DynDOLOD 3.00 Alpha 196

[AiElias]

28 minutes ago, AiElias said:

Edit: So after moving replacing the textures to the snowpine leaves I once again saw a regression back to the less dense lod. So the issue is something related to the texture. But I'm not sure what since it is mipmapped.

So going one step further and copying the texture from the leaves of Pine01 and replacing the leave texture of Snow01, I'm still seeing the same issue which seems counter intuitive if it was an issue with the texture. Yet it is the case.

[sheson]

51 minutes ago, AiElias said:

What is going on here with my tree lods. I'm using hybrid lods. These are TreePineForest01 and TreePineForestSnow01.
To the left are the loaded trees. To the right the hybrid lods.

For some reason the snow01 hybrid lod crown looks less dense. This is despite the fact that the loaded tree models are exactly the same with only differing textures.
I have looked through my hybrid lod nifs and they are also appear the same with again only differing texture paths. Yet despite this, when I copy over the Pine01 hybrid nif and replace the Snow01 hybrid the right lod looks correct albeit with incorrect textures.

Edit: So after moving replacing the textures to the snowpine leaves I once again saw a regression back to the less dense lod. So the issue is something related to the texture. But I'm not sure what since it is mipmapped.

The different texture has mipmaps generated without alpha-to-coverage. They basically fade into full transparancy.

That is why it is suggested that the hyrbrid crowns should use the same shader and textures as the full models. DynDOLOD will automatically adjust the alpha threshold for LOD and create mipmaps with alpha coverage. Also it will ensure compatibility in case the full textures are replaced. Existing mipmaps are ignored at the time textures are added to the texture atlas. The mipmaps on the texture atlas will always be created from the largest resolution with alpha-to-coverage.

See ..\DynDOLOD\docs\trees.ultra\tools\DynDOLOD_CreateStaticTree.html and ..\DynDOLOD\docs\trees.ultra\tools\hybrids\DynDOLOD_CreateHybrid.html

[AiElias]

39 minutes ago, sheson said:

The different texture has mipmaps generated without alpha-to-coverage. They basically fade into full transparancy.

I'm probably not understanding something here but if this is the case, why does using the same diffuse on both lods result in a variation? It's because the diffuse crown of the loaded model for snow01 doesn't have alpha-to-coverage mipmap generation?

46 minutes ago, sheson said:

That is why it is suggested that the hyrbrid crowns should use the same shader and textures as the full models.

Is there an option in Dyn 3 at the moment to downscale crown textures? I see no reason that they should be their native 2048x2048 resolution.

[sheson]

30 minutes ago, AiElias said:

I'm probably not understanding something here but if this is the case, why does using the same diffuse on both lods result in a variation? It's because the diffuse crown of the loaded model for snow01 doesn't have alpha-to-coverage mipmap generation?

Is there an option in Dyn 3 at the moment to downscale crown textures? I see no reason that they should be their native 2048x2048 resolution.

If two shapes have the same shader settings and textures but the alpha looks different it is most likely because of the Alpha threshold set on the NiAlphaProperty.

"Max tile size full sets the maximum resolution a single full texture can occupy on the object LOD texture atlas. Typically set to 1/4 of the vertical screen resolution, e.g. 256 for 1080p/1440p, 512 for 2160p etc."

Pay attention to the fact that LOD may not use the texture on the texture atlas because of the UV.

[AiElias]

33 minutes ago, sheson said:

If two shapes have the same shader settings and textures but the alpha looks different it is most likely because of the Alpha threshold set on the NiAlphaProperty.

Ah I think I see the issue. I deleted the NiAlphaProperty entirely from the hybrid nifs.

[AiElias]

32 minutes ago, AiElias said:

Ah I think I see the issue. I deleted the NiAlphaProperty entirely from the hybrid nifs.

Ok great that was it.

I also think I discovered a potential bug. I'm not getting my tree lods being created in my flat test world at cells (0,0) and (X,-Y). (X, +Y ) seems to be generating correctly. Anything I could be doing wrong on my end?

Edited September 4, 2021 by AiElias

[sheson]

19 minutes ago, AiElias said:

Ok great that was it.

I also think I discovered a potential bug. I'm not getting my tree lods being created in my flat test world at cells (0,0) and (X,-Y). (X, +Y ) seems to be generating correctly. Anything I could be doing wrong on my end?

LOD is generated for references that have base records for which LOD assets are defined / discovered. The tools print messages and log files to check on these things.

[AiElias]

36 minutes ago, sheson said:

LOD is generated for references that have base records for which LOD assets are defined / discovered. The tools print messages and log files to check on these things.

I can tell you for certain that the base records of those trees definitely had their lods defined because they were generated elsewhere in the worldspace.

[sheson]

14 minutes ago, AiElias said:

I can tell you for certain that the base records of those trees definitely had their lods defined because they were generated elsewhere in the worldspace.

I you require help or want to troubleshoot a problem I suggest to provide actual information. Read the first post what log file to include when making posts.

[AiElias]

23 minutes ago, sheson said:

I you require help or want to troubleshoot a problem I suggest to provide actual information. Read the first post what log file to include when making posts.

Yea I always forget. Here.

The Debug_log was too big and I couldn't truncate. Here's link.
https://discord.com/channels/525782915028877323/610980524818694191/883824313717882920

DynDOLOD_SSE_log.txt

[sheson]

7 hours ago, AiElias said:

Yea I always forget. Here.

The Debug_log was too big and I couldn't truncate. Here's link.
https://discord.com/channels/525782915028877323/610980524818694191/883824313717882920

DynDOLOD_SSE_log.txt 36.63 kB · 0 downloads

The log prints 4 lines 

Potentially wild edit reference with z = 0.0
Wild edits often happen when the Creation Kit is used for creating mods. In the game they often manifest as floating objects in the middle of the map. Look up the form id in xEdit and remove the reference, clean the plugin afterwards. Notify mod author.

The message also means the 4 mentioned references are ignored for LOD. Change the z value slightly.

 

[maxis123]

Hello. 

Hello. After upgrading from alpha 35 to 39, Dyndolodx64.exe suddenly stopped working. Lod generation takes 50 seconds and completes successfully. What could be the reason?

Edited September 5, 2021 by maxis123

[PRieST]

Without the logs, mentioned in the first post, nobody can tell.

[maxis123]

log.txt

[SkyrimGirus]

In DynDOLOD 2, there was a setting in the oreset files, RemoveUnseenIgnoreWater, that when toggled on would prevent the program from culling underwater LOD.

This setting is not present in the preset files of DynDOLOD 3, nor does there appear to be another way to achieve the same ends using another setting.

Thus the question: Is there a means of preventing the culling of underwater LOD in DynDOLOD 3, either directly or more crudely by disabling the culling of all unseen faces?