DynDOLOD 3.00 Alpha 198

[Jonado]

After updating to Alpha-198, all tree LODs just render as billboards, despite that I have set Level 4 tree LOD to use the Level0 model, checked Ultra Tree LOD, and have a lot of hybrid LOD models installed. I have never had any issues with this on previous DynDOLOD versions, despite using the same settings. I don't understand what is going on, so could you look into this?

Logs can be found here: https://www.dropbox.com/scl/fi/mhvt3nxzsjc1hb2dz2er2/Logs.7z?rlkey=zot7643q61cmmt0am8hi8zcoh&dl=0

EDIT: I figured it out. Most of my tree models had gotten a replacer with different checksums, and I forgot to disable those for LOD generation. You may ignore this report.

Edited yesterday at 01:23 AM by Jonado

[sheson]

6 hours ago, Jonado said:

After updating to Alpha-198, all tree LODs just render as billboards, despite that I have set Level 4 tree LOD to use the Level0 model, checked Ultra Tree LOD, and have a lot of hybrid LOD models installed. I have never had any issues with this on previous DynDOLOD versions, despite using the same settings. I don't understand what is going on, so could you look into this?

Logs can be found here: https://www.dropbox.com/scl/fi/mhvt3nxzsjc1hb2dz2er2/Logs.7z?rlkey=zot7643q61cmmt0am8hi8zcoh&dl=0

EDIT: I figured it out. Most of my tree models had gotten a replacer with different checksums, and I forgot to disable those for LOD generation. You may ignore this report.

Disabling something is a troubleshooting step and not really a fix. Whatever mod(s) replace tree full models should also include updated 3D tree LOD models with matching CRC32 in their filenames so everything works as intended.

[mostwanted11]

Before I start debugging things, can I ask if you've changed anything regarding tree lods and their mipmapping between alpha 196 and the latest version? 

Also is there a way to batch change the crowns of passthru files to have an alpha of 1 like you've told me before?

Edited yesterday at 03:27 PM by mostwanted11

[sheson]

1 hour ago, mostwanted11 said:

Before I start debugging things, can I ask if you've changed anything regarding tree lods and their mipmapping between alpha 196 and the latest version? 

Also is there a way to batch change the crowns of passthru files to have an alpha of 1 like you've told me before?

See https://dyndolod.info/Changelog which lists the changes between versions.

https://dyndolod.info/Official-DynDOLOD-Support-Forum
Report the actual problem or error message without making unverified assumptions or asking leading questions.

This script should work https://www.afkmods.com/index.php?/topic/5155-fnvedit-editing-alpha-flags-in-nif-files/#findComment-171206

[mostwanted11]

Logs: https://limewire.com/d/bDlNu#FVTlUVSG0k

Tree IDs: https://imgbox.com/g/Egxz38Zfxj

Mods used:

https://www.nexusmods.com/skyrimspecialedition/mods/161936?tab=description

https://www.nexusmods.com/skyrimspecialedition/mods/63604

https://www.nexusmods.com/skyrimspecialedition/mods/133619?tab=description

and pic of the issue in the attachment. Lods are way too thin from afar. This is after changing every passthru crown's alpha to 1. (the issue did not get resolved with or without that change)

Edited yesterday at 06:37 PM by mostwanted11

[sheson]

25 minutes ago, mostwanted11 said:

Logs: https://limewire.com/d/bDlNu#FVTlUVSG0k

Tree IDs: https://imgbox.com/g/Egxz38Zfxj

Mods used:

https://www.nexusmods.com/skyrimspecialedition/mods/161936?tab=description

https://www.nexusmods.com/skyrimspecialedition/mods/63604

https://www.nexusmods.com/skyrimspecialedition/mods/133619?tab=description

and pic of the issue in the attachment. Lods are way too thin from afar. This is after changing every passthru crown's alpha to 1. (the issue did not get resolved with or without that change)

The useful screenshots with more informative console show 3 different tree base record. The uploaded logs reports this:
xx01516E, spruce_forest_bigest01_summer.nif, uses unmodified spruce_forest_bigest01_summer_1b80523fpassthru_lod.nif with alpha threshold of 110
xx357AAC, spruce_common_big02_summer.nif, seems to use a modified version of spruce_common_big02_summer_lod_0.nif, it seems to set an alpha threshold of 1
xx408AC0, pruce_forest_slim01_summer.nif, uses unmodified spruce_forest_slim01_summer_16985565passthru_lod.nif with alpha threshold of 110

All of these trees seem to use the same spruce03_summer.dds for the crown. The object LOD atlas should have one version of spruce03_summer.dds adjusted to alpha threshold 1 and a second adjusted to 110.

Also upload L:\DynDOLOD\Edit Scripts\Export\LODGen_SSE_ObjectAtlasMap_Tamriel.lst and LODGen_SSE_ObjectAtlasMap_Tamriel.txt
From the generated output, upload ..\meshes\Terrain\Tamriel\Objects\Tamriel.4.4.-8.bto, ..\textures\DynDOLOD\LOD\DynDOLOD_Tamriel_Alpha.dds and DynDOLOD_Tamriel_Alpha_n.dds

[mostwanted11]

14 minutes ago, sheson said:

The useful screenshots with more informative console show 3 different tree base record. The uploaded logs reports this:
xx01516E, spruce_forest_bigest01_summer.nif, uses unmodified spruce_forest_bigest01_summer_1b80523fpassthru_lod.nif with alpha threshold of 110
xx357AAC, spruce_common_big02_summer.nif, seems to use a modified version of spruce_common_big02_summer_lod_0.nif, it seems to set an alpha threshold of 1
xx408AC0, pruce_forest_slim01_summer.nif, uses unmodified spruce_forest_slim01_summer_16985565passthru_lod.nif with alpha threshold of 110

All of these trees seem to use the same spruce03_summer.dds for the crown. The object LOD atlas should have one version of spruce03_summer.dds adjusted to alpha threshold 1 and a second adjusted to 110.

Also upload L:\DynDOLOD\Edit Scripts\Export\LODGen_SSE_ObjectAtlasMap_Tamriel.lst and LODGen_SSE_ObjectAtlasMap_Tamriel.txt
From the generated output, upload ..\meshes\Terrain\Tamriel\Objects\Tamriel.4.4.-8.bto, ..\textures\DynDOLOD\LOD\DynDOLOD_Tamriel_Alpha.dds and DynDOLOD_Tamriel_Alpha_n.dds

Quote

All of these trees seem to use the same spruce03_summer.dds for the crown. The object LOD atlas should have one version of spruce03_summer.dds adjusted to alpha threshold 1 and a second adjusted to 110.

I'm not sure if you're making a request or a conclusion sorry

Here are the files requested: https://limewire.com/d/u3RTp#wbcQRJAwrt

I have another issue, the grass lods are always properly lit at low sun angles (dawn/dusk) but sooo dark from 9am to 4pm. I tried changing the ComplexGrassBacklightMask to 0 or 90, it made no difference. Uploaded a pic without ENB or CS. 

Making the grass brighter would make the grass too bright at dawn/dusk, I uploaded logs for that worldspace

Edited yesterday at 07:34 PM by mostwanted11

[sheson]

12 hours ago, mostwanted11 said:

I'm not sure if you're making a request or a conclusion sorry

Here are the files requested: https://limewire.com/d/u3RTp#wbcQRJAwrt

I have another issue, the grass lods are always properly lit at low sun angles (dawn/dusk) but sooo dark from 9am to 4pm. I tried changing the ComplexGrassBacklightMask to 0 or 90, it made no difference. Uploaded a pic without ENB or CS. 

Making the grass brighter would make the grass too bright at dawn/dusk, I uploaded logs for that worldspace

I requested the files to verify if the result is matches what was reported about the assets in the logs.

There are 2 versions of spruce03_summer.dds on texture atlas, one with alpha channel unchanged for the LOD models that still set a threshold of 110 (left) and one with the alpha channel adjusted for the LOD models that set a threshold of 1 (right).

As you can see the right image is "thicker".

Here is how it looks 2 mipmap levels smaller:

The difference becomes less obvious. However, it is also not really thin either way. You could try to increase the AlphaFactor in the L:\DynDOLOD\Edit Scripts\DynDOLOD\DynDOLOD_SSE.ini. See what happens if you use 0.5 for example.

Otherwise you could create a custom LOD texture in your preferred image program and update the LOD models to use it directly. Though before doing that you might want to experiment using the full texture by adding either "usemipmaps" or "noatlas" to the Name of the BSTriShape "crown" - for example "crown noatlas" to see if it fades better without any alpha adjustment. See https://dyndolod.info/Help/3D-Tree-LOD-Model#Shape-Names

I also colored the two texture red and green in the atlas texture to verify that the BTO in fact uses them both depending on the different LOD models:

To summarize, not all 3D tree LOD have a threshold of 1. The threshold of 1 causes the alpha channel of the texture to be adjusted and to be added to the texture atlas and used. It might simply be not thick enough for your taste for this particular texture. Make sure that the output is really used in the game and not overwritten or wrong paths etc.

https://dyndolod.info/Help/Grass-LOD#Settings

When using complex grass and the side facing away from the light direction is too dark it can be brightened with the backlightmask. In the ..\DynDOLOD\Edit Scripts\DynDOLOD\DynDOLOD_SSE.ini set ComplexGrassBillboard=5 and set ComplexGrassBacklightMask to a value like 25, which means 25% of the light is applied to the side facing away from the light source. If the result is too dark, raise the value. If the result is too bright, lower the value.

If brightness of the grass billboard sides is fine when the sun is low, then backlighting must be working. Otherwise the grass billboard side facing the sun will be bright and its opposite side facing the other way would be too dark. So you would end up with a mix of bright and dark planes instead of a more uniform brightness. If you do not see any differences in the game between 0 and 100, then the NIF with the backlight is not used  - no complex grass billboards or the output is not active or being overwritten.

See this post https://stepmodifications.org/forum/topic/19903-dyndolod-300-alpha-198/page/717/#findComment-288563 how to verify backlighting is set/used.
This is the users follow up post: https://stepmodifications.org/forum/topic/19903-dyndolod-300-alpha-198/page/719/#findComment-288593

Since you set ComplexGrassBillboard=5, you could also test if this ..\DynDOLOD\Edit Scripts\DynDOLOD\  DynDOLOD_flat_4x2alt2_lod.nif works better with changing sun positions. Use expert mode to just Execute LODGen again to update object LOD meshes. See https://dyndolod.info/Help/Grass-LOD#Updating

 

[mostwanted11]

9 hours ago, sheson said:

I requested the files to verify if the result is matches what was reported about the assets in the logs.

There are 2 versions of spruce03_summer.dds on texture atlas, one with alpha channel unchanged for the LOD models that still set a threshold of 110 (left) and one with the alpha channel adjusted for the LOD models that set a threshold of 1 (right).

As you can see the right image is "thicker".

Here is how it looks 2 mipmap levels smaller:

The difference becomes less obvious. However, it is also not really thin either way. You could try to increase the AlphaFactor in the L:\DynDOLOD\Edit Scripts\DynDOLOD\DynDOLOD_SSE.ini. See what happens if you use 0.5 for example.

Otherwise you could create a custom LOD texture in your preferred image program and update the LOD models to use it directly. Though before doing that you might want to experiment using the full texture by adding either "usemipmaps" or "noatlas" to the Name of the BSTriShape "crown" - for example "crown noatlas" to see if it fades better without any alpha adjustment. See https://dyndolod.info/Help/3D-Tree-LOD-Model#Shape-Names

I also colored the two texture red and green in the atlas texture to verify that the BTO in fact uses them both depending on the different LOD models:

To summarize, not all 3D tree LOD have a threshold of 1. The threshold of 1 causes the alpha channel of the texture to be adjusted and to be added to the texture atlas and used. It might simply not be thick enough for your taste for this particular texture. Make sure that the output is really used in the game and not overwritten or wrong paths etc.

https://dyndolod.info/Help/Grass-LOD#Settings

When using complex grass and the side facing away from the light direction is too dark it can be brightened with the backlightmask. In the ..\DynDOLOD\Edit Scripts\DynDOLOD\DynDOLOD_SSE.ini set ComplexGrassBillboard=5 and set ComplexGrassBacklightMask to a value like 25, which means 25% of the light is applied to the side facing away from the light source. If the result is too dark, raise the value. If the result is too bright, lower the value.

If brightness of the grass billboard sides is fine when the sun is low, then backlighting must be working. Otherwise the grass billboard side facing the sun will be bright and its opposite side facing the other way would be too dark. So you would end up with a mix of bright and dark planes instead of a more uniform brightness. If you do not see any differences in the game between 0 and 100, then the NIF with the backlight is not used  - no complex grass billboards or the output is not active or being overwritten.

See this post https://stepmodifications.org/forum/topic/19903-dyndolod-300-alpha-198/page/717/#findComment-288563 how to verify backlighting is set/used.
This is the users follow up post: https://stepmodifications.org/forum/topic/19903-dyndolod-300-alpha-198/page/719/#findComment-288593

Since you set ComplexGrassBillboard=5, you could also test if this ..\DynDOLOD\Edit Scripts\DynDOLOD\  DynDOLOD_flat_4x2alt2_lod.nif works better with changing sun positions. Use expert mode to just Execute LODGen again to update object LOD meshes. See https://dyndolod.info/Help/Grass-LOD#Updating

 

Quote

If brightness of the grass billboard sides is fine when the sun is low, then backlighting must be working

It's fine regardless of what backlight value I use.

Quote

So you would end up with a mix of bright and dark planes

That's exactly what's happening yea

Quote

See this post https://stepmodifications.org/forum/topic/19903-dyndolod-300-alpha-198/page/717/#findComment-288563

My gray_70 is being used and it exists in the dyndolod output and its not being overwritten by anything. I don't know how to move to that cell to try if changing it to white.dds would work.

Do custom world spaces receive different grass lods somehow? I just tried grass lods in my custom grass lod testing world space and they looked good with billboard=1 under the same light and ENB but when I went into tamriel with those settings they look completely different. This never happened before

Logs: https://limewire.com/d/Sb7eD#jVGfcWbc80

DyndolodTestTundra is the name of the world space

Can I somehow PM you on discord for this issue instead of using the forum? I'd find a conversation much easier for me to troubleshoot this with you since I don't really grasp everything you say, but it's fine if you don't want to. I'll try the nif you provided next time maybe it'll help.

Here's a comparison of the same grass type under the same weather and ENB but in different world spaces: https://imgbox.com/g/ZSJAQOyBe9

 

I'll try your instructions for tree another time, thanks.

[mostwanted11]

https://streamable.com/fyw86b

Wth, that nif you gave me actually made the lods perfect or something. Thanks so much I can finally sleep lol

[mostwanted11]

The tree tweak worked(setting alpha to .5) I think that did it for me, they're like so much thicker now, thanks again!

Edited 14 hours ago by mostwanted11

[sheson]

8 hours ago, mostwanted11 said:

It's fine regardless of what backlight value I use.

That's exactly what's happening yea

My gray_70 is being used and it exists in the dyndolod output and its not being overwritten by anything. I don't know how to move to that cell to try if changing it to white.dds would work.

Do custom world spaces receive different grass lods somehow? I just tried grass lods in my custom grass lod testing world space and they looked good with billboard=1 under the same light and ENB but when I went into tamriel with those settings they look completely different. This never happened before

Logs: https://limewire.com/d/Sb7eD#jVGfcWbc80

DyndolodTestTundra is the name of the world space

Can I somehow PM you on discord for this issue instead of using the forum? I'd find a conversation much easier for me to troubleshoot this with you since I don't really grasp everything you say, but it's fine if you don't want to. I'll try the nif you provided next time maybe it'll help.

Here's a comparison of the same grass type under the same weather and ENB but in different world spaces: https://imgbox.com/g/ZSJAQOyBe9

 

I'll try your instructions for tree another time, thanks.

Grass LOD generation is not affected by any worldspace/weather settings and always the same. 

Not sure how grass LOD is different in the 2 screenshots. Next time use tfc to get closer to LOD.

7 hours ago, mostwanted11 said:

https://streamable.com/fyw86b

Wth, that nif you gave me actually made the lods perfect or something. Thanks so much I can finally sleep lol

Great. For future posterity, this is with ENB and complex grasses right?

4 minutes ago, mostwanted11 said:

The tree tweak worked, they're like so much thicker now, thanks again!

By tree tweak you mean changing the AlphaFactor to 0.5 or something else?

[mostwanted11]

1 hour ago, sheson said:

Grass LOD generation is not affected by any worldspace/weather settings and always the same. 

Not sure how grass LOD is different in the 2 screenshots. Next time use tfc to get closer to LOD.

Great. For future posterity, this is with ENB and complex grasses right?

By tree tweak you mean changing the AlphaFactor to 0.5 or something else?

Yes, enb and a complex grass mod with complexgrassbillboard=5

About the trees, so the two areas with the most glaring mipmap issues were the pine forest in falkreath/riverwood and the marsh in morthal. I changed some of the lod models in the pine forest to have an alpha of 1 and changed all the ones in the marsh but to no avail. I then set alphafactor to 0.5 and then went back and changed the rest of the lod models of the pineforest and re-ran dyndolod but this time everything looks right (in both regions, actually some other regions look better too) so I will say it must be changing alphafactor to 0.5 is what had the most visible effect.

I have a technical question, does playing in 4K or 1080p make a difference regarding mipmaps and whatnot? Does having a mid range card vs a high end one also make a difference?

Also I'm curious, what did you do with that nif you uploaded?

Edited 12 hours ago by mostwanted11

[sheson]

59 minutes ago, mostwanted11 said:

Yes, enb and a complex grass mod with complexgrassbillboard=5

About the trees, so the two areas with the most glaring mipmap issues were the pine forest in falkreath/riverwood and the marsh in morthal. I changed some of the lod models in the pine forest to have an alpha of 1 and changed all the ones in the marsh but to no avail. I then set alphafactor to 0.5 and then went back and changed the rest of the lod models of the pineforest and re-ran dyndolod but this time everything looks right (in both regions, actually some other regions look better too) so I will say it must be changing alphafactor to 0.5 is what had the most visible effect.

I have a technical question, does playing in 4K or 1080p make a difference regarding mipmaps and whatnot? Does having a mid range card vs a high end one also make a difference?

Also I'm curious, what did you do with that nif you uploaded?

See https://dyndolod.info/Help/Texture-Resolution. The higher the texture resolution, the more mipmap levels exist.
The hardware, driver and settings can control antialiasing techniques, their strength or quality, mipmap bias and whatnot that all affect image quality.

The NIF sets the SphereNormals shape name to change the direction of the normal vectors. Basically, the planes receive more light from above than the actual direction it faces.
See https://dyndolod.info/Help/3D-Tree-LOD-Model#Shape-Names and the last paragraph and image of https://dyndolod.info/Help/Ultra-Tree-LOD#Internal-Billboards-and-External-Billboards