WebGLRenderer: Create WebGL2 context by default

[mrdoob]

Fixes #19040

[fernandojsg]

weee 👍

[mrdoob]

Hmm... I'm seeing regressions...

https://raw.githack.com/mrdoob/three.js/dev/examples/webgl_loader_gltf.html

Before:

After:

[mrdoob]

@elalish Seems to be the RoughnessMipmapper 🤔

You can test this on model-viewer by passing WebGL2 context to the renderer:

var canvas = document.createElement( 'canvas' );
var context = canvas.getContext( 'webgl2', { alpha: false } );
renderer = new THREE.WebGLRenderer( { canvas: canvas, context: context } );

[mrdoob]

Another regression...

https://raw.githack.com/mrdoob/three.js/dev/examples/webgl_shaders_tonemapping.html

Before:

After:

[Mugen87]

So if the user needs a WebGL 1 context, it's necessary to create it on application level and pass it into the renderer, right?

[Mugen87]

webgl_mirror_nodes is broken with shader compilation errors.

[Mugen87]

Should be fixed via #19625.

[mrdoob]

So if the user needs a WebGL 1 context, it's necessary to create it on application level and pass it into the renderer, right?

What would a use case be to not want WebGL 2?

[Mugen87]

For example a project does not have the budget to update its shader code (see #19627) but still needs an important bug fix in a newer three.js version.

[Mugen87]

Seems to be the RoughnessMipmapper 🤔

webgl_loader_gltf reports a WebGL warning now:

[.WebGL-0x7fd1ef043800]RENDER WARNING: texture bound to texture unit 0 is not renderable. It might be non-power-of-2 or have incompatible texture filtering (maybe)?

[mrdoob]

For example a project does not have the budget to update its shader code (see #19627) but still needs an important bug fix in a newer three.js version.

I think that's fair.

Here are 2 possible ways:

new THREE.WebGLRenderer( { forceWebGL1: true } );

Or make a new class that extends WebGLRenderer.

new THREE.WebGL1Renderer();

And the internal code would just do if ( scope.isWebGL1Renderer === true )...

I think I prefer the later?

[Mugen87]

I'm fine with both. Pick the one you like the best 😇 .

In any event, it's good if we provide users an easy fallback just for the case they struggle with the migration.