DynDOLOD Settings Compares

[z929669]

I did some testing of the indicated DynDOLOD settings with respect to the Step SkyrimSE 2.3 build.

Testing Parameters

• Step SkyrimSE 2.3 build
• Step Heavy ENB
• BethINI Ultra preset
• DynDOLOD settings used to generate different outputs are identical except for the setting tested
• Testing limited to Whiterun (other child worldspaces may have more/less pronounced outcomes)

All comparison screenshots were created after loading a new game, so the position of the camera is slightly juxtaposed among like compares. ToD is slightly different but weathers are identical. In other words, compares were not captured via loading the same savegame, which is typically the best approach for screenshot compares. This is not the case when comparing certain aspects of different DynDOLOD outputs, since the plugins need to be reinitialized for each output in many cases.

1. "Upgrade NearGrid large references to FarGrid"
2. "Parent > child"
3. "Child > parent 'Lose' vs 'High'"

DynDOLOD settings tested

---

Upgrade NearGrid

According to the DynDOLOD documentation, "Upgrade NearGrid large references to FarGrid" should not be enabled if the large reference system is used, because it adds an unnecessary performance cost. SSE uses the LR system.

I found only a slight performance difference and no visual difference:

Upgrade UNticked-55 FPS Upgrade ticked-53 FPS )

I tested other areas with similar results with no performance difference at all in some cases. Note that I have a relatively high-end PC, so let's assume that small differences I see will be more significant on the average PC. Note that ToD in the second shot is probably closer to noon than the first shot, which is closer to 10 AM. This is why there are some shadow diffs.

Step recommends: Leave this option off.

---

Reference child/parent documentation for the following...

Parent > child

This setting indeed has a performance cost as a tradeoff for more accurate LOD seen outside of child worldspaces from the inside. Is it worth it? Probably not, but you be the judge.

No p-c, c-p=high p-c, c-p=high

Results show a 11 FPS hit for me, with the only difference being a few buildings and some near-ish rocks/mountain LOD visible from inside WR when enabling this option (see left-center of images). There was no performance difference outside the city walls (data not shown).

Step recommends: Leave this option off.

Child > parent Low vs High

This setting had no performance impact for me. Nevertheless, the visual difference in these shots is extremely subtle.

No p-c, c-p=high No p-c, c-p=low

The only difference I could make our was that the LOD for Gildergreen and possibly some rocks are visible inside WR when High is used.

Step recommends: Use the 'High' option, but the default 'Low' is probably just fine.