Cacti in flower pots don't render fully
NoComment1105 opened this issue ยท 15 comments
Expected Behavior
Le cacti should be fully visible
Actual Behavior
Le cacti is business at the front, party out back, The front to faces only render the back two don't
Reproduction Steps
- Place a catci in a flower pot
- Observe missing rear
Attachments
System Information
I am using the pinned version in #sodium on the discord (sodium-fabric-mc1.16.4-0.1.1-SNAPSHOT.jar)
- Java Version: [1.16.4 (java 14.0.2 64-bit]
- CPU: [Intel i5-5300U @2.3Ghz]
- GPU: [Intel HD Graphics 5500]
Cannot reproduce this with the .jar I believe you are referring to (linked here, Sha1: 5cad7f323991d139d322a8a85b043697e7f78857
).
System Information
- Java Version: 1.8.0_51
- CPU: Intel i5-5500U
- GPU: Intel HD Graphics 5500
Sodium has had several commits to the 1.16.x/next
branch since this issue was opened. Can you test if you experience this issue with Sodium (and only Sodium) with the latest commit?
Try using programmer-art
resource pack.
Not sure this will fix the bug, but even if it doesn't fix that, at least you'll have true minecraft, not the horrible game ruined by Microsoft.
Not sure this will fix the bug, but even if it doesn't fix that, at least you'll have true minecraft, not the horrible game ruined by Microsoft.
That's subjective, you cannot say that people's opinions (on minecraft textures) are incorrect.
Try using
programmer-art
resource pack.
Not sure this will fix the bug, but even if it doesn't fix that, at least you'll have true minecraft, not the horrible game ruined by Microsoft.
Ah right I will try it soon, but I'm not to bigger fan of programmer art, I feel the new textures look better, however I will still try
Oh my god I hadn't that that may influence it, I guess it happens because sodium thinks there is a full block on top of it therefore there should be no need to render the top and back because if thinks there is a full block above it. Sodium being resource saving I guess
Yes, I have a quick fix but I'm not sure if it's problematic/performance-degrading. The problem is in BlockOcclusionCache.calculate
. The specific line is 65: boolean ret = VoxelShapes.matchesAnywhere(selfShape, adjShape, BooleanBiFunction.ONLY_FIRST);
. Changing the predicate function to BooleanBiFunction.FIRST
appears to fix it.
Thats very interesting. I will have a look and see if it works for me too. Could you please test with Oneshot 3.0 chunk renderer as my iGPU is blacklisted against 4.3
@DasBundesland Thank you so much for reproducing this bug! I cannot articulate how happy I am to see this figured out, I spent MONTHS trying to reproduce without success
@amnotbananaama lol you're welcome!
Ok, so I am unsure what exactly what the different BiBooleanFunctions do in the context of VoxelShapes#matchEverywhere
and after some testing, there are some options that include the case where both culling faces are empty (FIRST and OR worked without breaking everything). Hopefully someone smarter than I am can provide insight.
Looks like a Vanilla bug :P MC-195461, it has been fixed on 20w46a apparently.