internal views as well as all cockpit views and did not experiance any problems.

Good, looks like the old bug is finally gone.

One issue I did have is there was no atmosphere visible. Not sure if that is a related issue.

Related or not, i'd like to see details and a screenshot.
Does it look like something in the manual?

What you need from Ati users? It runs without any crashes on my R9600.

To look for crashes and BSODs where they were found before. Looks like there are none now.

+ Triangles problem? http://ipicture.ru/upload/100508/Ef2R1tfj99.jpg ;

How to reproduce? Or, does it just happen in DG every time?

+ Fonts readable... but still very poor;

Best i can do, for the moment.

+ MFD has some problems with drawings... empty lines... for example in MAP mfd or Orbit MFD...;

Screenshots please?
Does current one fixes any?

 

[Artlav]

artlav, could you please fix my raytrace atm 1fps problem? THX!

If you have an integrated graphics card of nVidia nForce 630M chipset (GeForce 7150M), then i don't think i can help much.

Unless the raytraced sky the only thing that is slow?
What about other advanced graphics settings (mapped shadows, air haze)?
Does just turning on advanced graphics make it drop to 1 FPS, or some specific feature bound to it?

 

[DaveS]

Or, better still, where can i get the sources, seeing SSU have GPL mentioned in it?

From here: http://sourceforge.net/projects/shuttleultra/develop

 

[Fixerger]

How to reproduce? Or, does it just happen in DG every time?

Every time, my video card has corrupted memory but it has problem only with that center MFD, works good on standard renderer.

Problem with empty lines is maybe fixed. I need to look at it more.

Crash found - when you press TGT in map mfd and choosing some target.

 

[anemazoso]

Here's the screenie. I should have known to provide one sorry

Also there is some wierd behaviour with the stock Orbiter F4 and F3 menus. When you press F4 for example the menu does not pop up unless you move your mouse cursor over the area where it is. And when it does appear it does not have a X button in the top right corner of the menu to close it although if you click where the X is suppose to be the given menu will close. This issue is also circled red by me in the screenie.

Hope this helps.

 

[Artlav]

To all testers:
Please look at the screenshot below.
Do you get the same kind of holes around MFD buttons, and were they there since which beta?

Every time, my video card has corrupted memory but it has problem only with that center MFD, works good on standard renderer.

Probably that's the case, the memory is likely distributed differently in the different renderers.

Crash found - when you press TGT in map mfd and choosing some target.

And select spaceport? Known, probably something core-related.

Also there is some wierd behaviour with the stock Orbiter F4 and F3 menus. When you press F4 for example the menu does not pop up unless you move your mouse cursor over the area where it is. And when it does appear it does not have a X button in the top right corner of the menu to close it although if you click where the X is suppose to be the given menu will close. This issue is also circled red by me in the screenie.

Both knows, lack of X is something inherently wrong in the new Orbiter, the non-upping is something undefindedly wrong in full-screen OGLA.

 

[anemazoso]

OK, went to check what you posted last. Tried to open Orbiter_ng. The frist thing that happened is the Orbiter icon appeared on the start bar at the bottom but would not open the launch pad. After right clicking on it to select it I got a "Unkown Author" screen for which I clicked "Run". The Launch pad then opened but without any scenarios and the Video button was missing. See attached screenie.

 

[Fixerger]

I dont have a problem with empty space near buttons.

 

[Artlav]

From here: http://sourceforge.net/projects/shuttleultra/develop

Okay, after a few hours of vandalising an abbreviation-rich 1 Mb of code into a 31-line test case, i found three problems, 2 in SSU, one in OGLA.
The one in OGLA was an assumption that add-ons are unanimously correct.
The ones in SSU were lack of texture flag and a weird assignment of texture coordinates.

So, you basically have two options to fix the C3 panel:
A. Meshes\SSU\VC.msh -> Put D after SSU/C3.dds texture, so it won't be loaded compressed.
or B. Use the update for OGLA provided below (into modules directory).
http://orbides.1gb.ru/orbf/ogla-100508-1.zip

The MFDs would stay white if off. Not sure where to look for a fix, so best case is to make them black in the mesh (EDIT: making them useless... not a solution).

Now, why does not the payload bay show up...?
Do you have any ideas?

Also, that VC is a damn awesome mesh, i tried it with 3D glasses, and it looks very cool, good job!

OK, went to check what you posted last. Tried to open Orbiter_ng. The frist thing that happened is the Orbiter icon appeared on the start bar at the bottom but would not open the launch pad.

Uh, how sure are you that you set up everything right?
I'm not familiar with such a failure mode of my software.

 

[DaveS]

Now, why does not the payload bay show up...?
Do you have any ideas

Well, this is the line used to set up the rendering of the payload bay as it's part of the main orbiter mesh:

mesh_orbiter = AddMesh (hOrbiterMesh, &ofs);
    SetMeshVisibilityMode (mesh_orbiter, MESHVIS_EXTERNAL|MESHVIS_VC|MESHVIS_EXTPASS);

It starts at line 2480 in Atlantis.cpp. The only difference between the same lines in the default Atlantis and the SSU is that we have the additional MESHVIS_EXTPASS to make sure that any vessels won't get blocked by the rendering of the payload bay.

Could this be the culprit?

Also, that VC is a damn awesome mesh, i tried it with 3D glasses, and it looks very cool, good job!
The ones in SSU were lack of texture flag and a weird assignment of texture coordinates.

Well, the VC is Donamy's work so all the credit goes to him!

 

[FordPrefect]

Artlav, I also do NOT get the empty spaces near the MFD buttons in the DG 2D cockpit. I used your latest OGLA client http://orbides.1gb.ru/orbf/oglaclien...eta-100503.zip

 

[DaveS]

Okay, after a few hours of vandalising an abbreviation-rich 1 Mb of code into a 31-line test case, i found three problems, 2 in SSU, one in OGLA.
The one in OGLA was an assumption that add-ons are unanimously correct.
The ones in SSU were lack of texture flag and a weird assignment of texture coordinates.

This is now fixed! So need for the special OGLA. With the addition of the D to the texture lines, things work nicely again! Thanks for your assistance!

 

[anemazoso]

Uh, how sure are you that you set up everything right?
I'm not familiar with such a failure mode of my software.

I just looked through everything and verified that everything is correct for installation. The wierd thing is it was working this morning (that's when I took the no-atmosphere screen shot) then later in the day I tried to start it up and it didn't work. I didn't change anything.

After it started doing that I downloaded the latest beta from your site and installed it. Now it's not even opening anything. It's just an icon and won't open a window.

I'm going to delete everything including Orbiter and reinstall everything.

---------- Post added at 02:53 PM ---------- Previous post was at 02:19 PM ----------

Ok, after deleting the whole Orbiter 2010 directory and reinstalling both Orbiter and OGLA. Everting is working fine so far. The atmosphere works now for me too.

 

[computerex]

Testing with:

AMD Turion(tm) II Dual-Core Mobile M520 (2 CPUs), ~2.3GHz 4096MB RAM
AMD M880G with ATI Mobility Radeon HD 4200

orbiter100503beta.zip
oglaclient-100508_beta-100503.zip

strange artifacts on the top left:

GDI issues:

Nothing seemingly wrong:

I think this is because of switching focus to ms paint to post the screenie:

That is because [nomedia="http://orbithangar.com/searchid.php?ID=3279"]Screen capture[/nomedia] produced this:

any plans to add a little cross on the top right hand corner?

same scenario:

Aside from the fact that the text looks horrible (no smoothing) this works nicely. Frame rates aren't too bad either

 

[Artlav]

OGLAClient 100509:
orbides.1gb.ru/orbf/oglaclient-100509_beta-100503.zip
Requires Orbiter beta 100503.
Raytraced haze tables: http://orbides.1gb.ru/orbf/ogla_earth_scatter.zip

Minor update, some visual fixes.

Changes:
*Calibrated 2D graphics to near-perfect match (Shuttle-A panel, many MFDs)
*Fixed texture coordinates assumptions (SSU panel C3)
*Fixed internal mesh visibility issues (SSU VC&PDB)

The only difference between the same lines in the default Atlantis and the SSU is that we have the additional MESHVIS_EXTPASS to make sure that any vessels won't get blocked by the rendering of the payload bay.

Could this be the culprit?

It is, somehow the two got mutually exclusive. Should be fixed now, as well as the docking view (that does not work in DX7).
Also, there are two "camera" views on the back panel - a wing and the docking port. What about an option of making them actual cameras? They don't even need to run all the time with 50% loss per one, 1 update per second won't affect FPS much.

strange artifacts on the top left:

Are they still there?

Aside from the fact that the text looks horrible (no smoothing) this works nicely.

With smoothing it would look like what you see in the 100501, unreadable at all.

EDIT:

I think this is because of switching focus to ms paint to post the screenie:
That is because Screen capture produced this:

Screen capture won't work for lack of GDI, and there is no use either - Ctrl+V makes a screenshot from inside.