GTiff -- GeoTIFF File Format

Driver short name


Driver built-in by default

This driver is built-in by default

Most forms of TIFF and GeoTIFF files are supported by GDAL for reading, and somewhat less varieties can be written.

GDAL also supports reading and writing BigTIFF files (evolution of the TIFF format to support files larger than 4 GB).

Currently band types of Byte, UInt16, Int16, UInt32, Int32, Float32, Float64, CInt16, CInt32, CFloat32 and CFloat64 are supported for reading and writing. Paletted images will return palette information associated with the band. The compression formats listed below should be supported for reading as well.

As well, one bit files, and some other unusual formulations of GeoTIFF file, such as YCbCr color model files, are automatically translated into RGBA (red, green, blue, alpha) form, and treated as four eight bit bands.

Driver capabilities

Supports CreateCopy()

This driver supports the GDALDriver::CreateCopy() operation

Supports Create()

This driver supports the GDALDriver::Create() operation

Supports Georeferencing

This driver supports georeferencing

Supports VirtualIO

This driver supports virtual I/O operations (/vsimem/, etc.)


Most GeoTIFF projections should be supported, with the caveat that in order to translate uncommon Projected, and Geographic coordinate systems into OGC WKT it is necessary to have the PROJ proj.db database available. It must be found at the location pointed to by the PROJ_LIB environment variable, or at one of the locations set programmatically via OSRSetPROJSearchPaths().

Georeferencing from GeoTIFF is supported in the form of one tiepoint and pixel size, a transformation matrix, or a list of GCPs.

If no georeferencing information is available in the TIFF file itself, GDAL will also check for, and use an ESRI world file with the extension .tfw, .tifw/.tiffw or .wld, as well as a MapInfo .tab file.

By default, information is fetched in following order (first listed is the most prioritary): PAM (Persistent Auxiliary metadata) .aux.xml sidecar file, INTERNAL (GeoTIFF keys and tags), TABFILE (.tab), WORLDFILE (.tfw, .tifw/.tiffw or .wld), XML (.xml)

Starting with GDAL 2.2, the allowed sources and their priority order can be changed with the GDAL_GEOREF_SOURCES configuration option (or GEOREF_SOURCES open option) whose value is a comma-separated list of the following keywords : PAM, INTERNAL, TABFILE, WORLDFILE, XML (added in 3.7), NONE. First mentioned sources are the most prioritary over the next ones. A non mentioned source will be ignored.

For example setting it to "WORLDFILE,PAM,INTERNAL" will make a geotransformation matrix from a potential worldfile prioritary over PAM or GeoTIFF.

Minimum support for extracting the CRS from ESRI .xml side car files has been added in GDAL 3.7, using the metadata.refSysInfo.RefSystem.refSysID.identCode.code CRS code.

GDAL can read and write the RPCCoefficientTag as described in the RPCs in GeoTIFF proposed extension. The tag is written only for files created with the default profile GDALGeoTIFF. For other profiles, a .RPB file is created. In GDAL data model, the RPC coefficients are stored into the RPC metadata domain. For more details, see the RFC 22: RPC Georeferencing. If .RPB or _RPC.TXT files are found, they will be used to read the RPCs, even if the RPCCoefficientTag tag is set.

Internal nodata masks

TIFF files can contain internal transparency masks. The GeoTIFF driver recognizes an internal directory as being a transparency mask when the FILETYPE_MASK bit value is set on the TIFFTAG_SUBFILETYPE tag. According to the TIFF specification, such internal transparency masks contain 1 sample of 1-bit data. Although the TIFF specification allows for higher resolutions for the internal transparency mask, the GeoTIFF driver only supports internal transparency masks of the same dimensions as the main image. Transparency masks of internal overviews are also supported.

When the GDAL_TIFF_INTERNAL_MASK configuration option is set to YES (which is the case starting with GDAL 3.9 when CreateMaskBand() is invoked on the dataset, or when it is invoked on a raster band with the GMF_PER_DATASET flag), and the GeoTIFF file is opened in update mode, the CreateMaskBand() method on a TIFF dataset or rasterband will create an internal transparency mask. Otherwise, the default behavior of nodata mask creation will be used, that is to say the creation of a .msk file, as per RFC 15: Band Masks.

1-bit internal mask band are deflate compressed. When reading them back, to make conversion between mask band and alpha band easier, mask bands are exposed to the user as being promoted to full 8 bits (i.e. the value for unmasked pixels is 255) unless the GDAL_TIFF_INTERNAL_MASK_TO_8BIT configuration option is set to NO. This does not affect the way the mask band is written (it is always 1-bit).


The GeoTIFF driver supports reading, creation and update of internal overviews. Internal overviews can be created on GeoTIFF files opened in update mode (with gdaladdo for instance). If the GeoTIFF file is opened as read only, the creation of overviews will be done in an external .ovr file. Overview are only updated on request with the BuildOverviews() method.

The block size (tile width and height) used for overviews (internal or external) can be specified by setting the GDAL_TIFF_OVR_BLOCKSIZE environment variable to a power-of-two value between 64 and 4096. The default is 128, or starting with GDAL 3.1 to use the same block size as the full-resolution dataset if possible (i.e. block height and width are equal, a power-of-two, and between 64 and 4096).

Overviews and nodata masks

The following configurations can be encountered depending if overviews and nodata masks are internal or not.

  • Internal overviews, internal nodata mask: If a GeoTIFF file has a internal transparency mask and the GeoTIFF file is opened in update mode, BuildOverviews() will automatically create overviews for the internal transparency mask.

  • External overviews, external nodata mask: When opened in read-only mode, BuildOverviews() will automatically create overviews for the external transparency mask (in a .msk.ovr file)

  • Internal overviews, external nodata mask: when running BuildOverviews() in update mode on the .tif file, starting with GDAL 3.9, overviews of the main bands and of the external .msk will be generated ( as internal overview of the .msk). Prior to GDAL 3.9, only the overviews of the main bands are generated. Overviews of the external .msk file had to be explicitly generated by running BuildOverviews() on the .msk.

For the remaining configuration, behavior is less obvious:

  • External overviews, internal nodata mask: when running BuildOverviews() in read-only mode on the .tif file, only the overviews of the main bands will be generated. Generating the overviews of the internal nodata mask is not currently supported by the driver.

Practical note: for a command line point of view, BuildOverview() in update mode means "gdaladdo the.tiff" (without -ro). Whereas BuildOverviews() in read-only mode means "gdaladdo -ro the.tiff".


GDAL can deal with the following baseline TIFF tags as dataset-level metadata :













  • GEO_METADATA: This tag may be used for embedding XML-encoded instance documents prepared using 19139-based schema (GeoTIFF DGIWG) (GDAL >= 2.3)

  • TIFF_RSID: This tag specifies a File Universal Unique Identifier, or RSID, according to DMF definition (GeoTIFF DGIWG) (GDAL >= 2.3)

The name of the metadata item to use is one of the above names ("TIFFTAG_DOCUMENTNAME", ...). On creation, those tags can for example be set with

gdal_translate in.tif out.tif -mo {TAGNAME}=VALUE

Other non standard metadata items can be stored in a TIFF file created with the profile GDALGeoTIFF (the default, see below in the Creation issues section). Those metadata items are grouped together into a XML string stored in the non standard TIFFTAG_GDAL_METADATA ASCII tag (code 42112). When BASELINE or GeoTIFF profile are used, those non standard metadata items are stored into a PAM .aux.xml file.

The value of GDALMD_AREA_OR_POINT ("AREA_OR_POINT") metadata item is stored in the GeoTIFF key RasterPixelIsPoint for GDALGeoTIFF or GeoTIFF profiles.

XMP metadata can be extracted from the file, and will be reported as XML raw content in the xml:XMP metadata domain.

EXIF metadata can be extracted from the file, and will be reported in the EXIF metadata domain.

Color Profile Metadata

GDAL can deal with the following color profile metadata in the COLOR_PROFILE domain:

  • SOURCE_ICC_PROFILE (Base64 encoded ICC profile embedded in file. If available, other tags are ignored.)

  • SOURCE_PRIMARIES_RED (xyY in "x,y,1" format for red primary.)

  • SOURCE_PRIMARIES_GREEN (xyY in "x,y,1" format for green primary)

  • SOURCE_PRIMARIES_BLUE (xyY in "x,y,1" format for blue primary)

  • SOURCE_WHITEPOINT (xyY in "x,y,1" format for whitepoint)






Note that these metadata properties can only be used on the original raw pixel data. If automatic conversion to RGB has been done, the color profile information cannot be used.

All these metadata tags can be overridden and/or used as creation options.

Nodata value

GDAL stores band nodata value in the non standard TIFFTAG_GDAL_NODATA ASCII tag (code 42113) for files created with the default profile GDALGeoTIFF. Note that all bands must use the same nodata value. When BASELINE or GeoTIFF profile are used, the nodata value is stored into a PAM .aux.xml file.

Sparse files

GDAL makes a special interpretation of a TIFF tile or strip whose offset and byte count are set to 0, that is to say a tile or strip that has no corresponding allocated physical storage. On reading, such tiles or strips are considered to be implicitly set to 0 or to the nodata value when it is defined. On writing, it is possible to enable generating such files through the Create() interface by setting the SPARSE_OK creation option to YES. Then, blocks that are never written through the IWriteBlock()/IRasterIO() interfaces will have their offset and byte count set to 0. This is particularly useful to save disk space and time when the file must be initialized empty before being passed to a further processing stage that will fill it. To avoid ambiguities with another sparse mechanism discussed in the next paragraphs, we will call such files with implicit tiles/strips "TIFF sparse files". They will be likely not interoperable with TIFF readers that are not GDAL based and would consider such files with implicit tiles/strips as defective.

Starting with GDAL 2.2, this mechanism is extended to the CreateCopy() and Open() interfaces (for update mode) as well. If the SPARSE_OK creation option (or the SPARSE_OK open option for Open()) is set to YES, even an attempt to write a all 0/nodata block will be detected so that the tile/strip is not allocated (if it was already allocated, then its content will be replaced by the 0/nodata content).

Starting with GDAL 2.2, in the case where SPARSE_OK is not defined (or set to its default value FALSE), for uncompressed files whose nodata value is not set, or set to 0, in Create() and CreateCopy() mode, the driver will delay the allocation of 0-blocks until file closing, so as to be able to write them at the very end of the file, and in a way compatible of the filesystem sparse file mechanisms (to be distinguished from the TIFF sparse file extension discussed earlier). That is that all the empty blocks will be seen as properly allocated from the TIFF point of view (corresponding strips/tiles will have valid offsets and byte counts), but will have no corresponding physical storage. Provided that the filesystem supports such sparse files, which is the case for most Linux popular filesystems (ext2/3/4, xfs, btfs, ...) or NTFS on Windows. If the file system does not support sparse files, physical storage will be allocated and filled with zeros.

Raw mode

For some TIFF formulations that have "odd" photometric color spaces, on-the-fly decoding as RGBA is done. This might not be desirable in some use cases. This behavior can be disabled by prefixing the filename with GTIFF_RAW:

For example to translate a CMYK file to another one :

gdal_translate GTIFF_RAW:in.tif out.tif -co PHOTOMETRIC=CMYK

Open options

NUM_THREADS=[<number_of_threads>/ALL_CPUS]: Enable multi-threaded compression by specifying the number of worker threads. Worth it for slow compression algorithms such as DEFLATE or LZMA. Default is compression in the main thread. Starting with GDAL 3.6, this option also enables multi-threaded decoding when RasterIO() requests intersect several tiles/strips. The GDAL_NUM_THREADS configuration option can also be used as an alternative to setting the open option.

GEOREF_SOURCES=comma-separated list with one or several of PAM, INTERNAL, TABFILE, WORLDFILE or XML (XML added in 3.7): Define which georeferencing sources are allowed and their priority order. See Georeferencing paragraph.

SPARSE_OK=[TRUE/FALSE]: Defaults to FALSE. Should empty blocks be omitted on disk? When this option is set, any attempt of writing a block whose all pixels are 0 or the nodata value will cause it not to be written at all (unless there is a corresponding block already allocated in the file). Sparse files have 0 tile/strip offsets for blocks never written and save space; however, most non-GDAL packages cannot read such files.

  • IGNORE_COG_LAYOUT_BREAK=YES/NO (GDAL >= 3.8): Updating a COG (Cloud Optimized GeoTIFF) file generally breaks part of the optimizations, but still produces a valid GeoTIFF file. Starting with GDAL 3.8, to avoid undesired loss of the COG characteristics, opening such a file in update mode will be rejected, unless this option is also set to YES (default is NO). This option has only effect on COG files and when opening in update mode, and is ignored on regular (Geo)TIFF files.

Creation Issues

GeoTIFF files can be created with any GDAL defined band type, including the complex types. Created files may have any number of bands. Files of type Byte with exactly 3 bands will be given a photometric interpretation of RGB, files of type Byte with exactly four bands will have a photometric interpretation of RGBA, while all other combinations will have a photometric interpretation of MIN_IS_BLACK. Starting with GDAL 2.2, non-standard (regarding to the intrinsics TIFF capabilities) band color interpretation, such as BGR ordering, will be handled in creation and reading, by storing them in the GDAL internal metadata TIFF tag.

The TIFF format only supports R,G,B components for palettes / color tables. Thus on writing the alpha information will be silently discarded.

You may want to read hints to generate and read cloud optimized GeoTIFF files

Creation Options

  • TFW=[YES/NO]: Force the generation of an associated ESRI world file (.tfw).See the World Files page for details.

  • RPB=[YES/NO]: Force the generation of an associated .RPB file to describe RPC (Rational Polynomial Coefficients), if RPC information is available. If not specified, this file is automatically generated if there's RPC information and that the PROFILE is not the default GDALGeoTIFF.

  • RPCTXT=[YES/NO]: Force the generation of an associated _RPC.TXT file to describe RPC (Rational Polynomial Coefficients), if RPC information is available.

  • INTERLEAVE=[BAND/PIXEL]: By default TIFF files with pixel interleaving (PLANARCONFIG_CONTIG in TIFF terminology) are created. These are slightly less efficient than BAND interleaving for some purposes, but some applications only support pixel interleaved TIFF files. Starting with GDAL 3.5, when copying from a source dataset with multiple bands which advertises a INTERLEAVE metadata item, if the INTERLEAVE creation option is not specified, the source dataset INTERLEAVE will be automatically taken into account, unless the COMPRESS creation option is specified.

  • TILED=[YES/NO]: Defaults to NO. By default striped TIFF files are created. This option can be used to force creation of tiled TIFF files.

  • BLOCKXSIZE=<integer>: Defaults to 256. Sets tile width. Must be divisible by 16.

  • BLOCKYSIZE=<integer>: Set tile or strip height. Tile height defaults to 256, strip height defaults to a value such that one strip is 8K or less. Must be divisible by 16 when TILED=YES.

  • NBITS=<value>: Create a file with less than 8 bits per sample by passing a value from 1 to 7. The apparent pixel type should be Byte. Values of n=9...15 (UInt16 type) and n=17...31 (UInt32 type) are also accepted. From GDAL 2.2, n=16 is accepted for Float32 type to generate half-precision floating point values.


    • JPEG should generally only be used with Byte data (8 bit per channel). Better compression for RGB images can be obtained by using the PHOTOMETRIC=YCBCR colorspace with a 4:2:2 subsampling of the Y,Cb,Cr components.

      Starting with GDAL 3.4, if GDAL is built with its internal libtiff, read and write support for JPEG-in-TIFF compressed images with 12-bit sample is enabled by default (if JPEG support is also enabled), using GDAL internal libjpeg (based on IJG libjpeg-6b, with additional changes for 12-bit sample support). Support for JPEG with 12-bit sample is independent of whether 8-bit JPEG support is enabled through internal IJG libjpeg-6b or external libjpeg (like libjpeg-turbo)

    • CCITTFAX3, CCITTFAX4 or CCITRLE compression should only be used with 1bit (NBITS=1) data

    • LZW, DEFLATE and ZSTD compressions can be used with the PREDICTOR creation option.

    • ZSTD is available since GDAL 2.3 when using internal libtiff and if GDAL built against libzstd >=1.0, or if built against external libtiff with zstd support.

    • LERC and LERC_DEFLATE are available only when using internal libtiff for GDAL < 3.3.0. Since GDAL 3.3.0, LERC compression is also available when building GDAL against external libtiff >= 4.3.0, built itself against

    • LERC_ZSTD is available when LERC and ZSTD are available.

    • JXL is for JPEG-XL, and is only available when using internal libtiff and building GDAL against . Supported data types are Byte, UInt16 and Float32 only. For GDAL < 3.6.0, JXL compression may only be used alongside INTERLEAVE=PIXEL (the default) on datasets with 4 bands or less.

    • NONE is the default.

  • NUM_THREADS=[<integer>/NUM_CPUS]: Defaults to 1. Enable multi-threaded compression by specifying the number of worker threads. Worthwhile for slow compression algorithms such as DEFLATE or LZMA. Will be ignored for JPEG. Default is compression in the main thread.

  • PREDICTOR=[1/2/3]: Defaults to 1. Set the predictor for LZW, DEFLATE and ZSTD compression. The default is 1 (no predictor), 2 is horizontal differencing and 3 is floating point prediction. PREDICTOR=2 is only supported for 8, 16, 32 and 64 bit samples (support for 64 bit was added in libtiff > 4.3.0). PREDICTOR=3 is only supported for 16, 32 and 64 bit floating-point data.

  • DISCARD_LSB=[<nbits>/nbits_band1,nbits_band2,...nbits_bandN]: Set the number of least-significant bits to clear, possibly different per band. Lossy compression scheme to be best used with PREDICTOR=2 and LZW/DEFLATE/ZSTD compression.

  • SPARSE_OK=[TRUE/FALSE]: Should newly created files (through Create() interface) be allowed to be sparse? Sparse files have 0 tile/strip offsets for blocks never written and save space; however, most non-GDAL packages cannot read such files. Starting with GDAL 2.2, SPARSE_OK=TRUE is also supported through the CreateCopy() interface. Starting with GDAL 2.2, even an attempt to write a block whose all pixels are 0 or the nodata value will cause it not to be written at all (unless there is a corresponding block already allocated in the file). The default is FALSE.

  • JPEG_QUALITY=1-100: Defaults to 75. Set the JPEG quality when using JPEG compression.

    Low values result in higher compression ratios, but poorer image quality with strong blocking artifacts. Values above 95 are not meaningfully better quality but can be substantially larger.

  • JPEGTABLESMODE=[0/1/2/3]: Defaults to 1. Configure how and where JPEG quantization and Huffman tables are written in the TIFF JpegTables tag and strip/tile.

    • 0: JpegTables is not written. Each strip/tile contains its own quantization tables and use optimized Huffman coding.

    • 1: JpegTables is written with only the quantization tables. Each strip/tile refers to those quantized tables and use optimized Huffman coding. This is generally the optimal choice for smallest file size, and consequently is the default.

    • 2: JpegTables is written with only the default Huffman tables. Each strip/tile refers to those Huffman tables (thus no optimized Huffman coding) and contains its own quantization tables (identical). This option has no anticipated practical value.

    • 3: JpegTables is written with the quantization and default Huffman tables. Each strip/tile refers to those tables (thus no optimized Huffman coding). This option could perhaps with some data be more efficient than 1, but this should only occur in rare circumstances.

  • ZLEVEL=[1-9] or [1-12]: Defaults to 6. Set the level of compression when using DEFLATE compression (or LERC_DEFLATE). A value of 9 (resp. 12) is best/slowest when using zlib (resp. libdeflate), and 1 is least/fastest compression.

  • ZSTD_LEVEL=[1-22]: Defaults to 9. Set the level of compression when using ZSTD compression (or LERC_ZSTD). A value of 22 is best (very slow), and 1 is least compression.

  • MAX_Z_ERROR=<threshold>: Defaults to 0. Set the maximum error threshold on values for LERC/LERC_DEFLATE/LERC_ZSTD compression. The default is 0 (lossless).

  • MAX_Z_ERROR_OVERVIEW=<threshold>: (GDAL >= 3.8) Set the maximum error threshold on values for LERC/LERC_DEFLATE/LERC_ZSTD compression, on overviews. The default is the value of MAX_Z_ERROR

  • WEBP_LEVEL=[1-100]: Defaults to 75. Set the WEBP quality level when using WEBP compression. A value of 100 is best quality (least compression), and 1 is worst quality (best compression).

  • WEBP_LOSSLESS=[TRUE/FALSE]: Requires libwebp >= 0.1.4. By default, lossy compression is used. If set to True, lossless compression will be used. There is a significant time penalty for each tile/strip with lossless WebP compression, so you may want to increase the BLOCKYSIZE value for strip layout.

  • JXL_LOSSLESS=[YES/NO]: Defaults to YES. Set whether JPEG-XL compression should be lossless (YES) or lossy (NO). For lossy compression, the underlying data should be either gray, gray+alpha, rgb or rgb+alpha. For lossy compression, the pixel data should span the whole range of the underlying pixel type (i.e. 0-255 for Byte, 0-65535 for UInt16)

  • JXL_EFFORT=[1-9]: Defaults to 5. Level of effort for JPEG-XL compression. The higher, the smaller file and slower compression time.

  • JXL_DISTANCE=[0.1-15]: Defaults to 1.0. Distance level for lossy JPEG-XL compression. It is specified in multiples of a just-noticeable difference. (cf butteraugli for the definition of the distance) That is, 0 is mathematically lossless, 1 should be visually lossless, and higher distances yield denser and denser files with lower and lower fidelity. The recommended range is [0.5,3].

  • JXL_ALPHA_DISTANCE=[-1/0/[0.1-15]]: (GDAL >= 3.7) Defaults to -1. Requires libjxl > 0.8.1. Distance level for alpha channel for lossy JPEG-XL compression. It is specified in multiples of a just-noticeable difference. (cf butteraugli for the definition of the distance) That is, 0 is mathematically lossless, 1 should be visually lossless, and higher distances yield denser and denser files with lower and lower fidelity. For lossy compression, the recommended range is [0.5,3]. The default value is the special value -1.0, which means to use the same distance value as non-alpha channel (ie JXL_DISTANCE).

  • PHOTOMETRIC=[MINISBLACK/MINISWHITE/RGB/CMYK/YCBCR/CIELAB/ICCLAB/ITULAB]: Set the photometric interpretation tag. Default is MINISBLACK, but if the input image has 3 or 4 bands of Byte type, then RGB will be selected. You can override default photometric using this option.

  • ALPHA=[YES/NON-PREMULTIPLIED/PREMULTIPLIED/UNSPECIFIED]: The first "extrasample" is marked as being alpha if there are any extra samples. This is necessary if you want to produce a greyscale TIFF file with an alpha band (for instance). YES is an alias for NON-PREMULTIPLIED alpha.

  • PROFILE=[GDALGeoTIFF/GeoTIFF/BASELINE]: Defaults to GDALGeoTIFF. Control what non-baseline tags are emitted by GDAL.

    • With GDALGeoTIFF (the default) various GDAL custom tags may be written.

    • With GeoTIFF only GeoTIFF tags will be added to the baseline.

    • With BASELINE no GDAL or GeoTIFF tags will be written. BASELINE is occasionally useful when writing files to be read by applications intolerant of unrecognized tags.

  • BIGTIFF=[YES/NO/IF_NEEDED/IF_SAFER]: Defaults to IF_NEEDED. Control whether the created file is a BigTIFF or a classic TIFF.

    • YES forces BigTIFF.

    • NO forces classic TIFF.

    • IF_NEEDED will only create a BigTIFF if it is clearly needed (in the uncompressed case, and image larger than 4GB. So no effect when using a compression).

    • IF_SAFER will create BigTIFF if the resulting file *might* exceed 4GB. Note: this is only a heuristic that might not always work depending on compression ratios.

    BigTIFF is a TIFF variant which can contain more than 4GiB of data (size of classic TIFF is limited by that value). This option is available if GDAL is built with libtiff library version 4.0 or higher.

    When creating a new GeoTIFF with no compression, GDAL computes in advance the size of the resulting file. If that computed file size is over 4GiB, GDAL will automatically decide to create a BigTIFF file. However, when compression is used, it is not possible in advance to known the final size of the file, so classical TIFF will be chosen. In that case, the user must explicitly require the creation of a BigTIFF with BIGTIFF=YES if the final file is anticipated to be too big for classical TIFF format. If BigTIFF creation is not explicitly asked or guessed and the resulting file is too big for classical TIFF, libtiff will fail with an error message like "TIFFAppendToStrip:Maximum TIFF file size exceeded".

  • PIXELTYPE=[DEFAULT/SIGNEDBYTE]: By setting this to SIGNEDBYTE, a new Byte file can be forced to be written as signed byte. Starting with GDAL 3.7, this option is deprecated and Int8 should rather be used.

  • COPY_SRC_OVERVIEWS=[YES/NO]: Defaults to NO. Used by (CreateCopy() only) By setting this to YES, the potential existing overviews of the source dataset will be copied to the target dataset without being recomputed. This option is typically used to generate Cloud Optimized Geotiff (starting with GDAL 3.1, the COG -- Cloud Optimized GeoTIFF generator driver can be used as a convenient shortcut). If overviews of mask band also exist, provided that the GDAL_TIFF_INTERNAL_MASK configuration option is set to YES (which is the case starting with GDAL 3.9), they will also be copied. Note that this creation option will have no effect if general options (i.e. options which are not creation options) of gdal_translate are used. Creation options related to compression are also applied to the output overviews.

  • STREAMABLE_OUTPUT=[YES/NO]: Generate a file with an Image File Directory and block order that allow streaming. See Streaming operations.

  • GEOTIFF_KEYS_FLAVOR=[STANDARD/ESRI_PE]: Defaults to STANDARD. Determine which "flavor" of GeoTIFF keys must be used to write the SRS information. The STANDARD way (default choice) will use the general accepted formulations of GeoTIFF keys, including extensions of the values accepted for ProjectedCSTypeGeoKey to new EPSG codes. The ESRI_PE flavor will write formulations that are (more) compatible of ArcGIS. At the time of writing, the ESRI_PE choice has mostly an effect when writing the EPSG:3857 (Web Mercator) SRS. For other SRS, the standard way will be used, with the addition of a ESRI_PE WKT string as the value of PCSCitationGeoKey.

  • GEOTIFF_VERSION=[AUTO/1.0/1.1]: (GDAL >= 3.1.0) Defaults to AUTO. Select the version of the GeoTIFF standard used to encode georeferencing information. 1.0 corresponds to the original 1995, GeoTIFF Revision 1.0, by Ritter & Ruth. 1.1 corresponds to the OGC standard 19-008, which is an evolution of 1.0, which clear ambiguities and fix inconsistencies mostly in the processing of the vertical part of a CRS. AUTO mode (default value) will generally select 1.0, unless the CRS to encode has a vertical component or is a 3D CRS, in which case 1.1 is used.


    Write support for GeoTIFF 1.1 requires libgeotiff 1.6.0 or later.


Multi-page TIFF files are exposed as subdatasets. On opening, a subdataset name is GTIFF_DIR:{index}:filename.tif, where {index} starts at 1.

Starting with GDAL 3.0, subdataset creation is possible by using the APPEND_SUBDATASET=YES creation option. The filename passed to Create() / CreateCopy() should be the regular filename (not with GTIFF_DIR: syntax. Creating overviews on a multi-page TIFF is not supported.

Starting with GDAL 3.2, read-only access to subdataset overviews and masks is possible provided that they are referenced by their parent IFD through the TIFFTAG_SUBIFD tag.

About JPEG compression of RGB images

When translating a RGB image to JPEG-In-TIFF, using PHOTOMETRIC=YCBCR can make the size of the image typically 2 to 3 times smaller than the default photometric value (RGB). When using PHOTOMETRIC=YCBCR, the INTERLEAVE option must be kept to its default value (PIXEL), otherwise libtiff will fail to compress the data.

Note also that the dimensions of the tiles or strips must be a multiple of 8 for PHOTOMETRIC=RGB or 16 for PHOTOMETRIC=YCBCR

Lossless conversion of JPEG into JPEG-in-TIFF

The conversion of a JPEG file (but not a JPEG-in-TIFF file) to a JPEG-in-TIFF file without decompression and compression cycles, and thus without any additional quality loss, can be done with gdal_translate (or the CreateCopy() API), if all the following conditions are met:

  • the source dataset is a JPEG file (or a VRT with a JPEG as a single SimpleSource)

  • the target dataset is a JPEG-in-TIFF file

  • no explicitly target JPEG quality is specified

  • no change in colorspace is specified

  • no sub-windowing is requested

  • and more generally, no change that alters pixel values

The generation of a tiled JPEG-in-TIFF from the original JPEG image is possible. Explicit assignment of target SRS and bounds are also possible.

So, the following commands will use the lossless copy method :

gdal_translate in.jpg out.tif -co COMPRESS=JPEG

gdal_translate in.jpg out.tif -co COMPRESS=JPEG -co TILED=YES

gdal_translate in.jpg out.tif -co COMPRESS=JPEG -a_srs EPSG:4326 -a_ullr -180 90 180 -90

whereas the following commands will not (and thus cause JPEG decompression and compression):

gdal_translate in.jpg out.tif -co COMPRESS=JPEG -co JPEG_QUALITY=60

gdal_translate in.jpg out.tif -srcwin 0 0 500 500 -co COMPRESS=JPEG

Streaming operations

The GeoTIFF driver can support reading or writing TIFF files (with some restrictions detailed below) in a streaming compatible way.

When reading a file from /vsistdin/, a named pipe (on Unix), or if forcing streamed reading by setting the TIFF_READ_STREAMING configuration option to YES, the GeoTIFF driver will assume that the TIFF Image File Directory (IFD) is at the beginning of the file, i.e. at offset 8 for a classical TIFF file or at offset 16 for a BigTIFF file. The values of the tags of array type must be contained at the beginning of file, after the end of the IFD and before the first image strip/tile. The reader must read the strips/tiles in the order they are written in the file. For a pixel interleaved file (PlanarConfiguration=Contig), the recommended order for a writer, and thus for a reader, is from top to bottom for a strip-organized file or from top to bottom, which a chunk of a block height, and left to right for a tile-organized file. For a band organized file (PlanarConfiguration=Separate), the above order is recommended with the content of the first band, then the content of the second band, etc... Technically this order corresponds to increasing offsets in the TileOffsets/StripOffsets tag. This is the order that the GDAL raster copy routine will assume.

If the order is not the one described above, the UNORDERED_BLOCKS=YES dataset metadata item will be set in the TIFF metadata domain. Each block offset can be determined by querying the "BLOCK_OFFSET_[xblock]_[yblock]" band metadata items in the TIFF metadata domain (where xblock, yblock is the coordinate of the block), and a reader could use that information to determine the appropriate reading order for image blocks.

The files that are streamed into the GeoTIFF driver may be compressed, even if the GeoTIFF driver cannot produce such files in streamable output mode (regular creation of TIFF files will produce such compatible files for streamed reading).

When writing a file to /vsistdout/, a named pipe (on Unix), or when defining the STREAMABLE_OUTPUT=YES creation option, the CreateCopy() method of the GeoTIFF driver will generate a file with the above defined constraints (related to position of IFD and block order), and this is only supported for a uncompressed file. The Create() method also supports creating streamable compatible files, but the writer must be careful to set the projection, geotransform or metadata before writing image blocks (so that the IFD is written at the beginning of the file). And when writing image blocks, the order of blocks must be the one of the above paragraph, otherwise errors will be reported.

Some examples :

gdal_translate in.tif /vsistdout/ -co TILED=YES | gzip | gunzip | gdal_translate /vsistdin/ out.tif -co TILED=YES -co COMPRESS=DEFLATE


mkfifo my_fifo
gdalwarp in.tif my_fifo -t_srs EPSG:3857
gdal_translate my_fifo out.png -of PNG

Note: not all utilities are compatible with such input or output streaming operations, and even those which may deal with such files may not manage to deal with them in all circumstances, for example if the reading driver driven by the output file is not compatible with the block order of the streamed input.

Configuration options

This paragraph lists the configuration options that can be set to alter the default behavior of the GTiff driver.

  • GTIFF_IGNORE_READ_ERRORS=[TRUE/FALSE]: Can be set to TRUE to avoid turning libtiff errors into GDAL errors. Can help reading partially corrupted TIFF files.

  • ESRI_XML_PAM=[TRUE/FALSE]: Can be set to TRUE to force metadata in the xml:ESRI domain to be written to PAM.

  • BIGTIFF_OVERVIEW=value: Control whether BigTIFF should be used for overviews. Options are the same as BIGTIFF creation option.

  • COMPRESS_OVERVIEW=value: See COMPRESS creation option. Set the compression type to use for overviews. For internal overviews, only honoured since GDAL 3.6.

  • INTERLEAVE_OVERVIEW=[BAND/PIXEL]: Control whether pixel or band interleaving is used for overviews.

  • PHOTOMETRIC_OVERVIEW=value: Set the photometric color space for overview creation

  • PREDICTOR_OVERVIEW=[1/2/3]: Set the predictor to use for overviews with LZW, DEFLATE and ZSTD compression.

  • JPEG_QUALITY_OVERVIEW=Integer between 0 and 100: Defaults to 75. Quality of JPEG compressed overviews, either internal or external.

  • WEBP_LEVEL_OVERVIEW=Integer between 0 and 100: Defaults to 75. WEBP quality level of overviews, either internal or external.

  • WEBP_LOSSLESS_OVERVIEW=[YES/NO]: (GDAL >= 3.6) Defaults to NO. Whether WEBP compression is lossless or not.

  • ZLEVEL_OVERVIEW=Integer between 1 and 9 (or 12 when libdeflate is used): (GDAL >= 3.4.1) Defaults to 6. Deflate compression level of overviews, for COMPRESS_OVERVIEW=DEFLATE or LERC_DEFLATE, either internal or external.

  • ZSTD_LEVEL_OVERVIEW=Integer between 1 and 22: (GDAL >= 3.4.1) Defaults to 9. ZSTD compression level of overviews, for COMPRESS_OVERVIEW=DEFLATE or LERC_ZSTD, either internal or external.

  • MAX_Z_ERROR_OVERVIEW=value: (GDAL >= 3.4.1) Defaults to 0 (lossless). Maximum error threshold on values for LERC/LERC_DEFLATE/LERC_ZSTD compression of overviews, either internal or external.

  • SPARSE_OK_OVERVIEW=[ON/OFF]: (GDAL >= 3.4.1) Defaults to OFF. When set to ON, blocks whose pixels are all at nodata (or 0 if no nodata is defined)

  • GDAL_TIFF_INTERNAL_MASK=[TRUE/FALSE]: Defaults to TRUE (since GDAL 3.9), FALSE (GDAL 3.8 or earlier). See Internal nodata masks section

  • GDAL_TIFF_INTERNAL_MASK_TO_8BIT=[TRUE/FALSE]: Defaults to TRUE. See Internal nodata masks section.

  • USE_RRD: Can be set to TRUE to force external overviews in the RRD format.

  • TIFF_USE_OVR=[TRUE/FALSE]: Defaults to FALSE. Can be set to TRUE to force external overviews in the GeoTIFF (.ovr) format.

  • TIFF_READ_STREAMING=[YES/NO]: If YES, assume that Image File Directory is at the beginning of the file. See Streaming operations.

  • GTIFF_POINT_GEO_IGNORE=[TRUE/FALSE]: Defaults to FALSE. Can be set to TRUE to revert back to the behavior of ancient GDAL versions regarding how PixelIsPoint is interpreted w.r.t geotransform. See RFC 33: GTiff - Fixing PixelIsPoint Interpretation for more details.

  • GTIFF_REPORT_COMPD_CS=[TRUE/FALSE]: Can be set to TRUE to avoid stripping the vertical CRS of compound CRS when reading the SRS of a file. Does not affect the writing side. Default value : FALSE for GeoTIFF 1.0 files, or TRUE (starting with GDAL 3.1) for GeoTIFF 1.1 files.

  • GDAL_ENABLE_TIFF_SPLIT=[TRUE/FALSE]: Defaults to TRUE. Can be set to FALSE to avoid all-in-one-strip files being presented as having.

  • GDAL_TIFF_OVR_BLOCKSIZE=value: See Overviews section.

  • GTIFF_LINEAR_UNITS=value: Can be set to BROKEN to read GeoTIFF files that have false easting/northing improperly set in meters when it ought to be in coordinate system linear units. (Ticket #3901).

  • TAB_APPROX_GEOTRANSFORM=[YES/NO]: Defaults to NO. To decide if an approximate geotransform is acceptable when reading a .tab file.

  • GTIFF_DIRECT_IO=[YES/NO]: Defaults to NO. Can be set to YES to use specialized RasterIO() implementations when reading un-compressed TIFF files (un-tiled only in GDAL 2.0, both un-tiled and tiled in GDAL 2.1) to avoid using the block cache. Setting it to YES even when the optimized cases do not apply should be safe (generic implementation will be used).

  • GTIFF_VIRTUAL_MEM_IO=[YES/NO/IF_ENOUGH_RAM]: Defaults to NO. Can be set to YES to use specialized RasterIO() implementations when reading un-compressed TIFF files to avoid using the block cache. This implementation relies on memory-mapped file I/O, and is currently only supported on Linux (64-bit build strongly recommended) or, starting with GDAL 2.1, on other POSIX-like systems. Setting it to YES even when the optimized cases do not apply should be safe (generic implementation will be used), but if the file exceeds RAM, disk swapping might occur if the whole file is read. Setting it to IF_ENOUGH_RAM will first check if the uncompressed file size is no bigger than the physical memory. If both GTIFF_VIRTUAL_MEM_IO and GTIFF_DIRECT_IO are enabled, the former is used in priority, and if not possible, the later is tried.

  • GDAL_NUM_THREADS enables multi-threaded compression by specifying the number of worker threads. Worth it for slow compression algorithms such as DEFLATE or LZMA. Will be ignored for JPEG. Default is compression in the main thread. Note: this configuration option also apply to other parts to GDAL (warping, gridding, ...). Starting with GDAL 3.6, this option also enables multi-threaded decoding when RasterIO() requests intersect several tiles/strips.

  • GTIFF_WRITE_TOWGS84=[AUTO/YES/NO]: (GDAL >= 3.0.3) When set to AUTO, a GeogTOWGS84GeoKey geokey will be written with TOWGS84 3 or 7-parameter Helmert transformation, if the CRS has no EPSG code attached to it, or if the TOWGS84 transformation attached to the CRS doesn't match the one imported from the EPSG code. If set to YES, then the TOWGS84 transformation attached to the CRS will be always written. If set to NO, then the transformation will not be written in any situation.

  • CHECK_DISK_FREE_SPACE=[YES/NO]: Defaults to YES. If YES, ensure that enough disk space is available before attempting to write an uncompressed, non-sparse GeoTIFF. Disabling this check is not expected to be necessary, unless GDAL is incorrectly determining the disk space available on the destination file system.

Codec Recommendations


If you don't know what to choose, choose this one.


The most commonly supported TIFF codec, especially with older non-geo software.


Used for storing quantized floating point data.


Smaller and faster than DEFLATE, but not as commonly supported.


A smaller and faster JPEG.


Next-gen JPG from the JPG group.


Slow but storage efficient.


Did you happen to have fax files from the 1990s? Use these.

See Also