Some HISM are moved at world origin when fastgeo is active

When fastgeo is activated by adding “FastGeoWorldPartitionRuntimeCellTransformer” in the Runtime Cells Transformer Stack of the world partition, some HISM or ISM objects are loaded at the world center instead of at their expected positions.

When viewed in editor, the tile is OK

Once exported with no fastgeo, then streamed from pak, the trees are still well placed :

But when exported with fastGeo some trees are missing :

and we can found them at the center of the Unreal world :

We have not found why this strange behavior occurs only on some actors and not on others.

On the pictures we can see that inside a same actor, some IHSM are still well placed (water tower) and others not (trees). And some trees on other actors are always OK with or without fastgeo activated.

Is it a known bug ?

thanks for your help.

Steps to Reproduce
We have a process to export in .pak files some terrain tiles composed of several actors for terrain, buildings, trees … each tile is exported in a level.

Some actors manage ISM and/or HISM.

At runtime levels from .pak files are streamed and some elements are managed by the World Partition.

Nanite is not active for our project.

We recently tried to activate the fastgeo plugin on exported levels, which led to the problems described in this post.

Hello!

We addressed a few parenting problem with the runtime cell transformers. I think that CL44081791 contains the fix for the problem described here.

If you are using the ISM runtime cell transformer, I recommend grabbing CL48817083 and CL49332332.

Regards,

Martin

Hi Martin,

Thanks for the answer. What is the best way to review changes from a CL ?

Also, what Unreal Engine version did receive the corrections ? Is this already in 5.7 branch or is it pending 5.8 future release ?

Best,

Basile

Hi,

I missed that you indicated using the Launcher. In this case, the changes can be see on GitHub if you have tied your EpicID with a GH account. https://www.unrealengine.com/en\-US/ue\-on\-github

Martin

Hi Martin,

Can you confirm the first link ?

On my side, I get the attached result which does not make much sense.

Also, for completeness, assuming we cannot upgrade to 5.7 in the short term and that the commit you mentioned does result the problem, can you precise whether the correction is required at cook time, at runtime or both ?

If the result is only at cook time, then we may restrain it to the database production, …

Thanks,

Basile

Populate Release-5.7 from Dev-Release-5.7 at CL45349008 · EpicGames_UnrealEngine@679703e.pdf(139 KB)

Hi,

I converted the CL using an internal tool and didn’t validate the GH link which points at the creation of the 5.7 branch. The commit on the Main stream is: https://github.com/EpicGames/UnrealEngine/commit/f11b2502b2e15f37333a80b55d0227756c0b4df1

Those problems do happen in the editor. The RCTs are executed during the cell generation phase. This phase will happen when running PIE and while cooking.

Regards,

Martin

Hi Martin,

I hope you are fine despite the recent announcements…

I took the time to cherry pick the mentioned revision in my local build and recook the test cell. Unfortunately, it is still locating the trees at the wrong place. Is there any other candidate you can think of for trying ?

Else, I guess I will have to test directly within UE5.7.4 to see if it was fix or still affecting the product.

Thanks,

Basile

HI,

We have this other case that had similar problems: [Content removed]

This thread contains a proposed code fix and is related to BPs that are parents to other actors at the Outliner level.

Martin