Contains global variables accessible from everywhere. Use [method get_setting], [method set_setting] or [method has_setting] to access them. Variables stored in [code]project.godot[/code] are also loaded into ProjectSettings, making this object very useful for reading custom game configuration options.
When naming a Project Settings property, use the full path to the setting including the category. For example, [code]"application/config/name"[/code] for the project name. Category and property names can be viewed in the Project Settings dialog.
[b]Feature tags:[/b] Project settings can be overridden for specific platforms and configurations (debug, release, ...) using [url=https://docs.godotengine.org/en/3.4/tutorials/export/feature_tags.html]feature tags[/url].
[b]Overriding:[/b] Any project setting can be overridden by creating a file named [code]override.cfg[/code] in the project's root directory. This can also be used in exported projects by placing this file in the same directory as the project binary. Overriding will still take the base project settings' [url=https://docs.godotengine.org/en/3.4/tutorials/export/feature_tags.html]feature tags[/url] in account. Therefore, make sure to [i]also[/i] override the setting with the desired feature tags if you want them to override base project settings on all platforms and configurations.
Returns the absolute, native OS path corresponding to the localized [code]path[/code] (starting with [code]res://[/code] or [code]user://[/code]). The returned path will vary depending on the operating system and user preferences. See [url=https://docs.godotengine.org/en/3.4/tutorials/io/data_paths.html]File paths in Godot projects[/url] to see what those paths convert to. See also [method localize_path].
[b]Note:[/b] [method globalize_path] with [code]res://[/code] will not work in an exported project. Instead, prepend the executable's base directory to the path when running from an exported project:
[codeblock]
var path = ""
if OS.has_feature("editor"):
# Running from an editor binary.
# `path` will contain the absolute path to `hello.txt` located in the project root.
Loads the contents of the .pck or .zip file specified by [code]pack[/code] into the resource filesystem ([code]res://[/code]). Returns [code]true[/code] on success.
[b]Note:[/b] If a file from [code]pack[/code] shares the same path as a file already in the resource filesystem, any attempts to load that file will use the file from [code]pack[/code] unless [code]replace_files[/code] is set to [code]false[/code].
[b]Note:[/b] The optional [code]offset[/code] parameter can be used to specify the offset in bytes to the start of the resource pack. This is only supported for .pck files.
Returns the localized path (starting with [code]res://[/code]) corresponding to the absolute, native OS [code]path[/code]. See also [method globalize_path].
[b]Note:[/b] This method is intended to be used by editor plugins, as modified [ProjectSettings] can't be loaded back in the running app. If you want to change project settings in exported projects, use [method save_custom] to save [code]override.cfg[/code] file.
Saves the configuration to a custom file. The file extension must be [code].godot[/code] (to save in text-based [ConfigFile] format) or [code].binary[/code] (to save in binary format). You can also save [code]override.cfg[/code] file, which is also text, but can be used in exported projects unlike other formats.
Comma-separated list of custom Android modules (which must have been built in the Android export templates) using their Java package path, e.g. [code]"org/godotengine/godot/MyCustomSingleton,com/example/foo/FrenchFriesFactory"[/code].
[b]Note:[/b] Since Godot 3.2.2, the [code]org/godotengine/godot/GodotPaymentV3[/code] module was deprecated and replaced by the [code]GodotPayment[/code] plugin which should be enabled in the Android export preset under [code]Plugins[/code] section. The singleton to access in code was also renamed to [code]GodotPayment[/code].
If [code]true[/code], scale the boot splash image to the full window length when engine starts. If [code]false[/code], the engine will leave it at the default pixel size.
If [code]true[/code], applies linear filtering when scaling the image (recommended for high resolution artwork). If [code]false[/code], uses nearest-neighbor interpolation (recommended for pixel art).
This user directory is used for storing persistent data ([code]user://[/code] filesystem). If left empty, [code]user://[/code] resolves to a project-specific folder in Godot's own configuration folder (see [method OS.get_user_data_dir]). If a custom directory name is defined, this name will be used instead and appended to the system-specific user data directory (same parent folder as the Godot configuration folder documented in [method OS.get_user_data_dir]).
Icon set in [code].icns[/code] format used on macOS to set the game's icon. This is done automatically on start by calling [method OS.set_native_icon].
The project's name. It is used both by the Project Manager and by exporters. The project name can be translated by translating its value in localization files. The window title will be set to match the project name automatically on startup.
[b]Note:[/b] Changing this value will also change the user data folder's path if [member application/config/use_custom_user_dir] is [code]false[/code]. After renaming the project, you will no longer be able to access existing data in [code]user://[/code] unless you rename the old folder to match the new project name. See [url=https://docs.godotengine.org/en/3.4/tutorials/io/data_paths.html]Data paths[/url] in the documentation for more information.
Specifies a file to override project settings. For example: [code]user://custom_settings.cfg[/code]. See "Overriding" in the [ProjectSettings] class description at the top for more information.
[b]Note:[/b] Regardless of this setting's value, [code]res://override.cfg[/code] will still be read to override the project settings.
If [code]true[/code], the project will save user data to its own user directory (see [member application/config/custom_user_dir_name]). This setting is only effective on desktop platforms. A name must be set in the [member application/config/custom_user_dir_name] setting for this to take effect. If [code]false[/code], the project will save user data to [code](OS user data directory)/Godot/app_userdata/(project name)[/code].
If [code]true[/code], the project will use a hidden directory ([code].import[/code]) for storing project-specific data (metadata, shader cache, etc.).
If [code]false[/code], a non-hidden directory ([code]import[/code]) will be used instead.
[b]Note:[/b] Restart the application after changing this setting.
[b]Note:[/b] Changing this value can help on platforms or with third-party tools where hidden directory patterns are disallowed. Only modify this setting if you know that your environment requires it, as changing the default can impact compatibility with some external tools or plugins which expect the default [code].import[/code] folder.
Icon set in [code].ico[/code] format used on Windows to set the game's icon. This is done automatically on start by calling [method OS.set_native_icon].
Time samples for frame deltas are subject to random variation introduced by the platform, even when frames are displayed at regular intervals thanks to V-Sync. This can lead to jitter. Delta smoothing can often give a better result by filtering the input deltas to correct for minor fluctuations from the refresh rate.
[b]Note:[/b] Delta smoothing is only attempted when [member display/window/vsync/use_vsync] is switched on, as it does not work well without V-Sync.
It may take several seconds at a stable frame rate before the smoothing is initially activated. It will only be active on machines where performance is adequate to render frames at the refresh rate.
[b]Experimental.[/b] Shifts the measurement of delta time for each frame to just after the drawing has taken place. This may lead to more consistent deltas and a reduction in frame stutters.
If [code]true[/code], disables printing to standard error. If [code]true[/code], this also hides error and warning messages printed by [method @GDScript.push_error] and [method @GDScript.push_warning]. See also [member application/run/disable_stdout].
Changes to this setting will only be applied upon restarting the application.
If [code]true[/code], disables printing to standard output. This is equivalent to starting the editor or project with the [code]--quiet[/code] command line argument. See also [member application/run/disable_stderr].
Changes to this setting will only be applied upon restarting the application.
If [code]true[/code], flushes the standard output stream every time a line is printed. This affects both terminal logging and file logging.
When running a project, this setting must be enabled if you want logs to be collected by service managers such as systemd/journalctl. This setting is disabled by default on release builds, since flushing on every printed line will negatively affect performance if lots of lines are printed in a rapid succession. Also, if this setting is enabled, logged files will still be written successfully if the application crashes or is otherwise killed by the user (without being closed "normally").
[b]Note:[/b] Regardless of this setting, the standard error stream ([code]stderr[/code]) is always flushed when a line is printed to it.
If [code]true[/code], enables low-processor usage mode. This setting only works on desktop platforms. The screen is not redrawn if nothing changes visually. This is meant for writing applications and editors, but is pretty useless (and can hurt performance) in most games.
Audio buses will disable automatically when sound goes below a given dB threshold for a given time. This saves CPU as effects assigned to that bus will no longer do any processing.
Audio buses will disable automatically when sound goes below a given dB threshold for a given time. This saves CPU as effects assigned to that bus will no longer do any processing.
Specifies the audio driver to use. This setting is platform-dependent as each platform supports different audio drivers. If left empty, the default audio driver will be used.
Safer override for [member audio/mix_rate] in the Web platform. Here [code]0[/code] means "let the browser choose" (since some browsers do not like forcing the mix rate).
Specifies the preferred output latency in milliseconds for audio. Lower values will result in lower audio latency at the cost of increased CPU usage. Low values may result in audible cracking on slower hardware.
Audio output latency may be constrained by the host operating system and audio hardware drivers. If the host can not provide the specified audio output latency then Godot will attempt to use the nearest latency allowed by the host. As such you should always use [method AudioServer.get_output_latency] to determine the actual audio output latency.
The default compression level for gzip. Affects compressed scenes and resources. Higher levels result in smaller files at the cost of compression speed. Decompression speed is mostly unaffected by the compression level. [code]-1[/code] uses the default gzip compression level, which is identical to [code]6[/code] but could change in the future due to underlying zlib updates.
The default compression level for Zlib. Affects compressed scenes and resources. Higher levels result in smaller files at the cost of compression speed. Decompression speed is mostly unaffected by the compression level. [code]-1[/code] uses the default gzip compression level, which is identical to [code]6[/code] but could change in the future due to underlying zlib updates.
The default compression level for Zstandard. Affects compressed scenes and resources. Higher levels result in smaller files at the cost of compression speed. Decompression speed is mostly unaffected by the compression level.
Largest size limit (in power of 2) allowed when compressing using long-distance matching with Zstandard. Higher values can result in better compression, but will require more memory when compressing and decompressing.
If [code]true[/code], displays getters and setters in autocompletion results in the script editor. This setting is meant to be used when porting old projects (Godot 2), as using member variables is the preferred style from Godot 3 onwards.
If [code]true[/code], enables specific GDScript warnings (see [code]debug/gdscript/warnings/*[/code] settings). If [code]false[/code], disables all GDScript warnings.
If [code]true[/code], enables warnings when a function is declared with the same name as a variable. This will turn into an error in a future version when first-class functions become supported in GDScript.
If [code]true[/code], enables warnings when passing a floating-point value to a function that expects an integer (it will be converted and lose precision).
If [code]true[/code], enables warnings when calling a function without using its return value (by assigning it to a variable or using it as a function argument). Such return values are sometimes used to denote possible errors using the [enum Error] enum.
If [code]true[/code], enables warnings when defining a local or subclass member variable that would shadow a variable at an upper level (such as a member variable).
If [code]true[/code], enables warnings when calling an expression that has no effect on the surrounding code, such as writing [code]2 + 2[/code] as a statement.
If [code]true[/code], enables warnings when calling a ternary expression that has no effect on the surrounding code, such as writing [code]42 if active else 0[/code] as a statement.
If [code]true[/code], enables warnings when assigning a variable using an assignment operator like [code]+=[/code] if the variable wasn't previously assigned.
If [code]true[/code], enables warnings when a variable is declared with the same name as a function. This will turn into an error in a future version when first-class functions become supported in GDScript.
<membername="debug/settings/crash_handler/message"type="String"setter=""getter=""default=""Please include this when reporting the bug on https://github.com/godotengine/godot/issues"">
Maximum number of frames per second allowed. The actual number of frames per second may still be below this value if the game is lagging. See also [member physics/common/physics_fps].
This setting is therefore mostly relevant for lowering the maximum FPS below VSync, e.g. to perform non-real-time rendering of static frames, or test the project under lag conditions.
Print more information to standard output when running. It displays information such as memory leaks, which scenes and resources are being loaded, etc.
If [code]true[/code], allows HiDPI display on Windows, macOS, and the HTML5 platform. This setting has no effect on desktop Linux, as DPI-awareness fallbacks are not supported there.
The default screen orientation to use on mobile devices.
[b]Note:[/b] When set to a portrait orientation, this project setting does not flip the project resolution's width and height automatically. Instead, you have to set [member display/window/size/width] and [member display/window/size/height] accordingly.
If [code]true[/code], allows per-pixel transparency for the window background. This affects performance, so leave it on [code]false[/code] unless you need it.
See [member OS.window_per_pixel_transparency_enabled] for more details.
[b]Note:[/b] This feature is implemented on HTML5, Linux, macOS, Windows, and Android.
Sets the main window to full screen when the project starts. Note that this is not [i]exclusive[/i] fullscreen. On Windows and Linux, a borderless window is used to emulate fullscreen. On macOS, a new desktop is used to display the running project.
Regardless of the platform, enabling fullscreen will change the window size to match the monitor's size. Therefore, make sure your project supports [url=https://docs.godotengine.org/en/3.4/tutorials/rendering/multiple_resolutions.html]multiple resolutions[/url] when enabling fullscreen mode.
Sets the game's main viewport height. On desktop platforms, this is the default window size. Stretch mode settings also use this as a reference when enabled.
Sets the game's main viewport width. On desktop platforms, this is the default window size. Stretch mode settings also use this as a reference when enabled.
If [code]true[/code], enables vertical synchronization. This eliminates tearing that may appear in moving scenes, at the cost of higher input latency and stuttering at lower framerates. If [code]false[/code], vertical synchronization will be disabled, however, many platforms will enforce it regardless (such as mobile platforms and HTML5).
If [code]Use Vsync[/code] is enabled and this setting is [code]true[/code], enables vertical synchronization via the operating system's window compositor when in windowed mode and the compositor is enabled. This will prevent stutter in certain situations. (Windows only.)
[b]Note:[/b] This option is experimental and meant to alleviate stutter experienced by some users. However, some users have experienced a Vsync framerate halving (e.g. from 60 FPS to 30 FPS) when using it.
The command-line arguments to append to Godot's own command line when running the project. This doesn't affect the editor itself.
It is possible to make another executable run Godot by using the [code]%command%[/code] placeholder. The placeholder will be replaced with Godot's own command line. Program-specific arguments should be placed [i]before[/i] the placeholder, whereas Godot-specific arguments should be placed [i]after[/i] the placeholder.
For example, this can be used to force the project to run on the dedicated GPU in a NVIDIA Optimus system on Linux:
Search path for project-specific script templates. Godot will search for script templates both in the editor-specific path and in this project-specific path.
Text-based file extensions to include in the script editor's "Find in Files" feature. You can add e.g. [code]tscn[/code] if you wish to also parse your scene files, especially if you use built-in scripts which are serialized in the scene files.
Default [InputEventAction] to confirm a focused button, menu or list item, or validate input.
[b]Note:[/b] Default [code]ui_*[/code] actions cannot be removed as they are necessary for the internal logic of several [Control]s. The events assigned to the action can however be modified.
Default [InputEventAction] to discard a modal or pending input.
[b]Note:[/b] Default [code]ui_*[/code] actions cannot be removed as they are necessary for the internal logic of several [Control]s. The events assigned to the action can however be modified.
Default [InputEventAction] to move down in the UI.
[b]Note:[/b] Default [code]ui_*[/code] actions cannot be removed as they are necessary for the internal logic of several [Control]s. The events assigned to the action can however be modified.
Default [InputEventAction] to go to the end position of a [Control] (e.g. last item in an [ItemList] or a [Tree]), matching the behavior of [constant KEY_END] on typical desktop UI systems.
[b]Note:[/b] Default [code]ui_*[/code] actions cannot be removed as they are necessary for the internal logic of several [Control]s. The events assigned to the action can however be modified.
Default [InputEventAction] to focus the next [Control] in the scene. The focus behavior can be configured via [member Control.focus_next].
[b]Note:[/b] Default [code]ui_*[/code] actions cannot be removed as they are necessary for the internal logic of several [Control]s. The events assigned to the action can however be modified.
Default [InputEventAction] to focus the previous [Control] in the scene. The focus behavior can be configured via [member Control.focus_previous].
[b]Note:[/b] Default [code]ui_*[/code] actions cannot be removed as they are necessary for the internal logic of several [Control]s. The events assigned to the action can however be modified.
Default [InputEventAction] to go to the start position of a [Control] (e.g. first item in an [ItemList] or a [Tree]), matching the behavior of [constant KEY_HOME] on typical desktop UI systems.
[b]Note:[/b] Default [code]ui_*[/code] actions cannot be removed as they are necessary for the internal logic of several [Control]s. The events assigned to the action can however be modified.
Default [InputEventAction] to move left in the UI.
[b]Note:[/b] Default [code]ui_*[/code] actions cannot be removed as they are necessary for the internal logic of several [Control]s. The events assigned to the action can however be modified.
Default [InputEventAction] to go down a page in a [Control] (e.g. in an [ItemList] or a [Tree]), matching the behavior of [constant KEY_PAGEDOWN] on typical desktop UI systems.
[b]Note:[/b] Default [code]ui_*[/code] actions cannot be removed as they are necessary for the internal logic of several [Control]s. The events assigned to the action can however be modified.
Default [InputEventAction] to go up a page in a [Control] (e.g. in an [ItemList] or a [Tree]), matching the behavior of [constant KEY_PAGEUP] on typical desktop UI systems.
[b]Note:[/b] Default [code]ui_*[/code] actions cannot be removed as they are necessary for the internal logic of several [Control]s. The events assigned to the action can however be modified.
Default [InputEventAction] to move right in the UI.
[b]Note:[/b] Default [code]ui_*[/code] actions cannot be removed as they are necessary for the internal logic of several [Control]s. The events assigned to the action can however be modified.
Default [InputEventAction] to select an item in a [Control] (e.g. in an [ItemList] or a [Tree]).
[b]Note:[/b] Default [code]ui_*[/code] actions cannot be removed as they are necessary for the internal logic of several [Control]s. The events assigned to the action can however be modified.
[b]Note:[/b] Default [code]ui_*[/code] actions cannot be removed as they are necessary for the internal logic of several [Control]s. The events assigned to the action can however be modified.
If [code]true[/code], key/touch/joystick events will be flushed just before every idle and physics frame.
If [code]false[/code], such events will be flushed only once per idle frame, between iterations of the engine.
Enabling this can greatly improve the responsiveness to input, specially in devices that need to run multiple physics frames per visible (idle) frame, because they can't run at the target frame rate.
[b]Note:[/b] Currently implemented only in Android.
This is used by servers when used in multi-threading mode (servers and visual). RIDs are preallocated to avoid stalling the server requesting them on threads. If servers get stalled too often when loading resources in a thread, increase this number.
Maximum amount of characters allowed to send as output from the debugger. Over this value, content is dropped. This helps not to stall the debugger connection.
Maximum number of errors allowed to be sent as output from the debugger. Over this value, content is dropped. This helps not to stall the debugger connection.
Maximum amount of messages allowed to send as output from the debugger. Over this value, content is dropped. This helps not to stall the debugger connection.
Maximum number of warnings allowed to be sent as output from the debugger. Over this value, content is dropped. This helps not to stall the debugger connection.
Default size of packet peer stream for deserializing Godot data (in bytes, specified as a power of two). The default value [code]16[/code] is equal to 65,536 bytes. Over this size, data is dropped.
The CA certificates bundle to use for SSL connections. If this is set to a non-empty value, this will [i]override[/i] Godot's default [url=https://github.com/godotengine/godot/blob/master/thirdparty/certs/ca-certificates.crt]Mozilla certificate bundle[/url]. If left empty, the default certificate bundle will be used.
Additional expansion applied to object bounds in the 2D physics bounding volume hierarchy. This can reduce BVH processing at the cost of a slightly coarser broadphase, which can stress the physics more in some situations.
The default value will work well in most situations. A value of 0.0 will turn this optimization off, and larger values may work better for larger, faster moving objects.
[b]Note:[/b] Used only if [member ProjectSettings.physics/2d/use_bvh] is enabled.
[b]Note:[/b] Good values are in the range [code]0[/code] to [code]1[/code]. At value [code]0[/code] objects will keep moving with the same velocity. Values greater than [code]1[/code] will aim to reduce the velocity to [code]0[/code] in less than a second e.g. a value of [code]2[/code] will aim to reduce the velocity to [code]0[/code] in half a second. A value equal to or greater than the physics frame rate ([member ProjectSettings.physics/common/physics_fps], [code]60[/code] by default) will bring the object to a stop in one iteration.
[b]Note:[/b] Good values are in the range [code]0[/code] to [code]1[/code]. At value [code]0[/code] objects will keep moving with the same velocity. Values greater than [code]1[/code] will aim to reduce the velocity to [code]0[/code] in less than a second e.g. a value of [code]2[/code] will aim to reduce the velocity to [code]0[/code] in half a second. A value equal to or greater than the physics frame rate ([member ProjectSettings.physics/common/physics_fps], [code]60[/code] by default) will bring the object to a stop in one iteration.
Threshold angular velocity under which a 2D physics body will be considered inactive. See [constant Physics2DServer.SPACE_PARAM_BODY_ANGULAR_VELOCITY_SLEEP_THRESHOLD].
Threshold linear velocity under which a 2D physics body will be considered inactive. See [constant Physics2DServer.SPACE_PARAM_BODY_LINEAR_VELOCITY_SLEEP_THRESHOLD].
Sets whether physics is run on the main thread or a separate one. Running the server on a thread increases performance, but restricts API access to only physics process.
[b]Warning:[/b] As of Godot 3.2, there are mixed reports about the use of a Multi-Threaded thread model for physics. Be sure to assess whether it does give you extra performance and no regressions when using it.
[b]Note:[/b] Good values are in the range [code]0[/code] to [code]1[/code]. At value [code]0[/code] objects will keep moving with the same velocity. Values greater than [code]1[/code] will aim to reduce the velocity to [code]0[/code] in less than a second e.g. a value of [code]2[/code] will aim to reduce the velocity to [code]0[/code] in half a second. A value equal to or greater than the physics frame rate ([member ProjectSettings.physics/common/physics_fps], [code]60[/code] by default) will bring the object to a stop in one iteration.
[b]Note:[/b] Good values are in the range [code]0[/code] to [code]1[/code]. At value [code]0[/code] objects will keep moving with the same velocity. Values greater than [code]1[/code] will aim to reduce the velocity to [code]0[/code] in less than a second e.g. a value of [code]2[/code] will aim to reduce the velocity to [code]0[/code] in half a second. A value equal to or greater than the physics frame rate ([member ProjectSettings.physics/common/physics_fps], [code]60[/code] by default) will bring the object to a stop in one iteration.
Additional expansion applied to object bounds in the 3D physics bounding volume hierarchy. This can reduce BVH processing at the cost of a slightly coarser broadphase, which can stress the physics more in some situations.
The default value will work well in most situations. A value of 0.0 will turn this optimization off, and larger values may work better for larger, faster moving objects.
[b]Note:[/b] Used only if [member ProjectSettings.physics/3d/godot_physics/use_bvh] is enabled.
If enabled, 2D and 3D physics picking behaves this way in relation to pause:
- When pause is started, every collision object that is hovered or captured (3D only) is released from that condition, getting the relevant mouse-exit callback, unless its pause mode makes it immune to pause.
- During pause, picking only considers collision objects immune to pause, sending input events and enter/exit callbacks to them as expected.
If disabled, the legacy behavior is used, which consists in queuing the picking input events during pause (so nodes won't get them) and flushing that queue on resume, against the state of the 2D/3D world at that point.
The number of fixed iterations per second. This controls how often physics simulation and [method Node._physics_process] methods are run. See also [member debug/settings/fps/force_fps].
[b]Note:[/b] This property is only read when the project starts. To change the physics FPS at runtime, set [member Engine.iterations_per_second] instead.
[b]Note:[/b] Only 8 physics ticks may be simulated per rendered frame at most. If more than 8 physics ticks have to be simulated per rendered frame to keep up with rendering, the game will appear to slow down (even if [code]delta[/code] is used consistently in physics calculations). Therefore, it is recommended not to increase [member physics/common/physics_fps] above 240. Otherwise, the game will slow down when the rendering framerate goes below 30 FPS.
Controls how much physics ticks are synchronized with real time. For 0 or less, the ticks are synchronized. Such values are recommended for network games, where clock synchronization matters. Higher values cause higher deviation of in-game clock and real clock, but allows smoothing out framerate jitters. The default value of 0.5 should be fine for most; values above 2 could cause the game to react to dropped frames with a noticeable delay and are not recommended.
[b]Note:[/b] For best results, when using a custom physics interpolation solution, the physics jitter fix should be disabled by setting [member physics/common/physics_jitter_fix] to [code]0[/code].
[b]Experimental.[/b] Calls [code]glBufferData[/code] with NULL data prior to uploading batching data. This may not be necessary but can be used for safety.
[b]Note:[/b] Use with care. You are advised to leave this as default for exports. A non-default setting that works better on your machine may adversely affect performance for end users.
[b]Experimental.[/b] If set to on, uses the [code]GL_STREAM_DRAW[/code] flag for batching buffer uploads. If off, uses the [code]GL_DYNAMIC_DRAW[/code] flag.
[b]Note:[/b] Use with care. You are advised to leave this as default for exports. A non-default setting that works better on your machine may adversely affect performance for end users.
[b]Experimental.[/b] If set to on, this applies buffer orphaning - [code]glBufferData[/code] is called with NULL data and the full buffer size prior to uploading new data. This can be important to avoid stalling on some hardware.
[b]Note:[/b] Use with care. You are advised to leave this as default for exports. A non-default setting that works better on your machine may adversely affect performance for end users.
[b]Experimental.[/b] If set to on, uses the [code]GL_STREAM_DRAW[/code] flag for legacy buffer uploads. If off, uses the [code]GL_DYNAMIC_DRAW[/code] flag.
[b]Note:[/b] Use with care. You are advised to leave this as default for exports. A non-default setting that works better on your machine may adversely affect performance for end users.
Some NVIDIA GPU drivers have a bug which produces flickering issues for the [code]draw_rect[/code] method, especially as used in [TileMap]. Refer to [url=https://github.com/godotengine/godot/issues/9913]GitHub issue 9913[/url] for details.
If [code]true[/code], this option enables a "safe" code path for such NVIDIA GPUs at the cost of performance. This option affects GLES2 and GLES3 rendering, but only on desktop platforms.
If [code]true[/code], performs 2D skinning on the CPU rather than the GPU. This provides greater compatibility with a wide range of hardware, and also may be faster in some circumstances.
Currently only available when [member rendering/batching/options/use_batching] is active.
[b]Note:[/b] Antialiased software skinned polys are not supported, and will be rendered without antialiasing.
[b]Note:[/b] Custom shaders that use the [code]VERTEX[/code] built-in operate with [code]VERTEX[/code] position [i]after[/i] skinning, whereas with hardware skinning, [code]VERTEX[/code] is the position [i]before[/i] skinning.
[b]Experimental.[/b] For regression testing against the old renderer. If this is switched on, and [code]use_batching[/code] is set, the renderer will swap alternately between using the old renderer, and the batched renderer, on each frame. This makes it easy to identify visual differences. Performance will be degraded.
Lights have the potential to prevent joining items, and break many of the performance benefits of batching. This setting enables some complex logic to allow joining items if their lighting is similar, and overlap tests pass. This can significantly improve performance in some games. Set to 0 to switch off. With large values the cost of overlap tests may lead to diminishing returns.
Sets the proportion of the total screen area (in pixels) that must be saved by a scissor operation in order to activate light scissoring. This can prevent parts of items being rendered outside the light area. Lower values scissor more aggressively. A value of 1 scissors none of the items, a value of 0 scissors every item. The power of 4 of the value is used, in order to emphasize the lower range, and multiplied by the total screen area in pixels to give the threshold. This can reduce fill rate requirements in scenes with a lot of lighting.
Enabling this setting uses the legacy method to draw batches containing only one rect. The legacy method is faster (approx twice as fast), but can cause flicker on some systems. In order to directly compare performance with the non-batching renderer you can set this to true, but it is recommended to turn this off unless you can guarantee your target hardware will work with this method.
Size of buffer reserved for batched vertices. Larger size enables larger batches, but there are diminishing returns for the memory used. This should only have a minor effect on performance.
Including color in the vertex format has a cost, however, not including color prevents batching across color changes. This threshold determines the ratio of [code]number of vertex color changes / total number of vertices[/code] above which vertices will be translated to colored format. A value of 0 will always use colored vertices, 1 will never use colored vertices.
In certain circumstances, the batcher can reorder items in order to better join them. This may result in better performance. An overlap test is needed however for each item lookahead, so there is a trade off, with diminishing returns. If you are getting no benefit, setting this to 0 will switch it off.
Sets the number of commands to lookahead to determine whether to batch render items. A value of 1 can join items consisting of single commands, 0 turns off joining. Higher values are in theory more likely to join, however this has diminishing returns and has a runtime cost so a small value is recommended.
On some platforms (especially mobile), precision issues in shaders can lead to reading 1 texel outside of bounds, particularly where rects are scaled. This can particularly lead to border artifacts around tiles in tilemaps.
This adjustment corrects for this by making a small contraction to the UV coordinates used. Note that this can result in a slight squashing of border texels.
The amount of UV contraction. This figure is divided by 1000000, and is a proportion of the total texture dimensions, where the width and height are both ranged from 0.0 to 1.0.
Use the default unless correcting for a problem on particular hardware.
Default background clear color. Overridable per [Viewport] using its [Environment]. See [member Environment.background_mode] and [member Environment.background_color] in particular. To change this default color programmatically, use [method VisualServer.set_default_clear_color].
[Environment] that will be used as a fallback environment in case a scene does not specify its own environment. The default environment is loaded in at scene load time regardless of whether you have set an environment or not. If you do not rely on the fallback environment, it is best to delete [code]default_env.tres[/code], or to specify a different default environment here.
The use of half-float vertex compression may be producing rendering errors on some platforms (especially iOS). These have been seen particularly in particles. Disabling half-float may resolve these problems.
iOS specific override for [member rendering/gles2/compatibility/disable_half_float], due to poor support for half-float vertex compression on many devices.
Max number of lights renderable per object. This is further limited by hardware support. Most devices only support 409 lights, while many devices (especially mobile) only support 102. Setting this low will slightly reduce memory usage and may decrease shader compile times.
Max amount of elements renderable in a frame. If more elements than this are visible per frame, they will not be drawn. Keep in mind elements refer to mesh surfaces and not meshes themselves. Setting this low will slightly reduce memory usage and may decrease shader compile times, particularly on web. For most uses, the default value is suitable, but consider lowering as much as possible on web export.
Max number of lights renderable in a frame. If more lights than this number are used, they will be ignored. Setting this low will slightly reduce memory usage and may decrease shader compile times, particularly on web. For most uses, the default value is suitable, but consider lowering as much as possible on web export.
Max number of reflection probes renderable in a frame. If more reflection probes than this number are used, they will be ignored. Setting this low will slightly reduce memory usage and may decrease shader compile times, particularly on web. For most uses, the default value is suitable, but consider lowering as much as possible on web export.
Shaders have a time variable that constantly increases. At some point, it needs to be rolled back to zero to avoid precision errors on shader animations. This setting specifies when (in seconds).
The default compression level for lossless WebP. Higher levels result in smaller files at the cost of compression speed. Decompression speed is mostly unaffected by the compression level. Supported values are 0 to 9. Note that compression levels above 6 are very slow and offer very little savings.
On import, mesh vertex data will be split into two streams within a single vertex buffer, one for position data and the other for interleaved attributes data. Recommended to be enabled if targeting mobile devices. Requires manual reimport of meshes after toggling.
Determines the maximum number of sphere occluders that will be used at any one time.
Although you can have many occluders in a scene, each frame the system will choose from these the most relevant based on a screen space metric, in order to give the best overall performance.
If enabled, while merging meshes, the system will also attempt to remove [Spatial] nodes that no longer have any children.
Reducing the number of [Node]s in the scene tree can make traversal more efficient, but can be switched off in case you wish to use empty [Spatial]s for markers or some other purpose.
Uses a simplified method of generating PVS (potentially visible set) data. The results may not be accurate where more than one portal join adjacent rooms.
[b]Note:[/b] Generally you should only use this option if you encounter bugs when it is set to [code]false[/code], i.e. there are problems with the default method.
If [code]true[/code], allocates the root [Viewport]'s framebuffer with high dynamic range. High dynamic range allows the use of [Color] values greater than 1. This must be set to [code]true[/code] for glow rendering to work if [member Environment.glow_hdr_threshold] is greater than or equal to [code]1.0[/code].
Lower-end override for [member rendering/quality/depth/hdr] on mobile devices, due to performance concerns or driver support. This must be set to [code]true[/code] for glow rendering to work if [member Environment.glow_hdr_threshold] is greater than or equal to [code]1.0[/code].
If [code]true[/code], performs a previous depth pass before rendering materials. This increases performance in scenes with high overdraw, when complex materials and lighting are used.
The directional shadow's size in pixels. Higher values will result in sharper shadows, at the cost of performance. The value will be rounded up to the nearest power of 2.
[b]Note:[/b] The backend in use can be overridden at runtime via the [code]--video-driver[/code] command line argument, or by the [member rendering/quality/driver/fallback_to_gles2] option if the target system does not support GLES3 and falls back to GLES2. In such cases, this property is not updated, so use [method OS.get_current_video_driver] to query it at run-time.
[b]Note:[/b] The two video drivers are not drop-in replacements for each other, so a game designed for GLES3 might not work properly when falling back to GLES2. In particular, some features of the GLES3 backend are not available in GLES2. Enabling this setting also means that both ETC and ETC2 VRAM-compressed textures will be exported on Android and iOS, increasing the data pack's size.
Maximum anisotropic filter level used for textures with anisotropy enabled. Higher values will result in sharper textures when viewed from oblique angles, at the cost of performance. Only power-of-two values are valid (2, 4, 8, 16).
Sets the number of MSAA samples to use. MSAA is used to reduce aliasing around the edges of polygons. A higher MSAA value results in smoother edges but can be significantly slower on some hardware.
[b]Note:[/b] MSAA is not available on HTML5 export using the GLES2 backend.
If set to a value greater than [code]0.0[/code], contrast-adaptive sharpening will be applied to the 3D viewport. This has a low performance cost and can be used to recover some of the sharpness lost from using FXAA. Values around [code]0.5[/code] generally give the best results. See also [member rendering/quality/filters/use_fxaa].
If [code]true[/code], uses a fast post-processing filter to make banding significantly less visible. In some cases, debanding may introduce a slightly noticeable dithering pattern. It's recommended to enable debanding only when actually needed since the dithering pattern will make lossless-compressed screenshots larger.
[b]Note:[/b] Only available on the GLES3 backend. [member rendering/quality/depth/hdr] must also be [code]true[/code] for debanding to be effective.
[b]Note:[/b] There are known issues with debanding breaking rendering on mobile platforms. Due to this, it is recommended to leave this option disabled when targeting mobile platforms.
Enables FXAA in the root Viewport. FXAA is a popular screen-space antialiasing method, which is fast but will make the image look blurry, especially at lower resolutions. It can still work relatively well at large resolutions such as 1440p and 4K. Some of the lost sharpness can be recovered by enabling contrast-adaptive sharpening (see [member rendering/quality/filters/sharpen_intensity]).
If [code]true[/code], uses nearest-neighbor mipmap filtering when using mipmaps (also called "bilinear filtering"), which will result in visible seams appearing between mipmap stages. This may increase performance in mobile as less memory bandwidth is used. If [code]false[/code], linear mipmap filtering (also called "trilinear filtering") is used.
Strategy used for framebuffer allocation. The simpler it is, the less resources it uses (but the less features it supports). If set to "2D Without Sampling" or "3D Without Effects", sample buffers will not be allocated. This means [code]SCREEN_TEXTURE[/code] and [code]DEPTH_TEXTURE[/code] will not be available in shaders and post-processing effects such as glow will not be available in [Environment].
Lower-end override for [member rendering/quality/intended_usage/framebuffer_allocation] on mobile devices, due to performance concerns or driver support.
Enable usage of bicubic sampling in baked lightmaps. This results in smoother looking lighting at the expense of more bandwidth usage. On GLES2, changes to this setting will only be applied upon restarting the application.
Size of the atlas used by reflection probes. A larger size can result in higher visual quality, while a smaller size will be faster and take up less memory.
If [code]true[/code], uses a high amount of samples to create blurred variants of reflection probes and panorama backgrounds (sky). Those blurred variants are used by rough materials.
Limits the size of the irradiance map which is normally determined by [member Sky.radiance_size]. A higher size results in a higher quality irradiance map similarly to [member rendering/quality/reflections/high_quality_ggx]. Use a higher value when using high-frequency HDRI maps, otherwise keep this as low as possible.
[b]Note:[/b] Low and mid range hardware do not support complex irradiance maps well and may crash if this is set too high.
If [code]true[/code], uses texture arrays instead of mipmaps for reflection probes and panorama backgrounds (sky). This reduces jitter noise on reflections, but costs more performance and memory.
Lower-end override for [member rendering/quality/reflections/texture_array_reflections] on mobile devices, due to performance concerns or driver support.
If [code]true[/code], forces vertex shading for all 3D [SpatialMaterial] and [ShaderMaterial] rendering. This can be used to improve performance on low-end mobile devices. The downside is that shading becomes much less accurate, with visible linear interpolation between vertices that are joined together. This can be compensated by ensuring meshes have a sufficient level of subdivision (but not too much, to avoid reducing performance). Some material features are also not supported when vertex shading is enabled.
See also [member SpatialMaterial.flags_vertex_lighting] which can be used to enable vertex shading on specific materials only.
[b]Note:[/b] This setting does not affect unshaded materials.
Lower-end override for [member rendering/quality/shading/force_vertex_shading] on mobile devices, due to performance concerns or driver support. If lighting looks broken after exporting the project to a mobile platform, try disabling this setting.
If [code]true[/code], enables new physical light attenuation for [OmniLight]s and [SpotLight]s. This results in more realistic lighting appearance with a very small performance cost. When physical light attenuation is enabled, lights will appear to be darker as a result of the new attenuation formula. This can be compensated by adjusting the lights' energy or attenuation values.
Changes to this setting will only be applied upon restarting the application.
Size for cubemap into which the shadow is rendered before being copied into the shadow atlas. A higher number can result in higher resolution shadows when used with a higher [member rendering/quality/shadow_atlas/size]. Setting higher than a quarter of the [member rendering/quality/shadow_atlas/size] will not result in a perceptible increase in visual quality.
Shadow filter mode. Higher-quality settings result in smoother shadows that flicker less when moving. "Disabled" is the fastest option, but also has the lowest quality. "PCF5" is smoother but is also slower. "PCF13" is the smoothest option, but is also the slowest.
[b]Note:[/b] When using the GLES2 backend, the "PCF13" option actually uses 16 samples to emulate linear filtering in the shader. This results in a shadow appearance similar to the one produced by the GLES3 backend.
Allows [MeshInstance] to perform skinning on the CPU when the hardware doesn't support the default GPU skinning process with GLES2.
If [code]false[/code], an alternative skinning process on the GPU is used in this case (slower in most cases).
See also [member rendering/quality/skinning/force_software_skinning].
[b]Note:[/b] When the software skinning fallback is triggered, custom vertex shaders will behave in a different way, because the bone transform will be already applied to the modelview matrix.
Additional expansion applied to object bounds in the 3D rendering bounding volume hierarchy. This can reduce BVH processing at the cost of a slightly reduced accuracy.
The default value will work well in most situations. A value of 0.0 will turn this optimization off, and larger values may work better for larger, faster moving objects.
[b]Note:[/b] Used only if [member ProjectSettings.rendering/quality/spatial_partitioning/use_bvh] is enabled.
If [code]true[/code], the texture importer will import VRAM-compressed textures using the BPTC algorithm. This texture compression algorithm is only supported on desktop platforms, and only when using the GLES3 renderer.
[b]Note:[/b] Changing this setting does [i]not[/i] impact textures that were already imported before. To make this setting apply to textures that were already imported, exit the editor, remove the [code].import/[/code] folder located inside the project folder then restart the editor (see [member application/config/use_hidden_project_data_directory]).
If [code]true[/code], the texture importer will import VRAM-compressed textures using the Ericsson Texture Compression algorithm. This algorithm doesn't support alpha channels in textures.
[b]Note:[/b] Changing this setting does [i]not[/i] impact textures that were already imported before. To make this setting apply to textures that were already imported, exit the editor, remove the [code].import/[/code] folder located inside the project folder then restart the editor (see [member application/config/use_hidden_project_data_directory]).
If [code]true[/code], the texture importer will import VRAM-compressed textures using the Ericsson Texture Compression 2 algorithm. This texture compression algorithm is only supported when using the GLES3 renderer.
[b]Note:[/b] Changing this setting does [i]not[/i] impact textures that were already imported before. To make this setting apply to textures that were already imported, exit the editor, remove the [code].import/[/code] folder located inside the project folder then restart the editor (see [member application/config/use_hidden_project_data_directory]).
If [code]true[/code], the texture importer will import VRAM-compressed textures using the PowerVR Texture Compression algorithm. This texture compression algorithm is only supported on iOS.
[b]Note:[/b] Changing this setting does [i]not[/i] impact textures that were already imported before. To make this setting apply to textures that were already imported, exit the editor, remove the [code].import/[/code] folder located inside the project folder then restart the editor (see [member application/config/use_hidden_project_data_directory]).
If [code]true[/code], the texture importer will import VRAM-compressed textures using the S3 Texture Compression algorithm. This algorithm is only supported on desktop platforms and consoles.
[b]Note:[/b] Changing this setting does [i]not[/i] impact textures that were already imported before. To make this setting apply to textures that were already imported, exit the editor, remove the [code].import/[/code] folder located inside the project folder then restart the editor (see [member application/config/use_hidden_project_data_directory]).
