Although 2D draws in painters order with strict ordering, in certain circumstances items can be reordered to increase batching / decrease state changes, without affecting the end result. This can be determined by an overlap test.
In situation with item:
A-B-A
providing the third item does not overlap the second, they can be reordered:
A-A-B
Items already contain an AABB which can be used for this overlap test.
1)
To utilise this, I have implemented item reordering (only for single rects for now), with the lookahead adjustable in project settings. This can increase performance in situations where items may not be grouped in the scene tree by texture. It can also be switched off (by setting lookahead to 0).
2)
This same trick can be used to help join items that are lit. Lit items previously would prevent joining completely, thus missing out on performance gains other than multi-command items such as tilemaps.
In this PR, lights are assigned as bits in a bitfield (up to 64, the optimization is disabled above this), and on each try_item (for joining), the bitfield for lights and shadows is constructed and compared with the previous items. If these match the 2 items can potentially be joined. However, this can only be done without changing the rendered result if an overlap test is successful.
This overlap test can be adjusted to join items up to a specific number of item references, selectable in project settings, or turned off.
3)
The legacy uniform single rect drawing routine seems to have been identified as the source of flicker, particularly on nvidia. However, it can also be up to 2x as fast. Because of the speed the batching contains a fallback where it can use the legacy single rect method, but I have now added a project setting to make this switchable. In most cases with batching it should not be necessary (as single rects are drawn less frequently) and thus the flickering can be totally avoided.
4)
This PR also fixes a color modulate bug when drawing light passes, in certain situations (particularly custom _draw routines with multiple rects).
5)
This PR also fixes#38291, a bug in the legacy renderer where light passes could draw rects in wrong position.
- Resurrect it for GL ES 2
- Apply roll over with `fmod()` instead of resetting it to 0
- Expose the setting from the `VisualServer`, since it does not belong in any specific rasterizer
When reading SCREEN_TEXTURE in a shader, this previously only worked succesfully for the first read of the screen, because state.canvas_texscreen_used was never getting reset. This PR resets state.canvas_texscreen_used at the beginning of each joined item, so that further screen reads can happen.
Joining items across z_indices can interfere with light culling for lights which only affect certain z ranges. This PR disables joining across z_indices when lights are present, except specifically for lights with both z_min set to the global minimum (-4096) and z_max set to the global maximum (4096).
In addition, the z_index is now stored on the joined_item for accurate light culling. The z_index is also displayed in frame diagnostics.
In rare circumstances default batches were being joined incorrectly, causing visual regressions. This logic has been fixed.
In addition slightly more output information has been added to frame diagnosis mode.
Batching across z_index layers was not preserving the batch_break flag, which determines whether to not join the previous item. This is fixed by storing the flag in RenderItemState and preserving it across canvas_render_items calls.
Added project setting to enable / disable print frame diagnostics every 10 seconds. This prints out a list of batches and info, which is useful to optimize games and identify performance problems.
This adds 2 new values (items and draw calls) to the performance monitor in a '2d' section, rather than reusing the 3d values in the 'raster' section.
This makes it far easier to optimize games to minimize drawcalls.
Defers sending 'transform' commands within a RasterizerCanvas::Item until they are needed for default batches. Instead locally caches the extra matrix and applies it using software transform, preventing unnecessary batch breaks.
The logic is relatively complex, and the whole 'extra matrix' of the legacy renderer in addition to the final_transform is not ideal. However this is required to accelerate some user drawing techniques, and later the lines in the IDE.
Extra functions canvas_render_items_begin and canvas_render_items_end are added to RasterizerCanvas, with noop stubs for non-GLES2 renderers. This enables batching to be spready over multiple z_indices, and multiple calls to canvas_render_items.
It does this by only performing item joining within canvas_render_items, and deferring rendering until canvas_render_items_end().
Determined that a large reason for the decrease in performance in unbatchable scenes was due to the new routine being analogous to the 'nvidia workaround' code, that is about half the speed. So this simply uses the old routine in the case of single unbatchable rects. Hopefully we will be able to remove the old path at a later stage.
Where the final_modulate color varies between render_items this can prevent batching. This PR solves this by baking final_modulate into the vertex colors, and setting the uniform 'final_modulate' to white, and allowing the joining of items that have different final_modulate values. The previous batching system can then cope with vertex color changes as normal.
2d rendering is currently bottlenecked by drawing primitives one at a time, limiting OpenGL efficiency. This PR batches primitives and renders in fewer drawcalls, resulting in significant performance improvements. This also speeds up text rendering.
This PR batches across canvas items as well as within items.
The code dynamically chooses between a vertex format with and without color, depending on the input data for a frame, in order to optimize throughput and maximize batch size. It also adds an option to use glScissor to reduce fillrate in light passes.
Fixes#26637.
Fixes#19900.
The viewport_size returned by get_viewport_size was previously incorrect, being half the correct value. The function is renamed to get_viewport_half_extents, and now returns a Vector2.
Code which called this function has also been modified accordingly.
This PR also fixes shadow culling when using ortho cameras, because the correct input for CameraMatrix::set_orthogonal should be the full HEIGHT from get_viewport_half_extents, and not half the width.
It also fixes state.ubo_data.viewport_size in rasterizer_scene_gles3.cpp to be the width and the height of the viewport in pixels as stated in the documentation, rather than the current value which is half the viewport extents in worldspace, presumed to be a bug.
Some cases were not handled properly for Polygon2D after making changes in common code to fix Line2D antialiasing. Added an option for drawing polygons to differentiate the two use cases.
Fixes#34568
Happy new year to the wonderful Godot community!
We're starting a new decade with a well-established, non-profit, free
and open source game engine, and tons of further improvements in the
pipeline from hundreds of contributors.
Godot will keep getting better, and we're looking forward to all the
games that the community will keep developing and releasing with it.
Polygon2D:
The property wasn't used anymore after switching from canvas_item_add_polygon() to canvas_item_add_triangle_array() for drawing.
Line2D:
Added the same property as for Polygon2D & fixed smooth line drawing to use indices correctly.
Fixes#26823
This changes the code path so that `glRenderBufferStorage*` always uses
values appropriate for renderbuffers and `glTexImage2D` never uses an
internalformat meant for buffers.
Fixes#33825.
As discussed in #32657, this can't be done here as lines can be used
with a canvas scale, and this breaks them.
A suggestion is to do the pixel shifting at matrix level instead.
Fixes#33393.
Fixes#33421.
When rendering to an external texture and MSAA was active (as happened
in the Oculus Mobile ARVR plugin) no MSAA was rendered as the correct
depth buffer and multisample texture target was not used.
This also fixes https://github.com/GodotVR/godot_oculus_mobile/issues/54
While OpenGL ES 3.0 and WebGL 2.0 both support non power-of-2 (NPOT)
textures in their specification, the situation seems to be less clear
about *compressed* NPOT textures using repeat or mipmap flags.
At least Chrome on Linux doesn't seem to support this combination,
and a variety of mobile hardware have similar limitations.
As a workaround, we force decompressing such textures when running on
WebGL 2.0, at the cost of loading time and memory usage.
Fixes#33058.
OpenGL uses the diamond exit rule to rasterize lines. If we don't shift
the points down and to the right by 0.5, the line can sometimes miss a
pixel when it shouldn't. The final fragment of a line isn't drawn. By
drawing the lines clockwise, we can avoid a missing pixel in the rectangle.
See section 3.4.1 in the OpenGL 1.5 specification.
Fixes#32279
This fixes an issue that was fixed for gles3 in #31419 but not applied
to gles2. The fix consists of using a constant scale for cube_normal of -1.0
instead of -1000000. It results in broken panorama rendering on the
oculus quest (see https://github.com/GodotVR/godot_oculus_mobile/issues/29)
Although the backup USE_SKELETON_SOFTWARE skinning path is currently used when float texture is not supported, the default skinning path still fails when float texture is supported but GL_MAX_VERTEX_TEXTURE_IMAGE_UNITS is 0, i.e. the device cannot read from texture during vertex shader. This PR adds the logic to activate the SKELETON_SOFTWARE path if either of these cases occur, preventing crashes on devices which have this combination of features.
In 2.1 and 3.0, light_vec could be modified for altering shadow_computations.
But it broke shadows when rotating light. shadow_vec would do the same, but without breaking
shadows in rotated lights if not used.
Add inverse light transformation to shadow vec, so it's not affected when rotating lights;
Added usage define for shadow vec.
For shadow vec working properly when rotating a light, it's needed to multiply it by light_matrix normalized. Added usage define in order to don't do that if shadow_vec not used.
Condensed some if and ERR statements. Added dots to end of error messages
Couldn't figure out EXPLAINC. These files gave me trouble: core/error_macros.h, core/io/file_access_buffered_fa.h (where is it?),
core/os/memory.cpp,
drivers/png/png_driver_common.cpp,
drivers/xaudio2/audio_driver_xaudio2.cpp (where is it?)
This makes height fog appear at the bottom of the scene
(instead of the top), which is generally the expected result.
This also tweaks the fog height setting hint to be more flexible.
This closes#30709.
This is a straight copy-paste of the code from
`drivers/gles3/rasterizer_canvas_gles3.cpp`. It is subject to the
same restrictions as the GLES3 implementation: it only works
on desktop platforms as they use OpenGL instead of OpenGL ES.
For clarity, assign-to-release idiom for PoolVector::Read/Write
replaced with a function call.
Existing uses replaced (or removed if already handled by scope)
When setting the default precision type for uniforms (before compiling
the shader) prevent boolean uniforms from having one set. Booleans can't
have a precision type and on some Android devices this caused a
compilation failure.
Fixes#30317
Non-tools OpenGLES2 devices that use the USE_SKELETON_SOFTWARE path (i.e. do not support float texture) depend on surface->data being set containing the bone IDs and weights (rasterizer_scene_gles2.cpp, line 1456, RasterizerSceneGLES2::_setup_geometry). However currently if TOOLS_ENABLED is not defined, surface->data is not stored in main memory in rasterizer_storage_gles2.cpp. This causes a crash in rasterizer_scene_gles2.cpp when a rigged object comes into view.
This fix addresses the specific case of skinned objects when USE_SKELETON_SOFTWARE is active, and stores a copy of the bone data, as is done when TOOLS_ENABLED is defined. This fixes the crash by allowing the same mechanism as on desktop, without adding the memory overhead of storing all vertex data where not required.
Fixes#28298
This is a new singleton where camera sources such as webcams or cameras on a mobile phone can register themselves with the Server.
Other parts of Godot can interact with this to obtain images from the camera as textures.
This work includes additions to the Visual Server to use this functionality to present the camera image in the background. This is specifically targetted at AR applications.
The use of different default precision values (highp in vertex; mediump
in fragment) for uniform variables caused the shader program to not link properly on some android
devices/emulators.
If a non-imported texture resource file (e.g. DDS) gets updated the editor
doesn't reload it. The cause of the problem is two-fold:
First, the code of ImageTexture assumes that textures are always imported
from an image, but that's not the case for e.g. DDS. This change thus adds
code to issue a resource reload in case an image reload is not possible
(which is the case for non-imported texture resources).
Second, the code is filled with bogus calls to Image::get_image_data_size()
to determine the mipmap offset when that should be done using
Image::get_image_mipmap_offset(). Previous code literally passed the integer
mip level value to Image::get_image_data_size() where that actually expects
a boolean. Thus this part of the change might actually solve some other
issues as well.
To be pedantic, the texture_get_data() funciton of the rasterizer drivers is
still quite a mess, as it only ever returns the whole mipchain when
GLES_OVER_GL is set (practically only on desktop builds) but this change does
not attempt to resolve that.
The bake mode property of lights previously didn't affect GI probes.
This change makes the GI probe ignore lights that have their bake mode
set to disabled.
It seems to stay compatible with formatting done by clang-format 6.0 and 7.0,
so contributors can keep using those versions for now (they will not undo those
changes).
In canvas.glsl and scene.glsl, we were using texel2DFetch from stdlib.glsl,
which uses texture2DLod. In both cases, the stdlib.glsl include came before
the define of texture2DLod.
Might fix issues for drivers that don't support GL_EXT_shader_texture_lod.
