Texture

class Texture

Bases: TypedWritableReferenceCount, Namable

Represents a texture object, which is typically a single 2-d image but may also represent a 1-d or 3-d texture image, or the six 2-d faces of a cube map texture.

A texture’s image data might be stored in system RAM (see get_ram_image()) or its image may be represented in texture memory on one or more GraphicsStateGuardians (see prepare()), or both. The typical usage pattern is that a texture is loaded from an image file on disk, which copies its image data into system RAM; then the first time the texture is rendered its image data is copied to texture memory (actually, to the graphics API), and the system RAM image is automatically freed.

Inheritance diagram

enum ComponentType

enumerator T_unsigned_byte = 0

enumerator T_unsigned_short = 1

enumerator T_float = 2

enumerator T_unsigned_int_24_8 = 3: Packed

enumerator T_int = 4

enumerator T_byte = 5

enumerator T_short = 6

enumerator T_half_float = 7

enumerator T_unsigned_int = 8

enum CompressionMode

enumerator CM_default = 0: on or off, according to compressed-textures

enumerator CM_off = 1: uncompressed image

enumerator CM_on = 2: whatever compression the driver supports

enumerator CM_fxt1 = 3: Specific compression modes. Use only when you really want to use a particular compression algorithm. Use with caution; not all drivers support all compression modes. You can use GSG::get_supports_compressed_texture_format() to query the available compression modes for a particular GSG.

enumerator CM_dxt1 = 4: BC1: RGB with optional binary alpha.

enumerator CM_dxt2 = 5: Like DXT3, but assumes premultiplied alpha

enumerator CM_dxt3 = 6: BC2: RGB with uncompressed 4-bit alpha.

enumerator CM_dxt4 = 7: Like DXT5, but assumes premultiplied alpha

enumerator CM_dxt5 = 8: BC3: RGB with separately compressed 8-bit alpha.

enumerator CM_pvr1_2bpp = 9

enumerator CM_pvr1_4bpp = 10

enumerator CM_rgtc = 11: BC4/BC5: 1 or 2 channels, individually compressed.

enumerator CM_etc1 = 12

enumerator CM_etc2 = 13

enumerator CM_eac = 14: EAC: 1 or 2 channels.

enum DeprecatedFilterType

Deprecated. See SamplerState.FilterType.

enumerator FT_nearest = 0

enumerator FT_linear = 1

enumerator FT_nearest_mipmap_nearest = 2

enumerator FT_linear_mipmap_nearest = 3

enumerator FT_nearest_mipmap_linear = 4

enumerator FT_linear_mipmap_linear = 5

enumerator FT_shadow = 6

enumerator FT_default = 7

enumerator FT_invalid = 8

enum DeprecatedWrapMode

Deprecated. See SamplerState.WrapMode.

enumerator WM_clamp = 0

enumerator WM_repeat = 1

enumerator WM_mirror = 2

enumerator WM_mirror_once = 3

enumerator WM_border_color = 4

enumerator WM_invalid = 5

enum Format

enumerator F_depth_stencil = 1

enumerator F_color_index = 2

enumerator F_red = 3

enumerator F_green = 4

enumerator F_blue = 5

enumerator F_alpha = 6

enumerator F_rgb = 7: any suitable RGB mode, whatever the hardware prefers

enumerator F_rgb5 = 8: 5 bits per R,G,B channel

enumerator F_rgb8 = 9: 8 bits per R,G,B channel

enumerator F_rgb12 = 10: 12 bits per R,G,B channel

enumerator F_rgb332 = 11: 3 bits per R & G, 2 bits for B

enumerator F_rgba = 12: any suitable RGBA mode, whatever the hardware prefers

enumerator F_rgbm = 13: as above, but only requires 1 bit for alpha (i.e. mask)

enumerator F_rgba4 = 14: 4 bits per R,G,B,A channel

enumerator F_rgba5 = 15: 5 bits per R,G,B channel, 1 bit alpha

enumerator F_rgba8 = 16: 8 bits per R,G,B,A channel

enumerator F_rgba12 = 17: 12 bits per R,G,B,A channel

enumerator F_luminance = 18

enumerator F_luminance_alpha = 19: 8 bits luminance, 8 bits alpha

enumerator F_luminance_alphamask = 20: 8 bits luminance, only needs 1 bit of alpha

enumerator F_rgba16 = 21: 16 bits per R,G,B,A channel

enumerator F_rgba32 = 22: 32 bits per R,G,B,A channel

enumerator F_depth_component = 23

enumerator F_depth_component16 = 24

enumerator F_depth_component24 = 25

enumerator F_depth_component32 = 26

enumerator F_r16 = 27

enumerator F_rg16 = 28

enumerator F_rgb16 = 29

enumerator F_srgb = 30: These formats are in the sRGB color space. RGB is 2.2 gamma corrected, alpha is always linear.

enumerator F_srgb_alpha = 31

enumerator F_sluminance = 32

enumerator F_sluminance_alpha = 33

enumerator F_r32i = 34: 32-bit integer, used for atomic access

enumerator F_r32 = 35

enumerator F_rg32 = 36

enumerator F_rgb32 = 37

enumerator F_r8i = 38: 8 integer bits per R channel

enumerator F_rg8i = 39: 8 integer bits per R,G channel

enumerator F_rgb8i = 40: 8 integer bits per R,G,B channel

enumerator F_rgba8i = 41: 8 integer bits per R,G,B,A channel

enumerator F_r11_g11_b10 = 42: unsigned floating-point, 11 Red, 11 Green, 10 Blue Bits

enumerator F_rgb9_e5 = 43

enumerator F_rgb10_a2 = 44

enumerator F_rg = 45

enumerator F_r16i = 46

enumerator F_rg16i = 47

enumerator F_rgb16i = 48: not recommended

enumerator F_rgba16i = 49

enumerator F_rg32i = 50

enumerator F_rgb32i = 51

enumerator F_rgba32i = 52

enum QualityLevel

enumerator QL_default = 0: according to texture-quality-level

enumerator QL_fastest = 1

enumerator QL_normal = 2

enumerator QL_best = 3

enum TextureType

enumerator TT_1d_texture = 0

enumerator TT_2d_texture = 1

enumerator TT_3d_texture = 2

enumerator TT_2d_texture_array = 3

enumerator TT_cube_map = 4

enumerator TT_buffer_texture = 5

enumerator TT_cube_map_array = 6

enumerator TT_1d_texture_array = 7

explicit Texture(std::string const &name = string())

Constructs an empty texture. The default is to set up the texture as an empty 2-d texture; follow up with one of the variants of setup_texture() if this is not what you want.

Use Texture::make_copy() to make a duplicate copy of an existing Texture.

PointerTo<AsyncFuture> async_ensure_ram_image(bool allow_compression = true, int priority = 0)

Schedules a background task that reloads the the Texture from its disk file if there is not currently a RAM image (or uncompressed RAM image, if allow_compression is false).

A higher priority value indicates that this texture should be reloaded sooner than textures with a lower priority value. If the reload hasn’t taken place yet, you can call this again to update the priority value.

If someone else reloads the texture using an explicit call to reload() while an async reload request is pending, the async reload request is cancelled.

void clear(void): Reinitializes the texture to its default, empty state (except for the name).

void clear_alpha_filename(void): Removes the alpha filename, if it was previously set. See set_alpha_filename().

void clear_alpha_fullpath(void): Removes the alpha fullpath, if it was previously set. See set_alpha_fullpath().

void clear_aux_data(std::string const &key): Removes a record previously recorded via set_aux_data().

void clear_clear_color(void): The opposite of set_clear_color. If the image is cleared after setting this, its contents may be undefined (or may in fact not be cleared at all).

void clear_filename(void): Removes the alpha filename, if it was previously set. See set_filename().

void clear_fullpath(void): Removes the alpha fullpath, if it was previously set. See set_fullpath().

void clear_image(void)

Clears the texture data without changing its format or resolution. The texture is cleared on both the graphics hardware and from RAM, unlike clear_ram_image, which only removes the data from RAM.

If a clear color has been specified using set_clear_color, the texture will be cleared using a solid color.

The texture data will be cleared the first time in which the texture is used after this method is called.

void clear_ram_image(void): Discards the current system-RAM image.

void clear_ram_mipmap_image(int n): Discards the current system-RAM image for the nth mipmap level.

void clear_ram_mipmap_images(void): Discards the current system-RAM image for all mipmap levels, except level 0 (the base image).

void clear_simple_ram_image(void): Discards the current “simple” image.

bool compress_ram_image(Texture::CompressionMode compression = CM_on, QualityLevel quality_level = QL_default, GraphicsStateGuardianBase *gsg = nullptr)

Attempts to compress the texture’s RAM image internally, to a format supported by the indicated GSG. In order for this to work, the squish library must have been compiled into Panda.

If compression is CM_on, then an appropriate compression method that is supported by the indicated GSG is automatically chosen. If the GSG pointer is NULL, any of the standard DXT1/3/5 compression methods will be used, regardless of whether it is supported.

If compression is any specific compression method, that method is used regardless of whether the GSG supports it.

quality_level determines the speed/quality tradeoff of the compression. If it is QL_default, the texture’s own quality_level parameter is used.

Returns true if successful, false otherwise.

void consider_rescale(PNMImage &pnmimage)

static void consider_rescale(PNMImage &pnmimage, std::string const &name, AutoTextureScale auto_texture_scale = ::ATS_unspecified)

Asks the PNMImage to change its scale when it reads the image, according to the whims of the Config.prc file.

For most efficient results, this method should be called after pnmimage.read_header() has been called, but before pnmimage.read(). This method may also be called after pnmimage.read(), i.e. when the pnmimage is already loaded; in this case it will rescale the image on the spot. Also see rescale_texture().

static int down_to_power_2(int value): Returns the largest power of 2 less than or equal to value.

std::size_t estimate_texture_memory(void) const

Estimates the amount of texture memory that will be consumed by loading this texture. This returns a value that is not specific to any particular graphics card or driver; it tries to make a reasonable assumption about how a driver will load the texture. It does not account for texture compression or anything fancy. This is mainly useful for debugging and reporting purposes.

Returns a value in bytes.

static std::string format_component_type(Texture::ComponentType ct): Returns the indicated ComponentType converted to a string word.

static std::string format_compression_mode(Texture::CompressionMode cm): Returns the indicated CompressionMode converted to a string word.

static std::string format_format(Texture::Format f): Returns the indicated Format converted to a string word.

static std::string format_quality_level(Texture::QualityLevel tql): Returns the indicated QualityLevel converted to a string word.

static std::string format_texture_type(Texture::TextureType tt): Returns the indicated TextureType converted to a string word.

void generate_alpha_scale_map(void): Generates a special 256x1 1-d texture that can be used to apply an arbitrary alpha scale to objects by judicious use of texture matrix. The texture is a gradient, with an alpha of 0 on the left (U = 0), and 255 on the right (U = 1).

void generate_normalization_cube_map(int size): Generates a special cube map image in the texture that can be used to apply bump mapping effects: for each texel in the cube map that is indexed by the 3-d texture coordinates (x, y, z), the resulting value is the normalized vector (x, y, z) (compressed from -1..1 into 0..1).

void generate_ram_mipmap_images(void)

Automatically fills in the n mipmap levels of the Texture, based on the texture’s source image. This requires the texture’s uncompressed ram image to be available in system memory. If it is not already, it will be fetched if possible.

This call is not normally necessary, since the mipmap levels will be generated automatically if needed. But there may be certain cases in which you would like to call this explicitly.

void generate_simple_ram_image(void): Computes the “simple” ram image by loading the main RAM image, if it is not already available, and reducing it to 16x16 or smaller. This may be an expensive operation.

bool get_active(PreparedGraphicsObjects *prepared_objects) const: Returns true if this Texture was rendered in the most recent frame within the indicated GSG.

Filename const &get_alpha_filename(void) const: Returns the alpha_filename that has been set. If this is set, it represents the name of the alpha component, which is stored in a separate file. See also get_filename(), and get_alpha_fullpath().

Filename const &get_alpha_fullpath(void) const: Returns the alpha_fullpath that has been set. This is the full path to the alpha part of the image file as it was found along the texture search path.

int get_anisotropic_degree(void) const

Returns the degree of anisotropic filtering that should be applied to the texture. This value may return 0, indicating the default value; see also get_effective_anisotropic_degree.

This returns the default sampler state for this texture; it may still be overridden by a sampler state specified at a higher level.

AutoTextureScale get_auto_texture_scale(void) const: Returns the power-of-2 texture-scaling mode that will be applied to this particular texture when it is next loaded from disk. See set_textures_power_2().

TypedReferenceCount *get_aux_data(std::string const &key) const: Returns a record previously recorded via set_aux_data(). Returns NULL if there was no record associated with the indicated key.

LColor get_border_color(void) const

Returns the solid color of the texture’s border. Some OpenGL implementations use a border for tiling textures; in Panda, it is only used for specifying the clamp color.

This returns the default sampler state for this texture; it may still be overridden by a sampler state specified at a higher level.

static TypeHandle get_class_type(void)

LColor get_clear_color(void) const: Returns the color that was previously set using set_clear_color.

vector_uchar get_clear_data(void) const: Returns the raw image data for a single pixel if it were set to the clear color.

ComponentType get_component_type(void) const: Returns the numeric interpretation of each component of the texture.

int get_component_width(void) const: Returns the number of bytes stored for each color component of a texel. Typically this is 1, but it may be 2 for 16-bit texels.

CompressionMode get_compression(void) const

Returns the compression mode requested for this particular texture, or CM_off if the texture is not to be compressed.

If a value other than CM_off is returned, this is not a guarantee that the texture is actually successfully compressed on the GSG. It may be that the GSG does not support the requested compression mode, in which case the texture may actually be stored uncompressed in texture memory.

std::size_t get_data_size_bytes(PreparedGraphicsObjects *prepared_objects) const: Returns the number of bytes which the texture is reported to consume within graphics memory, for the indicated GSG. This may return a nonzero value even if the texture is not currently resident; you should also check get_resident() if you want to know how much space the texture is actually consuming right now.

SamplerState const &get_default_sampler(void) const: This returns the default sampler state for this texture, containing the wrap and filter properties specified on the texture level; it may still be overridden by a sampler state specified at a higher level.

int get_effective_anisotropic_degree(void) const

Returns the degree of anisotropic filtering that should be applied to the texture. This value will normally not return 0, unless there is an error in the config file.

This returns the default sampler state for this texture; it may still be overridden by a sampler state specified at a higher level.

SamplerState::FilterType get_effective_magfilter(void) const

Returns the filter mode of the texture for magnification, with special treatment for FT_default. This will normally not return FT_default, unless there is an error in the config file.

This returns the default sampler state for this texture; it may still be overridden by a sampler state specified at a higher level.

SamplerState::FilterType get_effective_minfilter(void) const

Returns the filter mode of the texture for minification, with special treatment for FT_default. This will normally not return FT_default, unless there is an error in the config file.

This returns the default sampler state for this texture; it may still be overridden by a sampler state specified at a higher level.

QualityLevel get_effective_quality_level(void) const: Returns the current quality_level hint, or the global default quality_level if this texture doesn’t specify a quality level. This value will not normally return QL_default (unless there is an error in the config file)

int get_expected_mipmap_num_pages(int n) const: Returns the total number of pages that the nth mipmap level should have, based on the texture’s size. This is usually the same as get_expected_mipmap_z_size(), except for a multiview texture, in which case it is get_expected_mipmap_z_size() * get_num_views().

int get_expected_mipmap_x_size(int n) const: Returns the x_size that the nth mipmap level should have, based on the texture’s size.

int get_expected_mipmap_y_size(int n) const: Returns the y_size that the nth mipmap level should have, based on the texture’s size.

int get_expected_mipmap_z_size(int n) const: Returns the z_size that the nth mipmap level should have, based on the texture’s size.

int get_expected_num_mipmap_levels(void) const

Returns the number of mipmap levels that should be defined for this texture, given the texture’s size.

Note that this returns a number appropriate for mipmapping, even if the texture does not currently have mipmapping enabled.

std::size_t get_expected_ram_image_size(void) const: Returns the number of bytes that ought to be used by the in-memory image, based on the texture parameters.

std::size_t get_expected_ram_mipmap_image_size(int n) const: Returns the number of bytes that ought to be used by the in-memory image for mipmap level n, based on the texture parameters.

std::size_t get_expected_ram_mipmap_page_size(int n) const: Returns the number of bytes that should be used per each Z page of the 3-d texture, for mipmap level n. For a 2-d or 1-d texture, this is the same as get_expected_ram_mipmap_view_size(n).

std::size_t get_expected_ram_mipmap_view_size(int n) const: Returns the number of bytes that ought to be used by each view of the in- memory image for mipmap level n, based on the texture parameters. For a normal, non-multiview texture, this is the same as get_expected_ram_mipmap_image_size(n).

std::size_t get_expected_ram_page_size(void) const: Returns the number of bytes that should be used per each Z page of the 3-d texture. For a 2-d or 1-d texture, this is the same as get_expected_ram_image_size().

Filename const &get_filename(void) const: Returns the filename that has been set. This is the name of the file as it was requested. Also see get_fullpath().

Format get_format(void) const: Returns the format of the texture, which represents both the semantic meaning of the texels and, to some extent, their storage information.

Filename const &get_fullpath(void) const: Returns the fullpath that has been set. This is the full path to the file as it was found along the texture search path.

UpdateSeq get_image_modified(void) const: Returns a sequence number which is guaranteed to change at least every time the texture image data (including mipmap levels) are modified.

SparseArray get_image_modified_pages(UpdateSeq since, int n = 0) const: Returns a SparseArray containing all the image pages that have been modified since the given UpdateSeq value.

virtual bool get_keep_ram_image(void) const: Returns the flag that indicates whether this Texture is eligible to have its main RAM copy of the texture memory dumped when the texture is prepared for rendering. See set_keep_ram_image().

bool get_loaded_from_image(void) const: Returns the flag that indicates the texture has been loaded from a disk file or PNMImage. See set_loaded_from_image().

bool get_loaded_from_txo(void) const: Returns the flag that indicates the texture has been loaded from a txo file.

SamplerState::FilterType get_magfilter(void) const

Returns the filter mode of the texture for magnification. The mipmap constants are invalid here. This may return FT_default; see also get_effective_minfilter().

This returns the default sampler state for this texture; it may still be overridden by a sampler state specified at a higher level.

bool get_match_framebuffer_format(void) const: Returns true if the special flag was set that indicates to the GSG that the Texture’s format should be chosen to exactly match the framebuffer’s format, presumably because the application intends to copy image data from the framebuffer into the Texture (or vice-versa).

SamplerState::FilterType get_minfilter(void) const

Returns the filter mode of the texture for minification. If this is one of the mipmap constants, then the texture requires mipmaps. This may return FT_default; see also get_effective_minfilter().

This returns the default sampler state for this texture; it may still be overridden by a sampler state specified at a higher level.

int get_num_components(void) const: Returns the number of color components for each texel of the texture image. This is 3 for an rgb texture or 4 for an rgba texture; it may also be 1 or 2 for a grayscale texture.

int get_num_loadable_ram_mipmap_images(void) const

Returns the number of contiguous mipmap levels that exist in RAM, up until the first gap in the sequence. It is guaranteed that at least mipmap levels [0, get_num_ram_mipmap_images()) exist.

The number returned will never exceed the number of required mipmap images based on the size of the texture and its filter mode.

This method is different from get_num_ram_mipmap_images() in that it returns only the number of mipmap levels that can actually be usefully loaded, regardless of the actual number that may be stored.

int get_num_pages(void) const: Returns the total number of pages in the texture. Each “page” is a 2-d texture image within the larger image–a face of a cube map, or a level of a 3-d texture. Normally, get_num_pages() is the same as get_z_size(). However, in a multiview texture, this returns get_z_size() * get_num_views().

int get_num_ram_mipmap_images(void) const

Returns the maximum number of mipmap level images available in system memory. The actual number may be less than this (that is, there might be gaps in the sequence); use has_ram_mipmap_image() to verify each level.

Also see get_num_loadable_ram_mipmap_images().

int get_num_views(void) const

Returns the number of “views” in the texture. A view is a completely separate image stored within the Texture object. Most textures have only one view, but a stereo texture, for instance, may have two views, a left and a right image. Other uses for multiple views are not yet defined.

If this value is greater than one, the additional views are accessed as additional pages beyond get_z_size().

int get_orig_file_x_size(void) const: Returns the X size of the original disk image that this Texture was loaded from (if it came from a disk file), before any automatic rescaling by Panda.

int get_orig_file_y_size(void) const: Returns the Y size of the original disk image that this Texture was loaded from (if it came from a disk file), before any automatic rescaling by Panda.

int get_orig_file_z_size(void) const: Returns the Z size of the original disk image that this Texture was loaded from (if it came from a disk file), before any automatic rescaling by Panda.

int get_pad_x_size(void) const: Returns size of the pad region. See set_pad_size.

int get_pad_y_size(void) const: Returns size of the pad region. See set_pad_size.

int get_pad_z_size(void) const: Returns size of the pad region. See set_pad_size.

bool get_post_load_store_cache(void) const: Returns the setting of the post_load_store_cache flag. See set_post_load_store_cache().

UpdateSeq get_properties_modified(void) const: Returns a sequence number which is guaranteed to change at least every time the texture properties (unrelated to the image) are modified.

QualityLevel get_quality_level(void) const: Returns the current quality_level hint. See set_quality_level(). This value may return QL_default; see get_effective_quality_level().

CPTA_uchar get_ram_image(void)

Returns the system-RAM image data associated with the texture. If the texture does not currently have an associated RAM image, and the texture was generated by loading an image from a disk file (the most common case), this forces the reload of the same texture. This can happen if keep_texture_ram is configured to false, and we have previously prepared this texture with a GSG.

Note that it is not correct to call has_ram_image() first to test whether this function will fail. A false return value from has_ram_image() indicates only that get_ram_image() may need to reload the texture from disk, which it will do automatically. However, you can call might_have_ram_image(), which will return true if the ram image exists, or there is a reasonable reason to believe it can be loaded.

On the other hand, it is possible that the texture cannot be found on disk or is otherwise unavailable. If that happens, this function will return NULL. There is no way to predict with 100% accuracy whether get_ram_image() will return NULL without calling it first; might_have_ram_image() is the closest.

CPTA_uchar get_ram_image_as(std::string const &requested_format)

Returns the uncompressed system-RAM image data associated with the texture. Rather than just returning a pointer to the data, like get_uncompressed_ram_image, this function first processes the data and reorders the components using the specified format string, and places these into a new char array.

The ‘format’ argument should specify in which order the components of the texture must be. For example, valid format strings are “RGBA”, “GA”, “ABRG” or “AAA”. A component can also be written as “0” or “1”, which means an empty/black or a full/white channel, respectively.

This function is particularly useful to copy an image in-memory to a different library (for example, PIL or wxWidgets) that require a different component order than Panda’s internal format, BGRA. Note, however, that this conversion can still be too slow if you want to do it every frame, and should thus be avoided for that purpose.

The only requirement for the reordering is that an uncompressed image must be available. If the RAM image is compressed, it will attempt to re-load the texture from disk, if it doesn’t find an uncompressed image there, it will return NULL.

CompressionMode get_ram_image_compression(void) const: Returns the compression mode in which the ram image is already stored pre- compressed. If this is other than CM_off, you cannot rely on the contents of the ram image to be anything predicatable (it will not be an array of x by y pixels, and it probably won’t have the same length as get_expected_ram_image_size()).

std::size_t get_ram_image_size(void) const: Returns the total number of bytes used by the in-memory image, across all pages and views, or 0 if there is no in-memory image.

CPTA_uchar get_ram_mipmap_image(int n) const: Returns the system-RAM image data associated with the nth mipmap level, if present. Returns NULL if the nth mipmap level is not present.

std::size_t get_ram_mipmap_image_size(int n) const: Returns the number of bytes used by the in-memory image for mipmap level n, or 0 if there is no in-memory image for this mipmap level.

std::size_t get_ram_mipmap_page_size(int n) const

Returns the number of bytes used by the in-memory image per page for mipmap level n, or 0 if there is no in-memory image for this mipmap level.

For a non-compressed texture, this is the same as get_expected_ram_mipmap_page_size(). For a compressed texture, this may be a smaller value. (We do assume that all pages will be the same size on a compressed texture).

void *get_ram_mipmap_pointer(int n) const: Similiar to get_ram_mipmap_image(), however, in this case the void pointer for the given ram image is returned. This will be NULL unless it has been explicitly set.

std::size_t get_ram_mipmap_view_size(int n) const

Returns the number of bytes used by the in-memory image per view for mipmap level n, or 0 if there is no in-memory image for this mipmap level.

A “view” is a collection of z_size pages for each mipmap level. Most textures have only one view, except for multiview or stereo textures.

For a non-compressed texture, this is the same as get_expected_ram_mipmap_view_size(). For a compressed texture, this may be a smaller value. (We do assume that all pages will be the same size on a compressed texture).

std::size_t get_ram_page_size(void) const

Returns the number of bytes used by the in-memory image per page, or 0 if there is no in-memory image.

For a non-compressed texture, this is the same as get_expected_ram_page_size(). For a compressed texture, this may be a smaller value. (We do assume that all pages will be the same size on a compressed texture).

std::size_t get_ram_view_size(void) const: Returns the number of bytes used by the in-memory image per view, or 0 if there is no in-memory image. Since each view is a stack of z_size pages, this is get_z_size() * get_ram_page_size().

bool get_render_to_texture(void) const

Returns a flag on the texture that indicates whether the texture is intended to be used as a direct-render target, by binding a framebuffer to a texture and rendering directly into the texture.

Normally, a user should not need to set this flag directly; it is set automatically by the low-level display code when a texture is bound to a framebuffer.

bool get_resident(PreparedGraphicsObjects *prepared_objects) const: Returns true if this Texture is reported to be resident within graphics memory for the indicated GSG.

CPTA_uchar get_simple_ram_image(void) const

Returns the image data associated with the “simple” texture image. This is provided for some textures as an option to display while the main texture image is being loaded from disk.

Unlike get_ram_image(), this function will always return immediately. Either the simple image is available, or it is not.

The “simple” image is always 4 components, 1 byte each, regardless of the parameters of the full texture. The simple image is only supported for ordinary 2-d textures.

std::size_t get_simple_ram_image_size(void) const: Returns the number of bytes used by the “simple” image, or 0 if there is no simple image.

int get_simple_x_size(void) const: Returns the width of the “simple” image in texels.

int get_simple_y_size(void) const: Returns the height of the “simple” image in texels.

LVecBase2 get_tex_scale(void) const

Returns a scale pair that is suitable for applying to geometry via NodePath::set_tex_scale(), which will convert texture coordinates on the geometry from the range 0..1 into the appropriate range to render the video part of the texture.

This is necessary only if a padding size has been set via set_pad_size() (or implicitly via something like “textures-power-2 pad” in the config.prc file). In this case, this is a convenient way to generate UV’s that reflect the built-in padding size.

TextureType get_texture_type(void) const: Returns the overall interpretation of the texture.

AutoTextureScale get_textures_power_2(void): This flag returns ATS_none, ATS_up, or ATS_down and controls the scaling of textures in general. It is initialized from the config variable of the same name, but it can be subsequently adjusted. See also get_auto_texture_scale().

CPTA_uchar get_uncompressed_ram_image(void)

Returns the system-RAM image associated with the texture, in an uncompressed form if at all possible.

If get_ram_image_compression() is CM_off, then the system-RAM image is already uncompressed, and this returns the same thing as get_ram_image().

If get_ram_image_compression() is anything else, then the system-RAM image is compressed. In this case, the image will be reloaded from the original file (not from the cache), in the hopes that an uncompressed image will be found there.

If an uncompressed image cannot be found, returns NULL.

GeomEnums::UsageHint get_usage_hint(void) const: Returns the usage hint specified for buffer textures, or UH_unspecified for all other texture types.

SamplerState::WrapMode get_wrap_u(void) const

Returns the wrap mode of the texture in the U direction.

This returns the default sampler state for this texture; it may still be overridden by a sampler state specified at a higher level.

SamplerState::WrapMode get_wrap_v(void) const

Returns the wrap mode of the texture in the V direction.

This returns the default sampler state for this texture; it may still be overridden by a sampler state specified at a higher level.

SamplerState::WrapMode get_wrap_w(void) const

Returns the wrap mode of the texture in the W direction. This is the depth direction of 3-d textures.

This returns the default sampler state for this texture; it may still be overridden by a sampler state specified at a higher level.

int get_x_size(void) const: Returns the width of the texture image in texels.

int get_y_size(void) const: Returns the height of the texture image in texels. For a 1-d texture, this will be 1.

int get_z_size(void) const: Returns the depth of the texture image in texels. For a 1-d texture or 2-d texture, this will be 1. For a cube map texture, this will be 6.

bool has_all_ram_mipmap_images(void) const: Returns true if all expected mipmap levels have been defined and exist in the system RAM, or false if even one mipmap level is missing.

bool has_alpha_filename(void) const: Returns true if the alpha_filename has been set and is available. See set_alpha_filename().

bool has_alpha_fullpath(void) const: Returns true if the alpha_fullpath has been set and is available. See set_alpha_fullpath().

bool has_auto_texture_scale(void) const: Returns true if set_auto_texture_scale() has been set to something other than ATS_unspecified for this particular texture.

bool has_clear_color(void) const: Returns true if a color was previously set using set_clear_color.

bool has_compression(void) const

Returns true if the texture indicates it wants to be compressed, either with CM_on or higher, or CM_default and compressed-textures is true.

If true returned, this is not a guarantee that the texture is actually successfully compressed on the GSG. It may be that the GSG does not support the requested compression mode, in which case the texture may actually be stored uncompressed in texture memory.

bool has_filename(void) const: Returns true if the filename has been set and is available. See set_filename().

bool has_fullpath(void) const: Returns true if the fullpath has been set and is available. See set_fullpath().

bool has_ram_image(void) const

Returns true if the Texture has its image contents available in main RAM, false if it exists only in texture memory or in the prepared GSG context.

Note that this has nothing to do with whether get_ram_image() will fail or not. Even if has_ram_image() returns false, get_ram_image() may still return a valid RAM image, because get_ram_image() will automatically load the texture from disk if necessary. The only thing has_ram_image() tells you is whether the texture is available right now without hitting the disk first.

Note also that if an application uses only one GSG, it may appear that has_ram_image() returns true if the texture has not yet been loaded by the GSG, but this correlation is not true in general and should not be depended on. Specifically, if an application ever uses multiple GSG’s in its lifetime (for instance, by opening more than one window, or by closing its window and opening another one later), then has_ram_image() may well return false on textures that have never been loaded on the current GSG.

bool has_ram_mipmap_image(int n) const: Returns true if the Texture has the nth mipmap level available in system memory, false otherwise. If the texture’s minfilter mode requires mipmapping (see uses_mipmaps()), and all the texture’s mipmap levels are not available when the texture is rendered, they will be generated automatically.

bool has_simple_ram_image(void) const: Returns true if the Texture has a “simple” image available in main RAM.

bool has_textures_power_2(void): If true, then get_textures_power_2 has been set using set_textures_power_2. If false, then get_textures_power_2 simply returns the config variable of the same name.

bool has_uncompressed_ram_image(void) const: Returns true if the Texture has its image contents available in main RAM and is uncompressed, false otherwise. See has_ram_image().

virtual bool is_cacheable(void) const: Returns true if there is enough information in this Texture object to write it to the bam cache successfully, false otherwise. For most textures, this is the same as has_ram_image().

bool is_prepared(PreparedGraphicsObjects *prepared_objects) const: Returns true if the texture has already been prepared or enqueued for preparation on the indicated GSG, false otherwise.

bool load(PNMImage const &pnmimage, LoaderOptions const &options = LoaderOptions())

bool load(PNMImage const &pnmimage, int z, int n, LoaderOptions const &options = LoaderOptions())

bool load(PfmFile const &pfm, LoaderOptions const &options = LoaderOptions())

bool load(PfmFile const &pfm, int z, int n, LoaderOptions const &options = LoaderOptions())

Replaces the texture with the indicated image.

Stores the indicated image in the given page and mipmap level. See read().

Texture *load_related(InternalName const *suffix) const: Loads a texture whose filename is derived by concatenating a suffix to the filename of this texture. May return NULL, for example, if this texture doesn’t have a filename.

bool load_sub_image(PNMImage const &pnmimage, int x, int y, int z = 0, int n = 0): Stores the indicated image in a region of the texture. The texture properties remain unchanged. This can be more efficient than updating an entire texture, but has a few restrictions: for one, you must ensure that the texture is still in RAM (eg. using set_keep_ram_image) and it may not be compressed.

PointerTo<Texture> make_copy(void) const

Returns a new copy of the same Texture. This copy, if applied to geometry, will be copied into texture as a separate texture from the original, so it will be duplicated in texture memory (and may be independently modified if desired).

If the Texture is a VideoTexture, the resulting duplicate may be animated independently of the original.

static PointerTo<Texture> make_from_txo(std::istream &in, std::string const &filename = "")

Constructs a new Texture object from the txo file. This is similar to Texture::read_txo(), but it constructs and returns a new object, which allows it to return a subclass of Texture (for instance, a movie texture).

Pass a real filename if it is available, or empty string if it is not.

PTA_uchar make_ram_image(void)

Discards the current system-RAM image for the texture, if any, and allocates a new buffer of the appropriate size. Returns the new buffer.

This does not affect keep_ram_image.

PTA_uchar make_ram_mipmap_image(int n)

Discards the current system-RAM image for the nth mipmap level, if any, and allocates a new buffer of the appropriate size. Returns the new buffer.

This does not affect keep_ram_image.

bool might_have_ram_image(void) const: Returns true if the texture’s image contents are currently available in main RAM, or there is reason to believe it can be loaded on demand. That is, this function returns a “best guess” as to whether get_ram_image() will succeed without actually calling it first.

PTA_uchar modify_ram_image(void)

Returns a modifiable pointer to the system-RAM image. This assumes the RAM image should be uncompressed. If the RAM image has been dumped, or is stored compressed, creates a new one.

This does not affect keep_ram_image.

PTA_uchar modify_ram_mipmap_image(int n)

Returns a modifiable pointer to the system-RAM image for the nth mipmap level. This assumes the RAM image is uncompressed; if this is not the case, raises an assertion.

This does not affect keep_ram_image.

PTA_uchar modify_simple_ram_image(void): Returns a modifiable pointer to the internal “simple” texture image. See set_simple_ram_image().

PTA_uchar new_simple_ram_image(int x_size, int y_size): Creates an empty array for the simple ram image of the indicated size, and returns a modifiable pointer to the new array. See set_simple_ram_image().

PointerTo<TexturePeeker> peek(void)

Returns a TexturePeeker object that can be used to examine the individual texels stored within this Texture by (u, v) coordinate.

If the texture has a ram image resident, that image is used. If it does not have a full ram image but does have a simple_ram_image resident, that image is used instead. If neither image is resident the full image is reloaded.

Returns NULL if the texture cannot find an image to load, or the texture format is incompatible.

PointerTo<AsyncFuture> prepare(PreparedGraphicsObjects *prepared_objects)

Indicates that the texture should be enqueued to be prepared in the indicated prepared_objects at the beginning of the next frame. This will ensure the texture is already loaded into texture memory if it is expected to be rendered soon.

Use this function instead of prepare_now() to preload textures from a user interface standpoint.

TextureContext *prepare_now(int view, PreparedGraphicsObjects *prepared_objects, GraphicsStateGuardianBase *gsg)

Creates a context for the texture on the particular GSG, if it does not already exist. Returns the new (or old) TextureContext. This assumes that the GraphicsStateGuardian is the currently active rendering context and that it is ready to accept new textures. If this is not necessarily the case, you should use prepare() instead.

Normally, this is not called directly except by the GraphicsStateGuardian; a texture does not need to be explicitly prepared by the user before it may be rendered.

bool read(Filename const &fullpath, LoaderOptions const &options = LoaderOptions())

bool read(Filename const &fullpath, Filename const &alpha_fullpath, int primary_file_num_channels, int alpha_file_channel, LoaderOptions const &options = LoaderOptions())

bool read(Filename const &fullpath, int z, int n, bool read_pages, bool read_mipmaps, LoaderOptions const &options = LoaderOptions())

bool read(Filename const &fullpath, Filename const &alpha_fullpath, int primary_file_num_channels, int alpha_file_channel, int z, int n, bool read_pages, bool read_mipmaps, BamCacheRecord *record = nullptr, LoaderOptions const &options = LoaderOptions())

Reads the named filename into the texture.

Combine a 3-component image with a grayscale image to get a 4-component image.

See the description of the full-parameter read() method for the meaning of the primary_file_num_channels and alpha_file_channel parameters.

Reads a single file into a single page or mipmap level, or automatically reads a series of files into a series of pages and/or mipmap levels.

See the description of the full-parameter read() method for the meaning of the various parameters.

Reads the texture from the indicated filename. If primary_file_num_channels is not 0, it specifies the number of components to downgrade the image to if it is greater than this number.

If the filename has the extension .txo, this implicitly reads a texture object instead of a filename (which replaces all of the texture properties). In this case, all the rest of the parameters are ignored, and the filename should not contain any hash marks; just the one named file will be read, since a single .txo file can contain all pages and mipmaps necessary to define a texture.

If alpha_fullpath is not empty, it specifies the name of a file from which to retrieve the alpha. In this case, alpha_file_channel represents the numeric channel of this image file to use as the resulting texture’s alpha channel; usually, this is 0 to indicate the grayscale combination of r, g, b; or it may be a one-based channel number, e.g. 1 for the red channel, 2 for the green channel, and so on.

If read pages is false, then z indicates the page number into which this image will be assigned. Normally this is 0 for the first (or only) page of the texture. 3-D textures have one page for each level of depth, and cube map textures always have six pages.

If read_pages is true, multiple images will be read at once, one for each page of a cube map or a 3-D texture. In this case, the filename should contain a sequence of one or more hash marks (“#”) which will be filled in with the z value of each page, zero-based. In this case, the z parameter indicates the maximum z value that will be loaded, or 0 to load all filenames that exist.

If read_mipmaps is false, then n indicates the mipmap level to which this image will be assigned. Normally this is 0 for the base texture image, but it is possible to load custom mipmap levels into the later images. After the base texture image is loaded (thus defining the size of the texture), you can call get_expected_num_mipmap_levels() to determine the maximum sensible value for n.

If read_mipmaps is true, multiple images will be read as above, but this time the images represent the different mipmap levels of the texture image. In this case, the n parameter indicates the maximum n value that will be loaded, or 0 to load all filenames that exist (up to the expected number of mipmap levels).

If both read_pages and read_mipmaps is true, then both sequences will be read; the filename should contain two sequences of hash marks, separated by some character such as a hyphen, underscore, or dot. The first hash mark sequence will be filled in with the mipmap level, while the second hash mark sequence will be the page index.

This method implicitly sets keep_ram_image to false.

bool read_dds(std::istream &in, std::string const &filename = "", bool header_only = false)

Reads the texture from a DDS file object. This is a Microsoft-defined file format; it is similar in principle to a txo object, in that it is designed to contain the texture image in a form as similar as possible to its runtime image, and it can contain mipmaps, pre-compressed textures, and so on.

As with read_txo, the filename is just for reference.

bool read_ktx(std::istream &in, std::string const &filename = "", bool header_only = false)

Reads the texture from a KTX file object. This is a Khronos-defined file format; it is similar in principle to a dds object, in that it is designed to contain the texture image in a form as similar as possible to its runtime image, and it can contain mipmaps, pre-compressed textures, and so on.

As with read_dds, the filename is just for reference.

bool read_txo(std::istream &in, std::string const &filename = "")

Reads the texture from a Panda texture object. This defines the complete Texture specification, including the image data as well as all texture properties. This only works if the txo file contains a static Texture image, as opposed to a subclass of Texture such as a movie texture.

Pass a real filename if it is available, or empty string if it is not.

bool release(PreparedGraphicsObjects *prepared_objects): Frees the texture context only on the indicated object, if it exists there. Returns true if it was released, false if it had not been prepared.

int release_all(void): Frees the context allocated on all objects for which the texture has been declared. Returns the number of contexts which have been freed.

bool reload(void)

Re-reads the Texture from its disk file. Useful when you know the image on disk has recently changed, and you want to update the Texture image.

Returns true on success, false on failure (in which case, the Texture may or may not still be valid).

bool rescale_texture(void): This method is similar to consider_rescale(), but instead of scaling a separate PNMImage, it will ask the Texture to rescale its own internal image to a power of 2, according to the config file requirements. This may be useful after loading a Texture image by hand, instead of reading it from a disk file. Returns true if the texture is changed, false if it was not.

void set_alpha_filename(Filename const &alpha_filename)

Sets the name of the file that contains the image’s alpha channel contents. Normally, this is set automatically when the image is loaded, for instance via Texture::read().

The Texture’s get_filename() function returns the name of the image file that was loaded into the buffer. In the case where a texture specified two separate files to load, a 1- or 3-channel color image and a 1-channel alpha image, this Filename is update to contain the name of the image file that was loaded into the buffer’s alpha channel.

void set_alpha_fullpath(Filename const &alpha_fullpath): Sets the full pathname to the file that contains the image’s alpha channel contents, as found along the search path. Normally, this is set automatically when the image is loaded, for instance via Texture::read().

void set_anisotropic_degree(int anisotropic_degree)

Specifies the level of anisotropic filtering to apply to the texture. Set this 0 to indicate the default value, which is specified in the texture- anisotropic-degree config variable.

To explicitly disable anisotropic filtering, set this value to 1. To explicitly enable anisotropic filtering, set it to a value higher than 1; larger numbers indicate greater degrees of filtering.

This sets the default sampler state for this texture; it may still be overridden by a sampler state specified at a higher level.

void set_auto_texture_scale(AutoTextureScale scale): Specifies the power-of-2 texture-scaling mode that will be applied to this particular texture when it is next loaded from disk. See set_textures_power_2().

void set_aux_data(std::string const &key, TypedReferenceCount *aux_data)

Records an arbitrary object in the Texture, associated with a specified key. The object may later be retrieved by calling get_aux_data() with the same key.

These data objects are not recorded to a bam or txo file.

void set_border_color(LColor const &color)

Specifies the solid color of the texture’s border. Some OpenGL implementations use a border for tiling textures; in Panda, it is only used for specifying the clamp color.

This sets the default sampler state for this texture; it may still be overridden by a sampler state specified at a higher level.

void set_clear_color(LColor const &color)

Sets the color that will be used to fill the texture image in absence of any image data. It is used when any of the setup_texture functions or clear_image is called and image data is not provided using read() or modify_ram_image().

This does not affect a texture that has already been cleared; call clear_image to clear it again.

void set_component_type(Texture::ComponentType component_type): Changes the data value for the texture components. This implicitly sets component_width as well.

void set_compression(Texture::CompressionMode compression)

Requests that this particular Texture be compressed when it is loaded into texture memory.

This refers to the internal compression of the texture image within texture memory; it is not related to jpeg or png compression, which are disk file compression formats. The actual disk file that generated this texture may be stored in a compressed or uncompressed format supported by Panda; it will be decompressed on load, and then recompressed by the graphics API if this parameter is not CM_off.

If the GSG does not support this texture compression mode, the texture will silently be loaded uncompressed.

void set_default_sampler(SamplerState const &sampler)

This sets the default sampler state for this texture, containing the wrap and filter properties specified on the texture level; it may still be overridden by a sampler state specified at a higher level. This encompasses the settings for get_wrap_u, get_minfilter, get_anisotropic_degree, etc.

This makes a copy of the SamplerState object, so future modifications of the same SamplerState will have no effect on this texture unless you call set_default_sampler again.

void set_filename(Filename const &filename)

Sets the name of the file that contains the image’s contents. Normally, this is set automatically when the image is loaded, for instance via Texture::read().

The Texture’s get_name() function used to return the filename, but now returns just the basename (without the extension), which is a more useful name for identifying an image in show code.

void set_format(Texture::Format format): Changes the format value for the texture components. This implicitly sets num_components as well.

void set_fullpath(Filename const &fullpath): Sets the full pathname to the file that contains the image’s contents, as found along the search path. Normally, this is set automatically when the image is loaded, for instance via Texture::read().

void set_keep_ram_image(bool keep_ram_image)

Sets the flag that indicates whether this Texture is eligible to have its main RAM copy of the texture memory dumped when the texture is prepared for rendering.

This will be false for most textures, which can reload their images if needed by rereading the input file. However, textures that were generated dynamically and cannot be easily reloaded will want to set this flag to true, so that the texture will always keep its image copy around.

void set_loaded_from_image(bool flag = true): Sets the flag that indicates the texture has been loaded from a disk file or PNMImage. You should also ensure the filename has been set correctly. When this flag is true, the texture may be automatically reloaded when its ram image needs to be replaced.

void set_loaded_from_txo(bool flag = true): Sets the flag that indicates the texture has been loaded from a txo file. You probably shouldn’t be setting this directly; it is set automatically when a Texture is loaded.

void set_magfilter(Texture::FilterType filter)

Sets the filtering method that should be used when viewing the texture up close.

This sets the default sampler state for this texture; it may still be overridden by a sampler state specified at a higher level.

void set_match_framebuffer_format(bool flag)

Sets the special flag that, if true, indicates to the GSG that the Texture’s format should be chosen to exactly match the framebuffer’s format, presumably because the application intends to copy image data from the framebuffer into the Texture (or vice-versa).

This sets only the graphics card’s idea of the texture format; it is not related to the system-memory format.

void set_minfilter(Texture::FilterType filter)

Sets the filtering method that should be used when viewing the texture from a distance.

This sets the default sampler state for this texture; it may still be overridden by a sampler state specified at a higher level.

void set_num_views(int num_views)

Sets the number of “views” within a texture. A view is a completely separate image stored within the Texture object. Most textures have only one view, but a stereo texture, for instance, may have two views, a left and a right image. Other uses for multiple views are not yet defined.

If this value is greater than one, the additional views are accessed as additional pages beyond get_z_size().

This also implicitly unloads the texture if it has already been loaded.

void set_orig_file_size(int x, int y, int z = 1): Specifies the size of the texture as it exists in its original disk file, before any Panda scaling.

void set_pad_size(int x = 0, int y = 0, int z = 0)

Sets the size of the pad region.

Sometimes, when a video card demands power-of-two textures, it is necessary to create a big texture and then only use a portion of it. The pad region indicates which portion of the texture is not really in use. All operations use the texture as a whole, including the pad region, unless they explicitly state that they use only the non-pad region.

Changing the texture’s size clears the pad region.

void set_post_load_store_cache(bool flag)

Sets the post_load_store_cache flag. When this is set, the next time the texture is loaded on a GSG, it will automatically extract its RAM image from the GSG and save it to the global BamCache.

This is used to store compressed RAM images in the BamCache. This flag should not be set explicitly; it is set automatically by the TexturePool when model-cache-compressed-textures is set true.

void set_quality_level(Texture::QualityLevel quality_level): Sets a hint to the renderer about the desired performance / quality tradeoff for this particular texture. This is most useful for the tinydisplay software renderer; for normal, hardware-accelerated renderers, this may have little or no effect.

void set_ram_image(PyObject *image, CompressionMode compression = CM_off, std::size_t page_size = 0)

Replaces the current system-RAM image with the new data. If compression is not CM_off, it indicates that the new data is already pre-compressed in the indicated format.

This does not affect keep_ram_image.

void set_ram_image_as(PyObject *image, std::string const &provided_format): Replaces the current system-RAM image with the new data, converting it first if necessary from the indicated component-order format. See get_ram_image_as() for specifications about the format. This method cannot support compressed image data or sub-pages; use set_ram_image() for that.

void set_ram_mipmap_image(int n, CPTA_uchar image, std::size_t page_size = 0)

Replaces the current system-RAM image for the indicated mipmap level with the new data. If compression is not CM_off, it indicates that the new data is already pre-compressed in the indicated format.

This does not affect keep_ram_image.

void set_ram_mipmap_pointer(int n, void *image, std::size_t page_size = 0)

Sets an explicit void pointer as the texture’s mipmap image for the indicated level. This is a special call to direct a texture to reference some external image location, for instance from a webcam input.

The texture will henceforth reference this pointer directly, instead of its own internal storage; the user is responsible for ensuring the data at this address remains allocated and valid, and in the correct format, during the lifetime of the texture.

void set_ram_mipmap_pointer_from_int(long long int pointer, int n, int page_size)

Accepts a raw pointer cast as an int, which is then passed to set_ram_mipmap_pointer(); see the documentation for that method.

This variant is particularly useful to set an external pointer from a language like Python, which doesn’t support void pointers directly.

void set_render_to_texture(bool render_to_texture)

Sets a flag on the texture that indicates whether the texture is intended to be used as a direct-render target, by binding a framebuffer to a texture and rendering directly into the texture.

This controls some low-level choices made about the texture object itself. For instance, compressed textures are disallowed when this flag is set true.

Normally, a user should not need to set this flag directly; it is set automatically by the low-level display code when a texture is bound to a framebuffer.

void set_simple_ram_image(CPTA_uchar image, int x_size, int y_size)

Replaces the internal “simple” texture image. This can be used as an option to display while the main texture image is being loaded from disk. It is normally a very small image, 16x16 or smaller (and maybe even 1x1), that is designed to give just enough sense of color to serve as a placeholder until the full texture is available.

The “simple” image is always 4 components, 1 byte each, regardless of the parameters of the full texture. The simple image is only supported for ordinary 2-d textures.

Also see generate_simple_ram_image(), modify_simple_ram_image(), and new_simple_ram_image().

void set_size_padded(int x = 1, int y = 1, int z = 1): Changes the size of the texture, padding if necessary, and setting the pad region as well.

void set_textures_power_2(AutoTextureScale scale): Set this flag to ATS_none, ATS_up, ATS_down, or ATS_pad to control the scaling of textures in general, if a particular texture does not override this. See also set_auto_texture_scale() for the per-texture override.

void set_wrap_u(Texture::WrapMode wrap)

This setting determines what happens when the texture is sampled with a U value outside the range 0.0-1.0. The default is WM_repeat, which indicates that the texture should repeat indefinitely.

This sets the default sampler state for this texture; it may still be overridden by a sampler state specified at a higher level.

void set_wrap_v(Texture::WrapMode wrap)

This setting determines what happens when the texture is sampled with a V value outside the range 0.0-1.0. The default is WM_repeat, which indicates that the texture should repeat indefinitely.

This sets the default sampler state for this texture; it may still be overridden by a sampler state specified at a higher level.

void set_wrap_w(Texture::WrapMode wrap)

The W wrap direction is only used for 3-d textures.

This sets the default sampler state for this texture; it may still be overridden by a sampler state specified at a higher level.

void set_x_size(int x_size): Changes the x size indicated for the texture. This also implicitly unloads the texture if it has already been loaded.

void set_y_size(int y_size): Changes the y size indicated for the texture. This also implicitly unloads the texture if it has already been loaded.

void set_z_size(int z_size): Changes the z size indicated for the texture. This also implicitly unloads the texture if it has already been loaded.

void setup_1d_texture(void)

void setup_1d_texture(int x_size, ComponentType component_type, Format format)

Sets the texture as an empty 1-d texture with no dimensions. Follow up with read() or load() to fill the texture properties and image data, or use set_clear_color to let the texture be cleared to a solid color.

Sets the texture as an empty 1-d texture with the specified dimensions and properties. Follow up with set_ram_image() or modify_ram_image() to fill the image data, or use set_clear_color to let the texture be cleared to a solid color.

void setup_2d_texture(void)

void setup_2d_texture(int x_size, int y_size, ComponentType component_type, Format format)

Sets the texture as an empty 2-d texture with no dimensions. Follow up with read() or load() to fill the texture properties and image data, or use set_clear_color to let the texture be cleared to a solid color.

Sets the texture as an empty 2-d texture with the specified dimensions and properties. Follow up with set_ram_image() or modify_ram_image() to fill the image data, or use set_clear_color to let the texture be cleared to a solid color.

void setup_2d_texture_array(int z_size = 1)

void setup_2d_texture_array(int x_size, int y_size, int z_size, ComponentType component_type, Format format)

Sets the texture as an empty 2-d texture array with no dimensions (though if you know the depth ahead of time, it saves a bit of reallocation later). Follow up with read() or load() to fill the texture properties and image data, or use set_clear_color to let the texture be cleared to a solid color.

Sets the texture as an empty 2-d texture array with the specified dimensions and properties. Follow up with set_ram_image() or modify_ram_image() to fill the image data, or use set_clear_color to let the texture be cleared to a solid color.

void setup_3d_texture(int z_size = 1)

void setup_3d_texture(int x_size, int y_size, int z_size, ComponentType component_type, Format format)

Sets the texture as an empty 3-d texture with no dimensions (though if you know the depth ahead of time, it saves a bit of reallocation later). Follow up with read() or load() to fill the texture properties and image data, or use set_clear_color to let the texture be cleared to a solid color.

Sets the texture as an empty 3-d texture with the specified dimensions and properties. Follow up with set_ram_image() or modify_ram_image() to fill the image data.

void setup_buffer_texture(int size, ComponentType component_type, Format format, GeomEnums::UsageHint usage)

Sets the texture as an empty buffer texture with the specified size and properties. Follow up with set_ram_image() or modify_ram_image() to fill the image data, or use set_clear_color to let the texture be cleared to a solid color.

Note that a buffer texture’s format needs to match the component type.

void setup_cube_map(void)

void setup_cube_map(int size, ComponentType component_type, Format format)

Sets the texture as an empty cube map texture with no dimensions. Follow up with read() or load() to fill the texture properties and image data, or use set_clear_color to let the texture be cleared to a solid color.

Sets the texture as an empty cube map texture with the specified dimensions and properties. Follow up with set_ram_image() or modify_ram_image() to fill the image data, or use set_clear_color to let the texture be cleared to a solid color.

Note that a cube map should always consist of six square images, so x_size and y_size will be the same, and z_size is always 6.

void setup_cube_map_array(int num_cube_maps)

void setup_cube_map_array(int size, int num_cube_maps, ComponentType component_type, Format format)

Sets the texture as cube map array with N cube maps. Note that this number is not the same as the z_size. Follow up with read() or load() to fill the texture properties and image data, or use set_clear_color to let the texture be cleared to a solid color.

New in version 1.10.0.

Sets the texture as cube map array with N cube maps with the specified dimensions and format. Follow up with set_ram_image() or modify_ram_image() to fill the image data, or use set_clear_color to let the texture be cleared to a solid color.

The num_cube_maps given here is multiplied by six to become the z_size of the image.

New in version 1.10.0.

void setup_texture(Texture::TextureType texture_type, int x_size, int y_size, int z_size, ComponentType component_type, Format format): Sets the texture to the indicated type and dimensions, presumably in preparation for calling read() or load(), or set_ram_image() or modify_ram_image(), or use set_clear_color to let the texture be cleared to a solid color.

bool store(PNMImage &pnmimage) const

bool store(PNMImage &pnmimage, int z, int n) const

bool store(PfmFile &pfm) const

bool store(PfmFile &pfm, int z, int n) const

Saves the texture to the indicated PNMImage, but does not write it to disk.

Saves the indicated page and mipmap level of the texture to the PNMImage.

Saves the texture to the indicated PfmFile, but does not write it to disk.

Saves the indicated page and mipmap level of the texture to the PfmFile.

static ComponentType string_component_type(std::string const &str): Returns the ComponentType corresponding to the indicated string word.

static CompressionMode string_compression_mode(std::string const &str): Returns the CompressionMode value associated with the given string representation.

static Format string_format(std::string const &str): Returns the Format corresponding to the indicated string word.

static QualityLevel string_quality_level(std::string const &str): Returns the QualityLevel value associated with the given string representation.

static TextureType string_texture_type(std::string const &str): Returns the TextureType corresponding to the indicated string word.

bool uncompress_ram_image(void)

Attempts to uncompress the texture’s RAM image internally. In order for this to work, the squish library must have been compiled into Panda, and the ram image must be compressed in a format supported by squish.

Returns true if successful, false otherwise.

static int up_to_power_2(int value): Returns the smallest power of 2 greater than or equal to value.

bool uses_mipmaps(void) const: Returns true if the minfilter settings on this texture indicate the use of mipmapping, false otherwise.

bool was_image_modified(PreparedGraphicsObjects *prepared_objects) const: Returns true if the texture needs to be re-loaded onto the indicated GSG, either because its image data is out-of-date, or because it’s not fully prepared now.

bool write(Filename const &fullpath)

bool write(Filename const &fullpath, int z, int n, bool write_pages, bool write_mipmaps)

void write(std::ostream &out, int indent_level) const

Writes the texture to the named filename.

Writes a single page or mipmap level to a single file, or automatically writes a series of pages and/or mipmap levels to a numbered series of files.

If the filename ends in the extension .txo, this implicitly writes a Panda texture object (.txo) instead of an image file. In this case, the remaining parameters are ignored, and only one file is written, which will contain all of the pages and resident mipmap levels in the texture.

If write_pages is false, then z indicates the page number to write. 3-D textures have one page number for each level of depth; cube maps have six pages number 0 through 5. Other kinds of textures have only one page, numbered 0. If there are multiple views, the range of z is increased; the total range is [0, get_num_pages()).

If write_pages is true, then all pages of the texture will be written. In this case z is ignored, and the filename should contain a sequence of hash marks (“#”) which will be filled in with the page index number.

If write_mipmaps is false, then n indicates the mipmap level number to write. Normally, this is 0, for the base texture image. Normally, the mipmap levels of a texture are not available in RAM (they are generated automatically by the graphics card). However, if you have the mipmap levels available, for instance because you called generate_ram_mipmap_images() to generate them internally, or you called GraphicsEngine::extract_texture_data() to retrieve them from the graphics card, then you may write out each mipmap level with this parameter.

If write_mipmaps is true, then all mipmap levels of the texture will be written. In this case n is ignored, and the filename should contain a sequence of hash marks (“#”) which will be filled in with the mipmap level number.

If both write_pages and write_mipmaps is true, then all pages and all mipmap levels will be written. In this case, the filename should contain two different sequences of hash marks, separated by a character such as a hyphen, underscore, or dot. The first hash mark sequence will be filled in with the mipmap level, while the second hash mark sequence will be the page index.

Not to be confused with write(Filename), this method simply describes the texture properties.

bool write_txo(std::ostream &out, std::string const &filename = "") const

Writes the texture to a Panda texture object. This defines the complete Texture specification, including the image data as well as all texture properties.

The filename is just for reference.