Model texture is rendered white

[BeverlyCode]

All versions.

Bug in entity rendering system. Causes only textures on entities to be full black, white, or use a random texture. Noticeable in Brachiosaurus from Paleotest in most instances, or mob entities in Livingfloatlands.

This is not limited to mobkit or mobs_redo, it can effect any entity. The bug is reproducible but the effects are random, doesn't always show up for everyone.

I have tested the same b3d models as meshnodes and the bug only appears when the models are loaded as entities.

Confirmed by GreenXenith: https://discord.com/channels/369122544273588224/369137254641303560/1089982985291452506

[SmallJoker]

Please do not delete the entire issue template next time and use it as a guide.

• How can this be reproduced? Preferably include a screenshot to visualize your problem.
• Which OS, architecture, GPU model, driver?

EDIT: The content you're referring to in the link is not visible by the public, hence suboptimal.

[Niklp09]

EDIT: The content you're referring to in the link is not visible by the public, hence suboptimal.

GreenXenith's message

Brach on jurassic park server, android (mt 5.5.1)

[BeverlyCode]

• Which OS, architecture, GPU model, driver?

This occurs on a variety of those combinations, I have tested this on many machines, GPU, iGPU, etc. Windows,Linux,Iris XE,AMD VEGA,Intel HD,RTX 2060,3090,various drivers over months,etc.

It's been a bug that has been in the engine for a long, long time.

[fluxionary]

i'm pretty sure this is a bug in paleotest, CalebJ from Tunnelers' Abyss fixed it at some point, but doesn't remember how. i installed paleotest in a fairly basic world, and spawned a few brachiosaurs, which were pure white. most mobs on most other servers are fine for me.

[fluxionary]

actually, something similar seems to have been reported previously: ElCeejo/paleotest#8

[BeverlyCode]

actually, something similar seems to have been reported previously: ElCeejo/paleotest#8

Yep that looks like exactly the same problem

i'm pretty sure this is a bug in paleotest, CalebJ from Tunnelers' Abyss fixed it at some point, but doesn't remember how. i installed paleotest in a fairly basic world, and spawned a few brachiosaurs, which were pure white. most mobs on most other servers are fine for me.

Like I mentioned in the first post it's not limited to Paleotest, or a specific mob library. Just that the Paleotest Brachiosaurus is the most reliable way to reproduce it. I have looked at the Paleotest code and see no reason why the Brachiosaurus would cause these rendering artefacts over other mobs in Paleotest.

I have had this happen to entities I have registered in custom mods with no dependencies.

I actually find it highly unlikely anyone has fixed this, one can easily delude themselves into thinking they have fixed it because it's not a consistent problem and what causes it to appear and disappear is unknown, so it seems random, but it is easy to change something and think that it has been fixed when really, it hasn't.

If this is not an engine bug, I would be really really quite shocked. I can assure you it is an engine bug, but I have already had this argument on the Discord for some hours with no solution or progress other than GreenXenith confirming its existence, which alone is a fight, because the default consensus of Minetest users on discord is to just not believe you; if it's never happened to them they assume it's "never happened to anyone but you" etc. (sometimes I wonder if they ever actually even play the game or just post on discord wasting everyone's time lol)

But if you want to go down the route of blaming Paleotest which I have confirmed in my own rigorous tests that it is not, feel free to go down that avenue - the more confirmation against it the better.

[ghost]

Petz has not that bug. I belive is a mod error.

[BeverlyCode]

Petz has not that bug. I belive is a mod error.

I believe you have not read this entire thread. It is not a mod error. It is very plausible there are mods that do not incur this problem, but that alone does not deduce it to being a "mod error", please refrain from making this claim without hard evidence.

[ghost]

I believe you have not read this entire thread. It is not a mod error. It is very plausible there are mods that do not incur this problem, but that alone does not deduce it to being a "mod error", please refrain from making this claim without hard evidence.

After many years of experience I will tell you, those errors can be due to functions that change textures, for example a bad use of object:set_properties. From the mod or from another mod that interferes (even in this case a mod that is not a dependency), it is very common with callbacks.

It is very easy to check, enter the model in PETZ. I am 100% sure there will be no problem.

[BeverlyCode]

I believe you have not read this entire thread. It is not a mod error. It is very plausible there are mods that do not incur this problem, but that alone does not deduce it to being a "mod error", please refrain from making this claim without hard evidence.

After many years of experience I will tell you, those errors can be due to functions that change textures, for example a bad use of object:set_properties. From the mod or from another mod that interferes (even in this case a mod that is not a dependency), it is very common with callbacks.

It is very easy to check, enter the model in PETZ. I am 100% sure there will be no problem.

I hear you, I checked use of set_properties in paleotest and it's only being used to set visual_size, any other thoughts? I don't doubt there is something triggering this bug and this may well being triggered by something specific such as a b3d mesh file when it is loaded.

If you read this thread you will findout that I have tested this with only one very minimal mod loaded that loads the b3d file as an entity and the bug still occurs which eliminates to possibility of it being a mod related bug.

I have had this happen to entities I have registered in custom mods with no dependencies.

It just can't be related to a mod. Maybe how the b3d file is loaded? Maybe, maybe.

[ghost]

send me the b3b

[BeverlyCode]

send me the b3b

models.zip

It's not actually limited to b3d there is also a .x file that is affected, but the bug only occurs when the models are loaded as entities (not meshnodes) and it's not 100% trigger-able, it's random, some times it will work, some times it will break. This is because we don't know exactly what causes it, but I have narrowed in it down to still affecting the models when no other mods are loaded (apart from minetest_game) and the models are loaded as simple entities.

When it does break it will appear as one of the three:

1. a random texture
2. all white texture
3. all black texture

[SmallJoker]

This looks optically the same as #9374 (comment) but it was fixed during the process.
I can confirm that /spawnentity paleotest:brachiosaurus from paleotest sometimes results in incorrectly rendered textures. This does not depend on shaders, e.g. enable_shaders = false makes no difference.

Have you yet checked whether the texture normals are correct? Aside from that it might be worthwhile to valgrind for uninitialized variables.

[BeverlyCode]

This looks optically the same as #9374 (comment) but it was fixed during the process. I can confirm that /spawnentity paleotest:brachiosaurus from paleotest sometimes results in incorrectly rendered textures. This does not depend on shaders, e.g. enable_shaders = false makes no difference.

Have you yet checked whether the texture normals are correct? Aside from that it might be worthwhile to valgrind for uninitialized variables.

Yeah texture normals was one of the first things I checked on all the affected models, including inverse scaling factors which would invert normals that seemingly point in the correct direction.

I would seem to be a memory-leak or uninitialized variables as you suggest.

It is also correct that it appears with or without shaders enabled, unrelated to shader code.

[sfan5]

Needs re-testing.