# What file format should I use for rendering?

I am new to Blender and want to render an image of my object. I see that there are a lot lot lot of file formats to chose from. Which one should I pick?

• Are you rendering an animation or a still image? – Fweeb Oct 22 '13 at 20:00
• Just one picture – Yppie Oct 22 '13 at 20:40

# Animation

It's generally good practice to render to an image sequence first, then combine all the images into a video format after the render is complete. This allows for the render to stop (whether the result of a crash or just canceling the render to use the computer for something else) and still be able to resume by setting the Start frame to the frame after the last rendered and restarting the render.

See my answer here for details.

# Stills

As a personal preference I use PNG as a kind of default, but each format has it's advantages. The format you use will also depend on what kind of data you need to keep and how much disk space you can spare.

See the wiki for descriptions and a list of supported image formats:

• BMP Bit-Mapped Paint loss-less format used by early paint programs.
• Iris The standard Silicon Graphics Inc (SGI) format used on some Unix OS machines.
• PNG Portable Network Graphics, a standard meant to replace old GIF inasmuch as it is loss-less, but supports full true color images. Supports Alpha channel.

Enable the RGBA button to save the Alpha channel.

The RGBA option is in Properties > Render settings > Output:

• Jpeg Joint Picture Expert Group (name of the consortium which defined it), an open format that supports very good compression with
little loss of quality. Only saves RGB values. Re-saving images
results in more and more compression and loss of quality.
• Jpeg 2000 Uses the Jpeg 2000 codec.

• TARGA and Targa raw Truevision Advanced Raster Graphics Adapter is a simple raster graphics format
established in 1984 and used by the original IBM PCs. Supports Alpha
Channel.

Enable the RGBA button to save the Alpha channel.

• Cineon format produced by a Kodak Cineon camera and used in high-end graphics software and more directed toward digital film.

• DPX Digital Moving-Picture eXchange format; an open professional format (close to Cineon) that also contains metainformation about the picture; 16-bit uncompressed bitmap (huge file size). Used in preservation.
• MultiLayer an OpenEXR format that supports storing multiple layers of images together in one file. Each layer stores a render pass, such as shadow, specularity, color, etc. You can specify the encoding used to save the MultiLayer file using the codec selector (ZIP (loss-less) is shown and used by default).
• OpenEXR an open and non-proprietary extended and highly dynamic range (HDR) image format, saving both Alpha and Z-depth buffer information. Enable the Half button to use the 16-bit format; otherwise 32-bit floating point precision color depth will be used. Enable the Zbuf button to save the Z-buffer (distance from camera) info. Choose a compression/decompression CODEC (ZIP by default) to save disk space. Enable the RGBA button to save the Alpha channel. Because OpenEXR is so new and previews are generally not supported by Operating Systems, enable Preview to save a JPG image
along with the EXR image so you can quickly and easily see what the
basic image looks like.

• Radiance HDR a High Dynamic Range image format that can store images in floating point (with light brighter than 1.0) - 32bits per channel.

• TIFF Often used for teletype and facsimile (FAX) images.

Some further notes on a few formats:

• PNG: PNG (Portable Network Graphics) is a lossless format, meaning that the data can be reconstructed exactly as it was before being compressed. PNG supports true color color depth. Blender also supports 16 bit PNG, which gives a larger range of possible colors.

Because PNG is lossless, the Compression slider in Render settings > Output only affects the file size (AFAIK, the only disadvantage to setting it to 100 is it might take more CPU time to compress/decompress)

As an example, I saved two pngs, one with 100% compression and one with 0% (Uncompressed). The uncompressed png is 2MB, and the fully compressed png is 18.7KB. The images are identical.

• JPEG: JPEG is a lossy format, which means that data can not be reconstructed exactly as it was before being compressed. (see a comparison to PNG here) However, as a result JPEG takes up a lot less file size.

Instead of a Compression slider there is a Quality slider when JPEG is selected as the output format. This refers to how closely the decompressed image resembles the original uncompressed image.

Here is an example. The left most image is a fully compressed PNG (36.3KB), the middle image is 100% quality JPEG (20.3KB), and the right most image is a 0% quality JPEG (8.5KB). At 1:1 they seem pretty much the same, but when scaled up the differences are more noticeable:

• OpenEXR: EXR Is a format for HDR (High Dynamic Range) floating point images, and is capable of storing deep color depth of 32 bits per channel. Blender supports Half precision (16 bits per channel) and full precision color depth EXR images (32 bits per channel). As a result, these files can get quite large, especially when using Multilayer EXR). For example, a 201x167 image saved with lossless ZIP compression is 199.1KB

These images are good for keeping every bit of data around (including render passes in the case of Multilayer), which can be useful for compositing.

Note that when rendering a still, the image is stored uncompressed in memory, so you can render an image then save it multiple times in different formats (with different compression ratios).

• Now this is where all of this info pays off. – Yppie Oct 22 '13 at 20:23
• You should include how to the compression works for jpeg and quality for png – user320 Oct 22 '13 at 20:24
• also some info on 8, 16, 32 bit would be nice. – user320 Oct 22 '13 at 20:27
• 8 bit gives you 256 shades for every RGB value, human eye cannot see difference between 128, 128.128 RGB and 129.129.129 RGB values, so it is fine when picture needs no editing, 16 bit has 65536 shades, so for example you can change contrast dramatically, and human eye still does not see any difference. 32 bit has 4294967296 shades for every color, which in my experience gives me full freedom of change – dimus Oct 22 '13 at 23:51

In a nutshell this is what I usually use --

PNG 8 bit is fine to save file without any changes

PNG 16bit is fine for some editing in Photoshop/Gimp

Heavy color editing and compositing -- OpenEXR -- has a huge dynamic range

Heavy compositing, animation/movie compositing -- OpenEXR multilayer

OpenEXR multilayer is amazingly powerful format. It allows to keep multiple frames for animation or movie, all/any Cycles rendering passes, and practically unlimited dynamic color range. All of it comes at a cost of huge file sizes. This format is completely supported by Blender compositor, Photoshop does not support it completely at the moment (there is a plugin)

To appreciate what is possible with Multilayer OpenEXR see http://cgcookie.com/blender/2012/12/10/compositing-cycles-render-passes-blender/

• Thanks dimus! Any difference between 8 and 16 bits in just casual viewing? Human eye? – NiCk Newman Oct 11 '16 at 1:19
• @NiCkNewman --- for human eye there is no difference between 8 and 16 bit. You can try an experiment of making any two adjacent grey values for example 56/256 and 57/256 --- you will find that your eye cannot distinguish between them. So 8bit is enough dynamic resolution for viewing. – dimus Oct 12 '16 at 14:25
• Thanks Dimus, saved me a lot of bytes from my renders! Thought I had to do 16 :) – NiCk Newman Oct 12 '16 at 23:08

Blender's manual has a table of formats supported features, as well as a hint on which formats to use - if you're not concerned with details of each format.

https://docs.blender.org/manual/en/2.83/files/media/image_formats.html

• link returns 404 :( – rob Oct 5 '17 at 7:44
• corrected link. – ideasman42 Oct 5 '17 at 10:26