Bug reports

[rgcotl]

sorry havet time to test it earlyer
but 80e didint fixed the background colors around the aplha channels
even more i believe it a bit more shows background color around the plume than before it was on 80d

edit:Alpha Test value didnt gived the wanted result either
the efect was the same exept with the 3760    but the plume then looked very much pixelated something like 1 bit for aplha channel

 

[mtarini]

80e didint fixed the background colors around the aplha channels
even more i believe it a bit more shows background color around the plume than before it was on 80d

Yes, so apparently OpenBRF does a good job.

The "background color around <transparent> things" is exactly the problem that you  address when you tune the "alpha test" part of the material flags.
OpenBRF is trying to let you preview the problem before you see it in-game.

The idea is: tune the flags until it looks good in OpenBRF, then it hopefully will be the same in-game.
Higher "alpha-test" values = less background around transparent things.
Too high "alpha-test" values = transparent things become too thin.

(That's the purpose of the preview provided by OpenBRF: not trying to make objects look good, but trying to make them look as good or as bad as they will be in-game. )

(Naturally it is only an approximation, mainly because: (1) it would be too work intensive to replicate the exact behavior of game shaders,  (2) the conditions,e.g. daylight, background, are different and (3) the exact meaning of the various flags and setting is approximated.)

 

[rgcotl]

well 77b did a good job with it
and plumes in game looks all ok (defnatly it dont included OpenBRF background color)

 

[mtarini]

well 77b did a good job with it

Once again, let me restate that it is not a matter of doing a "good" or a "bad" job per se:
it is a matter of doing a job which is as good, or as bad, as the one the game will do.

If you there are problems with the material flags (e.g. wrong alpha value), or with the texture,
I want OpenBRF to replicate the problems, not to look good. That's what a preview is for.

and plumes in game looks all ok (defnatly it dont included OpenBRF background color)

Sure, if things go as I expect, then sub-optimal material settings will result in some background "bleeding" inside the plumes.

In OpenBRF, "background" means background color.

In game, "background" would mean whatever happens to be on the screen when the plumes is rendered, e.g. maybe terrain (that is why you sometime want to use the "render order" material settings too).

 

[mtarini]

Here is a little alpha-test-flag testing with native meshes:

Here is how a native tree model looks like, according to OpenBRF, with
Alpha test = "8/256", "136/256", and "251/256", respectively i.e. the three settings you can set on the flags.

Spoiler

A close up:

Spoiler

(Native uses "136/256" for the material of this mesh: the middle image)

In this case:
- Small Alpha test value (8/256) = background problems (left)
- Medium Alpha test value (136/256) = decent (middle)
- Large Alpha test value (251/256) = too thin, disappearing branches, pixelated look (right)

I wish I had to patience and the time to modify the files and run the game so to test to which extent
the preview by OpenBRF replicate faithfully (or not) the game renderings.

 

[Slytacular]

I can not remember where in the forums I could find the solution to my problem, but my meshes in OpenBRF look normal, though when my models are shown in-game, some triangles stick out of place. Is it because I scaled my imported meshes from huge to small? For example, my helmet's aventail has two triangles sticking out of the end. I'll post a screenshot later.

 

[mtarini]

I'll post a screenshot later.

What a curious problem. Can you also post or pm me the brf file?

 

[Zimke Zlovoljni]

I can not remember where in the forums I could find the solution to my problem, but my meshes in OpenBRF look normal, though when my models are shown in-game, some triangles stick out of place. Is it because I scaled my imported meshes from huge to small? For example, my helmet's aventail has two triangles sticking out of the end. I'll post a screenshot later.

Are they rigged, maybe it's a rigging issue?

 

[cmpxchg8b]

Hey mtarini, small corrections to material flags:

0x40 Unknown (for alpha test?) -> Uniform lighting
0x80 Uniform lighting -> unused (?)
0x800 Blend factor -> Normalize normals
0x40000 LoD -> Allow batching
0x80000 unused (?) -> Allow batching if low poly
0x100 Blend, 0x200 Blend add and 0x400 Blend multiply are not actually flags but a 3 bit field. Here is a table of what the values do (src is object color and dst is framebuffer color):
0 (none): result = src
1 (alpha): result = src * src.a + dst * (1 - src.a)
2 (add alpha): result = src * src.a + dst
3 (multiply): result = src * dst
4 (add): result = src + dst
5 (undefined behavior)
6 (undefined behavior)
7 (auto-select): if material color is transparent it will use mode 1, else 0

 

[mtarini]

cmpxchg8b:
SUPER-EXTRA-COOL thanks.

Any idea if these apply to WB and pre-WB M&B (i.e. ver 1.011) alike?

 

[cmpxchg8b]

They apply to Warband, I don't know about M&B.

 

[dunde]

I failed to separate mesh into sub-meshes with 0.80e. It pushed me back to 0.79.

 

[mtarini]

Thanks Dunde, but: how did it fail?

I tried to replicate the bug (in 0.80e):
the sub-meshes are created and listed, but aren't shown in the 3D preview screen.

Is that it?
If so: got it, will be fixed (thanks!)
  (it's a visualization problem: the newly created sub-meshes are there, but aren't shown if multi-view-mode is "auto")
  (switching mode to "combo" or "aside", or restarting OpenBRF, resurrects them).
  (the problem was there also before, but didn't always emerge because in 0.80e I've decided to set default view-mode to "auto" at start-up).

 

[dunde]

Yes, it 's like that. Thanks mtarini. I can see them now.

 

[Somebody]

There is a problem with the wireframe checkbox - it will not display when either Bump/Specular are checked.

 

[dunde]

and it still display the wireframe if it's not animated.

 

[mtarini]

There is a problem with the wireframe checkbox - it will not display when either Bump/Specular are checked.

Thanks, hotfixed! (in ver 0.0.81b)
BTW, did you notice that the wireframe looks better in the last version (0.0.81)?

and it still display the wireframe if it's not animated.

You meant that the above bug occurred also when the mesh wasn't animated? If so, fixed too!

 

[rgcotl]

great thig with alpha transparency

but the normal maps seems dont works

Spoiler: spec+normal map preview

 

[mtarini]

great thig with alpha transparency

but the normal maps seems dont works

Thanks for the report...
I'm having troubles replicating that bug.
Native objects look good enough on my computer (e.g. lamellar_armor_c in armors_e.brf).
Do they in yours?
Can I have that helmet model, complete with its material? Nice model btw!

 

[rgcotl]

thanks
it is included in my mod
http://forums.taleworlds.com/index.php/topic,249788.0.html

samnite_helmets.brf name is h2_samnite

native meshes looks all ok for me to
and any meshes with havent using using alpha channels as transparency seems looks ok to