The Info Pass provides a view of the effects of the normals, UVs and geometric data provided in the scene. Many options here can help you in the compositing stage. For example, if you do not want to deal with the Render Layer Mask, you can get a render pass by selecting "MatID" or "ObjectID" here. and then extract and use this data in the compositing software. For example, to use MatID, you can extract each ID color using "EFX Render Elements" (which is a free plugin) in After Effects. So you get a result like layer mask.

Now let's explain the options:

Max. Samples:  

Sets the maximum number of samples per pixel only for the info channels. This parameter is independent from the "max samples" in the Kernel Settings.

samplIng mode:

This option has three sub-menus:

Distrubuted Rays:

Option to show distributed ray tracing, i.e. if enabled, motion blur and DOF are sampled. This is enabled by default.

Non-Distrubuted with Pixel Filtering:

Removes any DOF or Motion Blur from the selected info channel using Pixel filtering. For clean output image.

Non-Distrubuted without Pixel Filtering:

Removes any DOF or Motion Blur from the selected info channel without Pixel filtering. Jagged edges may appear in the picture.

Z-depth max:

It only becomes active when "Z-depth" type is selected. Determines the maximum depth of Z-depth info channel type as shown in the picture below.

UV max:

It only becomes active when "UV" type is selected. Sets the maximum value that can be shown for the texture UV coordinates.

max speed:

It only becomes active when "Motion Vector" type is selected. Speed mapped to the maximum intensity in the motion vector channel. A value of 1 means a maximum movement of 1 screen width in the shutter interval.

AO dIstance:

Controls the distance of the ambient occlusion shadowing spread. This setting should be adjusted in order to achieve realistic results depending on the scale of the objects in the scene. For example a small value is more appropriate for small objects such as toys and larger values for an object such as a house.

render out proper z-depth pass

Since one of the most used Pass type is Z-Depth, let's give a brief description of it. What you need to consider when taking a Z-Depth pass is that whether it's Tonemap or Linear. If you are getting a Z-Depth pass, then you'll use it for compositing for DOF effect (or depth compositing). If you choose tonemap, you get the same result as Live Viewer and render according to it. If you choose linear, you will see a much different result than LV, and you may need to extract this information when doing Comp. For example, you can use the "Extractor" effect in After Effects.

The following two diagrams show both the Linear and Tonemapped workflow for Z-Depth Pass output. In fact, these 2 diagrams are valid for all passes, not only for Z-Depth. Please remember them at the end of your work.

Finally, never use 8-bit and get rid of it forever, for both 3D and comp. Do not choose jpg, png or targa formats as a Render output. Such formats can not store extra data (except Alpha). Do not forget that EXR is your best friend both your final render and the compositing. After the comp stage, a final touch, you may use 8-bit format again.

On the other hand, the writer of these lines believes that the effects, such as DOF or Blur, should be done in 3D. Any blur or DOF made in the Comp never reaches a 3D quality. If you do not mind the time, do it in 3D and compare the result. The difference in quality is like the difference between black and white.