spacer spacer spacer

Tutorial: Using Velocity Channels for Compositing Motion Blur

(This page is horribly long, I'm going to split it into multiple pages and fix its terrible layout when I have time, my apologies.)

spacer spacer spacer
spacer spacer spacer
Summary: Using a velocity channel render element to create motion blur in post                      using max/vray's velocity element by running it through the RE: Vision                      FX Reel Smart Motion Blur in Fusion 5.
When you render frames in passes you cannot utilize in-render motion blur because it then eliminates the ability to single out individual pieces using a wirecolor element or mask. When the mask is blurred as much as the piece it is trying to mask the result is very inaccurate.
Rendering all the final frames with no motion blur and then compositing it in post, when assembling the final frames, allows full control of all the elements right up to the final output of the final frames.

For this tutorial I am using:

3D Max 9

Vray Renderer

Eyeon's Fusion 5

(I'll get a shake version posted)

I'll show how to set up a very simple scene multiple different ways for rendering options. I will compare in camera/renderer motion blur and the final output of the motion blurred images from fusion.
I've cropped most of these images so the small text is still readable. Click on any image to see the larger full screen version
spacer spacer spacer
This is just to show my light setup for the scene. I used 2 vray plane lights, one spot light with vray plane shadows, and the circular light over the teapot is a vray light set as a dome, being used as a way to utilize IBL with an HDRI. -This is the same scene I used in the render layers tutorial.
Just a still of the scene I'll use for this tutorial. There will be a red box that will fly through the frame from right to left so i can show a few different ways to render stuff out.

The box flies through with enough speed that the motion blur will be noticeable and will also help keep these video file sizes small.

I am going to render the box and the background separate. It may seem unnecessary for this simple scene but when you start applying this to larger scenes with more pieces it becomes clear as to why you would do this. For example, if I had multiple characters in the scene I would render them separately so I can control the amount of motion blur on each character, then I can even animate the value so if I had some sort of hero shot where I wanted the blur to decrease for a short period of time I could just animate that in post.
To start we need to separate the background objects from the box. We will open the layer manager along the top menu.
Once in the layer manager you will notice everything is on the same layer, the default layer.

Drag over all the objects that will be in the background.

Tip-One of my lights was in the way when I dragged over the objects, hitting shift+l will hide the lights from the view port but still allow them to render.

Once the objects are selected click the "Create New Layer" button on the upper left corner of the layer manager.
If you expand that layer you will notice it has already added all the objects that were selected at the time the layer was created.
Click the dash next to the checkmark to hide all the objects on that layer.
Now when the time slider is scrubbed, only the box is visible in the camera view.
Before we can render out this box we need to add the velocity render element. Just like in the render layers tutorial, in the Vray render dialog go to the tab all the way on the top right. Click add - then scroll to the vrayvelocity element and make sure to save it somewhere you can find it again.
Go ahead and render out the box sequence.
The bar with no motion blur will look like this.
The Velocity element will look similar to this. The velocity element is calculated pixel by pixel and the amount the previous pixel has moved... in its most simple form. The actual calculation is a bit more advanced, but that's an easy description.


spacer spacer spacer
Now that we have the foreground part of our sequence rendered we need to render the background.
In production with multiple characters or objects moving, most likely, the reflections and shadows will be needed, so using a static frame for the background wont work. If I didn't want the reflections and shadows for this piece I could just render one frame and save time.
To set up a render for just the background objects but still have the red box's appropriate reflections and shadowing we'll use a pretty simple setup.
spacer spacer spacer
After Un-hiding the objects for the background I have scrubbed to where the box is in frame. I right clicked on the animated box and will select "object properties."
The resulting box has a list of options that, hopefully, I will have time to cover in another tutorial. For this tutorial we want to deselect "Visible to camera."
Now if you scrub the time slider you will notice the bar still shows in the view port/camera, however, if you render a single frame where the bar should be visible, it is not. The bar can still be seen in the reflections and still casts a shadow, though.
The velocity element is not needed for the background, since it does not move, so you can go ahead and turn off all the elements by deselecting the "elements active" checkbox in the same menu we added elements in before.
Now you can render the background sequence.
Here's the background with no bar.
Now we have to composite the bar on top of the background and then use the velocity element to control how much we want the bar motion blurred.



spacer spacer spacer
I'm using Fusion to composite this for this tutorial.

To bring a series of images into Fusion go to Tools>I/O>Loader. It will bring up the typical windows search window, navigate to the images and select the series (it will appear as one file.)

I have already loaded up everything I will need for the tutorial. On the bottom left (green node) is the velocity series, which is displaying in the upper left view port. The yellow node above that is the colored bar which is displaying on the upper right view port.
To get the nodes to display their images, click on the node once and then hit the number 1 or 2 on the keyboard, depending on which view port you want it to display in.
The white dot on the bottom of the node shows which view port that node is displaying in.
Now to bring in the Reel Smart Motion Blur node. Tools>Re:Vision>ReelSmart Motion Blur 3

With the node in its time to connect the velocity element and the color images.

My mouse is hovering over the color input connector, the yellow triangle.

The yellow triangle: Color Source Input
Purple Triangles: Effect Mask Connector
Green Triangle (on RSMB): Track Source
White Triangle(on RSMB):Motion Blur Mask
Red Box: Output
I will connect the output of the color node to the yellow triangle.
Now connect the velocity element to the green triangle.
Notice the green triangle is now on the bottom of the node. Fusion tries to keep the noodles from intersecting as much as possible and will re-arrange the input positions as needed.
For the next step I have already imported my background set of images.
Select the node for the background images, mine is selected and showing as yellow. Then along the top bar is a button "Mrg" which is short for merge. Or you can get a merge node by going to Tools>3d>merge.
With the background node already selected when you select the merge tool it will automatically connect the two nodes together. As shown below.
Now connect the output of the RSMB node to the green triangle on the merge node. The green triangle on this node is for the foreground input.
Again, fusion keeps my workspace tidy so I don't have to.
Now, when you scrub through the video you will notice a halo effect around the bar. That's caused because the merge node is trying to cut a sort of mask so it can layer it on top of the background. For us, this won't work. The video below demonstrates the issue.
To remedy the issue we will use the purple triangle on the merge node to tell it exactly where to mask. Since the mask needs to be the same as the colored bar we will just double the output from the RSMB node.
Drag from the purple triangle to the output of the RSMB node or the other way around, either way works.
By loading the video I rendered out from Max that utilizes Max's camera motion blur you can see the RSMB motion blur is not as extreme. This is because the default value of RSMB is .5 and the output from max is 1. To correct this just click the RSMB node and look at the right menu where it says "Blur Amount." It's currently at .5.
Bringing it up to 1 ends up being a bit much so I've set mine to .85.
There are many more adjustments that can be made to get the exact same motion blur but those tweaks aren't necessary for such a simple example. Some people end up blurring their velocity node to get a smoother blur, I have had both positive and negative results with that technique.
To render out the final image sequence use a saver tool. Found under Tools>I/O>Saver. Input the name.extension and then select a location to render it out.
In the middle of the bottom of the screen is a green render button, hit that to bring up the render dialogue.
After checking all the settings hit "Start Render."
Fusion is pretty quick at putting out images and will display a "render complete" message with some statistics on it when it's done.
On the top is the sequence output from Max. The bottom is the composited sequence from Fusion with RSMB.
spacer spacer spacer


spacer spacer spacer
That's it! That's how you utilize a velocity channel to create post controlled motion blur. I hope that helps.

If this has helped in any way or if you found a flaw or have a suggestion please let me know.

spacer spacer spacer