The root cause of the issue is that OpenGL ES 2 does not support the `textureCubeLod` function.
There are (optional) extensions to support this, but they don't appear to be exposed with the ES2 renderer (even though the hardware needed to support LOD features are certainly available.)
The existing shim in `drivers/gles2/shaders/cubemap_filter.glsl` just creates a macro:
```
#define textureCubeLod(img, coord, lod) textureCube(img, coord)
```
But the third parameter of `textureCube` is actually a mip bias, not an absolute mip level.
(And it doesn't seem to work regardless.)
In this specific case, the `cubemap_filter` should only sample from the first level of the "source" panorama cubemap.
In lieu of a method to force a lod level of zero, I've chosen to comment out the switchover from a 2D equirectangular panorama to the cubemap version of the same image, therefore always sampling roughness values from the 2D equirectangular panorama.
This may cause additional artifacts or issues across the seam, but at least it prevents the glaringly obvious black areas.
---
This same issue (no fragment texture LOD support) has rather large repercussions elsewhere too; it means materials with larger cubemap density (i.e. planar or distant objects) will be far rougher than expected.
Since GLES 3 appears to properly support fragment `texture*Lod` functions, switching to the GLES 3 backend would solve this problem.
---
Root cause discovered with help from @KaadmY.
Image::resize_to_po2() now takes an optional p_interpolation parameter
that it passes directly to resize() with default value INTERPOLATE_BILINEAR.
GLES2: call resize_to_po2() with interpolate argument
Call resize_to_po2() in GLES2 rasterizer storage with either
INTERPOLATE_BILINEAR or INTERPOLATE_NEAREST depending on TEXTURE_FLAG_FILTER.
This avoids filtering issues with non power of two pixel art textures.
See #44379
This partially reverts #44041.
Also fix the bogus label about lack of GLES3 support being always shown...
It was meant to be behind a check but I had left it out while testing, and
forgot to put it back.
Better materials:
- default values will no longer trigger things like emission,
clear coat being enabled etc incorrectly, mostly importers misunderstand
this value, a value of 1 is not actually enabled emission, it must have a
non zero setting other than emission value for the emission flags to
actually be enabled correctly in our engine #42386
- lambert materials are warned against significantly,
do not use Lambert in Godot, use a PBR material like Ai Standard Surface
(it's like going from low quality to high definition in terms of output
in scenes and on assets)
- roughness values are calculated correctly in this version
Fixes for normal's array - some normal's from some files were dropped and
generated, this is now fixed.
- FBX indexing for items with (-1) index, for normal data is supported,
this is super helpful at increasing our range of compatibility of
FBX meshes.
Fix bone rest data **no longer requiring go to bind pose in Maya and various applications**
- Partial fix for #43688
Validation tools added for validating large projects
Provide package name, vendor and FBX vendor in the log.
Implemented metadata properties so we can read extra document
info required for bug reporting
**FBX 2011 (version 7200)** is unsupported now,
you must re-export your file in Maya or Max to upgrade the file format.
Fixes skin bind poses being generated based on node xforms in the scene
Fixes duplicate bones being added to skin and prevents add_bones from
registering skeleton bones it now registers skin bind poses,
but this should really be documented in the engine correctly right now
it doesn't tell you this is the case.
fix for issue #44231: snap_selected_nodes_to_floor() results in an editor
crash when a child collisionshape has invalid shape object
(cherry picked from commit 52f6e0b5ad)
