Supported Graphics Formats

Image Formats

This is the list of image file formats supported internally by Blender:

Format	Channel Depth	Alpha	Metadata	DPI	Extensions
BMP	8bit	✗	✗	✓	`.bmp`
Iris	8, 16bit	✓	✗	✗	`.sgi` `.rgb` `.bw`
PNG	8, 16bit	✓	✓	✓	`.png`
JPEG	8bit	✗	✓	✓	`.jpg` `.jpeg`
JPEG 2000	8, 12, 16bit	✓	✗	✗	`.jp2` `.jp2` `.j2c`
Targa	8bit	✓	✗	✗	`.tga`
Cineon & DPX	8, 10, 12, 16bit	✓	✗	✗	`.cin` `.dpx`
OpenEXR	float 16, 32bit	✓	✓	✓	`.exr`
Radiance HDR	float	✓	✗	✗	`.hdr`
TIFF	8, 16bit	✓	✗	✓	`.tif` `.tiff`

Hint

If you are not interested in technical details, a good rule of thumb for selecting output formats for your project is:

Use OpenEXR: if you intend to do compositing or color grading on these images.
Use PNG: if you intend on-screen output or encoding into multiple video formats.
Use JPEG: for on-screen output where file size is a concern and quality loss is acceptable.

All these formats support compression which can be important when rendering out animations.

Hint

Bit depths for image formats represent the following numbers of tonal levels per channel:

8: 256 levels
10: 1024 levels
12: 4096 levels
16: 65536 levels

Opening Images

Relative Path

Sets the file path to be relative to the currently opened blend-file.

See Relative Paths.

Detect Sequences

Automatically looks for image sequences in the selected images (based on the file name). Disable this when you do want to get single images that are part of a sequence.

Detect UDIMs

Automatically looks for UDIM tiles in the directory of the selected image; if matches are found they are loaded into Blender as UDIMs. This works by detecting if the filename has a .xxxx (four digit number) before the file extension.

Opening an Image Sequence

To load image sequence in any of the supported image file formats, the filename of the images must contain a digit to indicate the frame order (e.g. *-0001.jpg, *-0002.jpg, *-0003.jpg, etc, of any image format), indicating the frame.

The sequence could be opened by the selection of the images with any of the following methods by the confirmation with the Open Image button or Return.

Range: Navigate into the directory and LMB click and drag over a range of names to highlight multiple files. You can page down and continue Shift-LMB click-dragging to add more to the selection.
Batch: Shift-LMB click selected non-related stills for batch processing; each image will be one frame, in sort order, and can be a mix of file types (jpg, png, exr, etc.).
All: Press A to select/deselect all files in the directory.

Saving Images

File Format

Choose what format to save the image as.

Color Mode

Choose the color format to save the image (or video) to. Note that RGBA is not available for all image formats, check the list above for details.

BW, RGB, RGBA

Color Depth

Some image file formats support a varying number of bits per pixel. This affects the color quality and file size. Commonly used depths:

8-bit: Most common for on-screen graphics and video.
10, 12, 16-bit: Used for some formats focusing on photography and digital films (such as DPX and JPEG 2000).
16-bit Half Float: Since full 32bit float is often more than enough precision, half float can save drive space while still providing a high dynamic range.
32-bit Float: Highest quality color depth.

Note

Internally Blender's image system supports either:

8 bits per channel (4 × 8 bits).
32 bits float per channel (4 × 32 bits) -- using 4 times as much memory.

Images higher than 8 bits per channel will be converted into a float on loading into Blender.

Compression

Used to reduce the size of the image file. How this is done may vary depending on the file format and settings used.

Quality

Similar to Compression but is used for JPEG based file formats. The quality is a percentage, 0% being the maximum amount of compression and 100% is no compression.

Save As Render

Applies color transform to the saved image.

Copy

The Copy checkbox will define if the data-block will reference the newly created file or the reference will be unchanged, maintaining it with the original one.

Format Details

Cineon & DPX

Cineon is Kodak's standard for film scanning, 10 bits per channel and logarithmic. DPX has been derived from Cineon as the ANSI/SMPTE industry standard. DPX supports 16-bit colors/channels, linear as well as logarithmic. DPX is currently a widely adopted standard used in the film hardware/software industry.

DPX as well as Cineon only stores and converts the "visible" color range of values between 0.0 and 1.0 (as a result of rendering or composite).

OpenEXR

ILM's OpenEXR has become a software industry standard for HDR image files, especially because of its flexible and expandable structure.

An OpenEXR file can store multiple layers and passes. This means OpenEXR images can be loaded into a Compositor keeping render layers and passes intact.

Output Options

Available options for OpenEXR render output are:

Color Depth

Half saves images in a custom 16 bits per channel floating-point format. This reduces the actual "bit depth" to 10-bit, with a 5-bit power value and 1-bit sign.

Float (Half), Float (Full)

Codec

PXR24: Lossy algorithm from Pixar, converting 32-bit floats to 24-bit floats.
ZIP: Standard lossless compression using Zlib, operating on 16 scanlines at a time.
PIZ: Lossless wavelet compression. Compresses images with grain well.
RLE: Run-length encoded, lossless, works well when scanlines have same values.
ZIPS: Standard lossless compression using Zlib, operating on a single scanline at a time.
DWAA: JPEG-like lossy algorithm from DreamWorks.

Z Buffer

Save the depth information. In Blender, this now is written in floats too, denoting the exact distance from the camera.

預覽

On rendering animations (or single frames via command line), Blender saves the same image also as a JPEG, for quick preview or download.

Radiance HDR

Radiance is a suite of tools for lighting simulation. Since Radiance had the first (and for a long time the only) HDR image format, this format is supported by many other software packages.

Radiance .hdr files store colors still in 8 bits per component, but with an additional (shared) 8-bit exponent value, making it 32 bits per pixel.