Rendered Image Formats

I see a lot of questions on output formats for renderings and which one to use. The output format has a great impact on the quality of the image, not only in the resolution of detail but the amount of color and also the size of the output file. Before we can talk about the quality of the format, we need to first understand how images are stored by the computer.

Color Depth

A standard computer image is stored in 24 bits of color consisting of 8 bits of red, green and blue for each pixel and these images are referred to as having an 8 bit color depth. In addition to 8 bits per channel, some formats can also store 8 bits of alpha channel which holds transparency data. Most rendering engines can separate the alpha channel but not all formats can save it and it is essential when working with a compositing work flow where images will be stacked one upon another. An 8 bit image with alpha is what is commonly called a 32 bit image, 24 bits of RGB plus 8 bits of alpha.

Higher format levels include 16 bits per channel and 32 bits per channel, with or without alpha channel. When 3ds Max renders, it actually renders to a greater bit depth of 16 bits of color per channel. This is far beyond the ability of current monitors to display but it is very useful when working in post operations such as compositing in Adobe Photoshop, Autodesk Composite or Autodesk Smoke. The extra bits of color data allow you to adjust things like gamma and gain to improve appearance without re-rendering the image.

Formats

I’ll look at three separate groupings of formats, each with their own pros and cons.

The first groups are the loss based compression formats like .jpg and .gif. These formats are convenient because they are small in size and can be readily opened. The downside is that they lose image resolution during their compression which introduces noise and loss of detail. The use of these formats is really limited to preview images and they should never be used for final rendered output.

The second group are the main stay formats that are well established and include .tif, .tga and .png. The TIFF (.tif) format was developed for the printing industry and this format will allow you to set and embed the dpi(printed dots per inch) of the image. TIFF supports 8, 16 and 32 bits per channel and you can automatically save the alpha channel as a separate file but it suffers from very large file sizes. The Targa format can store in 16, 24 and 32 bit color, not color depth. The 16 bit color is a holdover from the old days when computer monitors could only display thousands of colors, not the millions in a 24 bit color, which is 8 bit color depth. The 32 bit color is 8 bit color depth with alpha. Targa is an older but still used format that has some compression so doesn’t suffer from the extremely large files size of the TIFF while still not losing any resolution as the compression is loss less. The Portable Network Graphics (.png) format can store in 8 bit or 16 bit color with or without the alpha channel embedded in the file. This is one of the most commonly used formats in visualization as it is easily opened, has a loss less compression to keep the file size down and can carry the alpha channel all in one file. While many programs can easily open and view .png files, you need to be careful with the color depth as only graphics specialized programs can properly open a 16 bit color depth image. Opening a 16 bit color depth image in Microsoft Outlook or PowerPoint will result in a washed out image while it will appear perfectly fine in Adobe Photoshop.

The last category of formats are the radiance image formats(.hdr and .exr). Radiance image formats contain 16 bits or more of data per color. Radiance formats are typically large and require specific software to open properly but they allow you to select the range of color brightness data to be viewed and are frequently used in design visualization to actually light a scene. Of the two formats, the OpenEXR is receiving quite a bit of attention as it supports the use of layers, allowing you to save all passes from the Render Elements settings to individual layers within the same file so that they can be used downstream in post processing. These formats are very useful if your pipeline already integrates high bit depth post processing but they are not the ones you’d want to use to send your output to your client.

Video Formats

The first question people ask is what format should I render my video to. The answer is DON’T. It is never a good idea to output renders to a video format and for a number of reasons. Video output is always compressed out of 3ds Max which means you are always losing image quality without any control. Video can only be output from a single machine which means your render farm is useless except for the one node rendering the video stream. And if you crash, well, you lose everything because the video file will not be complete and you won’t be able to open it up. So, always render everything to a still image format. Still images can be compressed into video through RAM Player or Video Post, both are in the Rendering top menu and there are instructions in the Help files. You can also use outside apps such as Adobe After Effects, Adobe Premiere, Apple Final Cut Pro, Autodesk Smoke or a whole host of other commercial, shareware or free applications.

Stephen Gabriel

Senior Application Engineer

MasterGraphics Inc.