Reach tree LOD troubles with Happy Little Trees

[SirDibble2]

Hey, so i've been trying to get LOD going with HLT and everything is generating ok, but about half the trees in the reach have what i've been referring to as the 'cherry blossom' effect, where all the branch tips in the LOD models form a white block that jarringly goes away as the full model loads in.

Following STEP guide for generating LOD, using HLT standard install, and the hybrid LOD add on. Did a new profile with only my utility/bug fix mods and HLT and LOd requirements to take pics and get logs without other stuff muddying it up.

1st pic is looking ahead at some of the LOD in question:

https://ibb.co/5MvQ0yN

2nd pic is walking forward just enough to get the closest trees to load in

https://ibb.co/b7cQ7k5

Last pic is just an aerial view looking down at the reach to show that only, if i'm not mistaken, the fully barren reach trees have this issue. Other LOD for reach trees looks fine.

https://ibb.co/tzN3VZz

I deleted all logs before i ran this, so they are all fresh if they're needed. Really just looking to see if there is a way to fix this. Thanks in advance!

 

Edited April 21, 2023 by SirDibble2

[z929669]

27 minutes ago, SirDibble2 said:

Hey, so i've been trying to get LOD going with HLT and everything is generating ok, but about half the trees in the reach have what i've been referring to as the 'cherry blossom' effect, where all the branch tips in the LOD models form a white block that jarringly goes away as the full model loads in.

Following STEP guide for generating LOD, using HLT standard install, and the hybrid LOD add on. Did a new profile with only my utility/bug fix mods and HLT and LOd requirements to take pics and get logs without other stuff muddying it up.

1st pic is looking ahead at some of the LOD in question:

https://ibb.co/5MvQ0yN

2nd pic is walking forward just enough to get the closest trees to load in

https://ibb.co/b7cQ7k5

Last pic is just an aerial view looking down at the reach to show that only, if i'm not mistaken, the fully barren reach trees have this issue. Other LOD for reach trees looks fine.

https://ibb.co/tzN3VZz

I deleted all logs before i ran this, so they are all fresh if they're needed. Really just looking to see if there is a way to fix this. Thanks in advance!

 

Thanks for the detail, but I think you forgot to post your logs as mentioned and as specified in the guidelines at top of this forum and DynDOLOD Alpha OP

[SirDibble2]

24 minutes ago, z929669 said:

Thanks for the detail, but I think you forgot to post your logs as mentioned and as specified in the guidelines at top of this forum and DynDOLOD Alpha OP

Edit: nevermind i used the interwebs to find a solution to how to share the logs... imagine that hahahaha

logs are all zipped here: https://ufile.io/74o6o0hs

 

Edited April 21, 2023 by SirDibble2

[sheson]

Read https://dyndolod.info/Official-DynDOLOD-Support-Forum, especially the part about screenshots.

However, you are using ultra tree LOD, which typically means the nearest LOD for trees are 3rd party 3D tree LOD models. If you have issues with 3rd party LOD assets you need to check with the mod author on their comments section/forum.

[SirDibble2]

1 hour ago, sheson said:

Read https://dyndolod.info/Official-DynDOLOD-Support-Forum, especially the part about screenshots.

However, you are using ultra tree LOD, which typically means the nearest LOD for trees are 3rd party 3D tree LOD models. If you have issues with 3rd party LOD assets you need to check with the mod author on their comments section/forum.

Hahahaha. The author is z929669 above and they directed me here. Sooooo who go to now?

Read through the article, think i did most of that except changing weather and getting a form ID. Would those things help diagnose any issues or does the issue lay completely with the mod authors provided assets?

Edited April 22, 2023 by SirDibble2

[sheson]

I have no idea what the problem is supposed to be in those screenshots.

Use the base record form id to look up the tree in the Tree_Report.txt.

https://dyndolod.info/Help/Ultra-Tree-LOD
While the LOD is generated, a log file with a detailed report for each tree is created to ..\DynDOLOD\Logs\DynDOLOD_[GAME MODE]_Tree_Report.txt. It will contain an entry for each which types of assets were found and used for the different LOD levels:

Since you set Level0 on the tree mesh mask rule, look up what the report says which NIF is used for Level:0. Check which textures it uses, Then check which mods the NIF and the used textures comes from by looking it up in MO2 right Window data tab.

[SirDibble2]

12 minutes ago, sheson said:

I have no idea what the problem is supposed to be in those screenshots.

Use the base record form id to look up the tree in the Tree_Report.txt.

https://dyndolod.info/Help/Ultra-Tree-LOD
While the LOD is generated, a log file with a detailed report for each tree is created to ..\DynDOLOD\Logs\DynDOLOD_[GAME MODE]_Tree_Report.txt. It will contain an entry for each which types of assets were found and used for the different LOD levels:

Since you set Level0 on the tree mesh mask rule, look up what the report says which NIF is used for Level:0. Check which textures it uses, Then check which mods the NIF and the used textures comes from by looking it up in MO2 right Window data tab.

I am so confused how neither you or z929669 can immediately see how odd the LOD is on the trees. They aren’t supposed to look like cherry blossoms with big white blocks on the tops, and the larger ones with leaves don’t, as can clearly be seen in the 3rd photo.

I assure you I don’t mean to be disrespectful or anything, and am willing to do my due diligence in finding the issue, so i will take the steps you deserve and check back in. But seriously, am i like way overthinking this or expecting the LOD to be more of an accurate match then is possible? To me the LOD very clearly looks like it has issues with the upper branches being white blocks, but neither you or z929669 seem to see the issue despite how clear i’ve tried to be. Starting to feel a little crazy hahahahaha 

Edited April 22, 2023 by SirDibble2

[DoubleYou]

I think you are merely noticing small differences based upon slight changes of the light reflecting off the normal maps between full and lod versions. I think you are expecting it to be more accurate than it is. I also think that the alpha on the LOD versions appears slightly thicker than in full, and that is probably overemphasizing it to you.

[SirDibble2]

9 minutes ago, DoubleYou said:

I think you are merely noticing small differences based upon slight changes of the light reflecting off the normal maps between full and lod versions. I think you are expecting it to be more accurate than it is. I also think that the alpha on the LOD versions appears slightly thicker than in full, and that is probably overemphasizing it to you.

Thank you. If I’m overthinking it then i will just call it here. I mean, literally everything else is gooorgeous. I can handle a little bit of imperfection hahaha.

 

I appreciate your opinion on that matter. Thanks to all of you for your patience and help.

[MisterMorden]

4 hours ago, SirDibble2 said:

Thank you. If I’m overthinking it then i will just call it here. I mean, literally everything else is gooorgeous. I can handle a little bit of imperfection hahaha.

 

I appreciate your opinion on that matter. Thanks to all of you for your patience and help.

You could try setting the full model for reach trees' level 4 lod and see if that bothers you less.  I've had to do this for specific types of trees instead of using billboards or level0/decimated 3d lod before to get the look I wanted.

Edited April 22, 2023 by MisterMorden

[SirDibble2]

1 hour ago, MisterMorden said:

You could try setting the full model for reach trees' level 4 lod and see if that bothers you less.  I've had to do this for specific types of trees instead of using billboards or level0/decimated 3d lod before to get the look I wanted.

That sounds like a great option. Which setting is that? I’ve seen in the rules section the full setting, but don’t recall seeing it under the lod 4 drop down for trees.

[z929669]

9 hours ago, DoubleYou said:

I think you are merely noticing small differences based upon slight changes of the light reflecting off the normal maps between full and lod versions. I think you are expecting it to be more accurate than it is. I also think that the alpha on the LOD versions appears slightly thicker than in full, and that is probably overemphasizing it to you.

This is pretty much what I was saying on Nexus. Reach trees are notoriously difficult to get to behave with respect to mipmap alpha threshold and/or the mesh NiAlphaproperty. In either case, it's something that would need to be fixed in HLT, but T4 isn't going to bother with it, since the full trees look fine.

9 hours ago, SirDibble2 said:

Thank you. If I’m overthinking it then i will just call it here. I mean, literally everything else is gooorgeous. I can handle a little bit of imperfection hahaha.

I appreciate your opinion on that matter. Thanks to all of you for your patience and help.

Fixing this would require messing with the texture alpha and possibly the mesh of the LOD model crown and regenerating that LOD with each change to check it's fixed. Either way, any changes would need to be checked in the full model to be sure it still looks right. THEN those changes would need to be made in the HLT mod itself AND the LOD tree would need to be re-made based on that.

A lot of work that either T4 or myself would need to do unless a user provided the fixed files to us to verify ... after verifying the root issue is reproducible in mine and T4's mod builds (which are undoubtedly quite different, BTW) in the first place and not caused or exacerbated by other factors like weather mods, ENB, etc.

I mainly wanted you to post here so that you could get suggestions from sheson about how best to investigate and sort it out in your build.

[SirDibble2]

1 hour ago, z929669 said:

This is pretty much what I was saying on Nexus. Reach trees are notoriously difficult to get to behave with respect to mipmap alpha threshold and/or the mesh NiAlphaproperty. In either case, it's something that would need to be fixed in HLT, but T4 isn't going to bother with it, since the full trees look fine.

Fixing this would require messing with the texture alpha and possibly the mesh of the LOD model crown and regenerating that LOD with each change to check it's fixed. Either way, any changes would need to be checked in the full model to be sure it still looks right. THEN those changes would need to be made in the HLT mod itself AND the LOD tree would need to be re-made based on that.

A lot of work that either T4 or myself would need to do unless a user provided the fixed files to us to verify ... after verifying the root issue is reproducible in mine and T4's mod builds (which are undoubtedly quite different, BTW) in the first place and not caused or exacerbated by other factors like weather mods, ENB, etc.

I mainly wanted you to post here so that you could get suggestions from sheson about how best to investigate and sort it out in your build.

Lordy lord. That is a lot of work. well, I appreciate everyone’s help, and am still happy with the rest of the tree and object lod, so no worries at all.

 

any advice on how to force those trees to use full models for lod4? Maybe i’ll give that aa shot.

[DoubleYou]

You would make a mesh mask rule and set lod4 to Full. I don't see much likelihood that this will help though, since I do believe that @z929669 uses the full crowns on the 3d hybrids anyway. There are two different lod trees available... decimated one and the full crowns, so switching between them, if you haven't already tried that, might yield different results. 

[z929669]

1 hour ago, DoubleYou said:

You would make a mesh mask rule and set lod4 to Full. I don't see much likelihood that this will help though, since I do believe that @z929669 uses the full crowns on the 3d hybrids anyway. There are two different lod trees available... decimated one and the full crowns, so switching between them, if you haven't already tried that, might yield different results. 

Yeah, the crowns are identical. I'm sure it's optimization of the mipmapping and possibly also the NiAlphaProperty. Reach trees are a ***** with the fat trunks and teeny, broom-like branches (or little groups of leaves on fat branches).

[sheson]

16 hours ago, SirDibble2 said:

I am so confused how neither you or z929669 can immediately see how odd the LOD is on the trees. They aren’t supposed to look like cherry blossoms with big white blocks on the tops, and the larger ones with leaves don’t, as can clearly be seen in the 3rd photo.

I assure you I don’t mean to be disrespectful or anything, and am willing to do my due diligence in finding the issue, so i will take the steps you deserve and check back in. But seriously, am i like way overthinking this or expecting the LOD to be more of an accurate match then is possible? To me the LOD very clearly looks like it has issues with the upper branches being white blocks, but neither you or z929669 seem to see the issue despite how clear i’ve tried to be. Starting to feel a little crazy hahahahaha 

You are using a 3rd party tree mod with third party 3D LOD assets. I have no idea how anything is supposed to look. From experience I know that DynDOLOD typically does not change the look of 3rd party LOD assets in the way described.

https://dyndolod.info/Official-DynDOLOD-Support-Forum#In-Game-Screenshots
To get closer to LOD, toggle to the free fly camera with tfc in console.

A close up screenshot of LOD and a close up screenshot of the full model that shows the differences between them would be helpful.

Is there a reason you are not chekcing checking the NIF and textures the 3D LOD models uses in order to see what mod(s) they come form? If I knew there mod(s) and filenames I might be able to look at things myself.