[Solved]freezeAllActiveMeshes issues after upgrade to 4.1

Hi,
after upgrading to version 4.1 stable noticed issue with one of my materials.
Its PBRMetalic material, that is transparent and uses global reflection texture, its used by cloned meshes

after calling freezeAllActiveMeshes on the scene, it turns black ands stays so until meshes are unfreezen.
Was working with alpha versions.

Trying to find cause but maybe someone have idea in with direction to check? :slight_smile:

Thank you

1 Like

Are you using multi scene ? I wonder if it could come from the BRDF texture.

Would be amazing if you could share a tiny PG with a repro so that I can fix it ASAP. It should definitely not do that.

@sebavan will try to reproduce in PG, but for your answer, no multi scene is not used

Getting closer to isolate, its related to shadow generation.
both meshes that are affected belong to shadow generator, and when I freeze meshes,
I animate shadow light (SpotLight) intensity from 0 to 1, at that point they become black

Thanks for the heads up and I am pretty sure it is a bug :frowning: I ll fix it in priority as soon as we have the repro.

I tried to reproduce your problem but could not:

https://playground.babylonjs.com/#P9T8KE

After 1s, the meshes are frozen, and after 1.2s the spot intensity is raised from 0 to 1 in ~ 1.5s

1 Like

Can you try on 4.2 preview ???

After animation is done I set map.refreshRate = 0; that’s when they become black, will try to find a way to reproduce in PG

@Evgeni_Popov @sebavan

Managed to update your PG @Evgeni_Popov to reproduce

PG

tested in Chrome

1 Like

Yes could not reproduce on safari, only on Chrome (Version 80.0.3987.132 (64 bit))
OSX: Catalina 10.15.3

Thanks a lot for the repro !!! I am on it I will fix it ASAP

2 Likes

Just in case somebody else is looking into this, here is a min repro:
https://playground.babylonjs.com/#0NR8TY#5

Basically it looks like it only happens if we are first without refresh then freeze and then set back to no refresh. It is also just in PCF as the colorWrite flag looks like being incorrectly reseting the meshes who are thinking to be in a depth pre pass.

Obviously a bug but might take a bit of time to fix.

The fix should be in in a couple of hours :slight_smile: thanks again for your report.

2 Likes

Ok due to meetings and other stuff I can not avoid (parent teacher meeting due to the virus…) those two hours will probably transform in a bit more, but I will try to have it by end of day in the US, sorry for the delay.

3 Likes

On behalf of everyone in the Babylon community, NO APOLOGIES NECESSARY Seb!!!

You are a rockstar and we all love you!

5 Likes

All fixed pushing a new nightly in about 10 minutes, sorry again for the delay.

All deployed !!!

1 Like

Thank you for your quick work, will test out as soon as npm version will be created :slight_smile:
And no apologies needed as this is way quicker then most enterprise/paid engine support is doing :smiley: