Parallax2D (and ParallaxLayer) applies repeat to children nodes before their rotation

[squirrel-bat]

Tested versions

v4.3.stable.official [77dcf97]

System information

Godot v4.3.stable - Pop!_OS 22.04 LTS - Wayland - GLES3 (Compatibility) - AMD Radeon Graphics (radeonsi, raphael_mendocino, LLVM 15.0.7, DRM 3.57, 6.9.3-76060903-generic) - AMD Ryzen 7 7800X3D 8-Core Processor (16 Threads)

Issue description

With the introduction of Parallax2D nodes, repeated children nodes will be placed in respect to their own rotation.
In the screenshot below, the sprite seen on the right is the repeat of the Sprite2D on the left side, with it's rotation set to 12° and Parallax2D's rotation left at 0°.

The same issue occurs with a setup of ParallaxBackground > ParallaxLayer > Sprite2D.

I am not sure if this is the intended behaviour? Please let me know if there is missing information or if I am misunderstanding how Parallax2D works! :)

Steps to reproduce

• Create a Parallax2D node and set any Repeat Size with a minimum Repeat Times of 1
• Add a child node (Sprite2D) and set it's rotation to anything other than 0°

Minimal reproduction project (MRP)

parallax_demo_MRP.zip

[markdibarry]

You know, I have no idea. I assumed it should be the same between the two versions, but I'm not sure what is desired. If it's more desired and doesn't have any bad side effects, I'd be happy to look into changing how it works.

Edit:
After our convo, I think you're right. We should explore this, but the behavior should be that the repeat should respect the item's rotation, just not when a child of a parallax node. I'll see if I can come up with something soon! Thanks for the great catch!

[kleonc]

As pointed out in #95666 (comment) ParallaxLayer/ParallaxBackground behave the same way in v4.3.stable.official [77dcf97] which is a regression compared to v4.2.stable.official [46dc277]. Parallax2D was just added so not a regression but both me and @markdibarry consider this a bug. Hence marking as such.

v4.2.stable.official [46dc277]	v4.3.stable.official [77dcf97]

Edit: Just noticed this issue was mentioning ParallaxBackground/ParallaxLayer from the start, but I somehow got focused only on the Parallax2D part, my bad!

[Macksaur]

I'm on 83d54ab and this seems to be still broken, I can't get the MRP from #95666 to produce the new behaviour. Actually, wait, it does work but only on Compatibility and does not work on Forward+.

[kleonc]

I'm on 83d54ab and this seems to be still broken, I can't get the MRP from #95666 to produce the new behaviour. Actually, wait, it does work but only on Compatibility and does not work on Forward+.

Seems like #92797 was not properly rebased to account for the changes in servers/rendering/renderer_rd/renderer_canvas_render_rd.cpp done in 1bd8372 / #95666. cc @stuartcarnie

Basically this part from the current master is pre-#95666 because of this:

godot/servers/rendering/renderer_rd/renderer_canvas_render_rd.cpp

Lines 2149 to 2166 in 74de05a

	Transform2D base_transform = p_canvas_transform_inverse * ci->final_transform;
	if (!ci->repeat_size.x && !ci->repeat_size.y) {
	_record_item_commands(ci, p_to_render_target, base_transform, current_clip, p_lights, instance_index, batch_broken, r_sdf_used);
	} else {
	Point2 start_pos = ci->repeat_size * -(ci->repeat_times / 2);
	Point2 end_pos = ci->repeat_size * ci->repeat_times + ci->repeat_size + start_pos;
	Point2 pos = start_pos;
	do {
	do {
	Transform2D transform = base_transform * Transform2D(0, pos / ci->xform_curr.get_scale());
	_record_item_commands(ci, p_to_render_target, transform, current_clip, p_lights, instance_index, batch_broken, r_sdf_used);
	pos.y += ci->repeat_size.y;
	} while (pos.y < end_pos.y);
	pos.x += ci->repeat_size.x;
	pos.y = start_pos.y;
	} while (pos.x < end_pos.x);
	}

So that's a new regression and, if anything, should be rather reported as a new issue.