View Full Version : Lightwave and motion vectors

01-27-2012, 09:30 PM
Has anyone figured out have to get a motion vector pass for motion blur? If not, how do you use the X/Y motion passes? Thanks

01-28-2012, 12:19 AM
We've been using motion vectors for years now. I think we started shortly after our first Call of Duty commercial, which was four or five years ago. I don't think we've had to render 'real' motion blur since, and thank goodness! Now that everything has to be rendered HD, we don't have time for that 'nonsense' anymore. :)

There are a few different plug-ins that let you access the buffers and save them as embedded channels in .exr or as separate images.

I use exrTrader (commercial version) in LW which lets you do either. I prefer to embed motion vectors inside my rendered image, which is automatically recognized as vectors in Fusion. Then, in Fusion, I simply add a Vector Motion Blur node and tweak. FYI, I also use Lightwave's depth and Object/Material ID channels in Fusion. Using these channels isn't perfect because LW doesn't output a coverage channel, but I can still use them pretty effectively in Fusion--just feed the channel into a mask node, make your M/O ID selection and add a tiny touch of blur to the result. Usually (but not always,) this trick works quite well.

At work, we have several custom LW and Fusion tools with some very interesting controls, like being able to expand and constrain regions of the vectors to apply motion blur to fur and hair. (It's a cheat but the effect is surprisingly convincing. Now that FFX exports vectors, it may not be necessary though. Need to test this before we do our next 'hair/fur' job.) It also has a few methods visualizing and for cleaning up errant vectors. The most recent feature added to our LW EXR saver is the ability to embed camera position and settings in meta data and world coordinates. In Fusion, we can use this data to literally 'pin' elements to a pixel in 3D space, which sometimes turns out to be far more accurate than tracking. But I digress...

Anyway, LW does have a couple of native buffer exporters which can save motion vectors as images, and you can use this data in many compositing programs.

On the compositing side, some users prefer to use RE:Vision FX's ReelSmart motion blur (http://revisionfx.com/products/rsmb/) plug-in. Many years ago I used to use the RSMB After Effects plug-in in Fusion (before Eyeon added a native tool for this) and it worked great.

Good luck! If you need more info, just do a search in the forums. I believe this topic has been discussed in detail a few times in the past year or two.


01-28-2012, 01:15 AM
Thanks for the info! So does exrtrader give you access to buffers (like motion vector) that the new buffer exporter doesn't? I just upgraded to AE 5.5 to get the extended exr support so maybe exr trader should be my next purchase.

01-28-2012, 04:35 AM
also see Gerardo's tip on RSMB,...


01-28-2012, 06:41 AM
Thanks for the info! So does exrtrader give you access to buffers (like motion vector) that the new buffer exporter doesn't? I just upgraded to AE 5.5 to get the extended exr support so maybe exr trader should be my next purchase.

No buffers are buffers so you'll get the same one's in exrtrader. you can find a script for fusion here.... http://www.db-w.com/products if you want to merge the xy vectors to rsmb vectors

01-28-2012, 10:23 AM
Thanks for the info! So does exrtrader give you access to buffers (like motion vector) that the new buffer exporter doesn't? I just upgraded to AE 5.5 to get the extended exr support so maybe exr trader should be my next purchase.
Sorry, I haven't used the new exporter yet but I imagine it would be similar. I use exrTrader at home because it has worked well for me for several years and we use a proprietary saver at work; I haven't compared it with the current native exporter though.

With some apps you can use embedded channels in an .exr file. When saving with exrTrader or the saver I use at work, Fusion for example just picks up on the data automatically.

FYI, some programs can save .exr with the embedded vectors natively, but you may need to flip a channel either in export or import. With Vue, for example, I need to flip Y channel in the Vector Blur tool to make it work properly in Fusion. This may vary depending on which 3D and compositing programs you use.

As 3DGFXStudios notes, some vector blur tools, like RSMB, expect a single merged vector image. The script mentioned will conveniently do this at render in LW. That's fine but if you need to manipulate the data for other uses (as we sometimes do at work,) it's better to have the channels available separately. If your compositing program reads .exr, it *should* recognize the channels and combine them for you internally. I don't know if AE does this but you should be able to merge the data in any compositing program before applying a vector blur tool.

(Sorry, I work entirely between LW and Fusion these days so I've gotten rusty regarding other compositing programs. Other users here will be more helpful about using vectors in AE.)


01-28-2012, 10:41 AM
You can't use the separate xy vectors in AE. Not that I know of. The XY vectors are better because the are floating point (if you save them in float of course ;)) RSMB only accepts 8 an 16 bit images.

Andy Meyer
01-29-2012, 03:43 AM
i started to use motion and depth buffers some time ago.
but i need to render the scene twice, with and without transparent objects for vector mb and for depth.

in some cases i can use RSMB without vectors, so even transparent objects and the stuff behind the transparent objects are blured very well by using pixel tracing method...
but i dont know a solution to add depth blur on scenes with transparent objects with just one render.

what is your way to render scenes with transparency for post depth and motion blur?

01-30-2012, 11:28 AM
what is your way to render scenes with transparency for post depth and motion blur?

Andy Meyer
02-01-2012, 10:08 PM

02-02-2012, 01:24 AM
I'm not sure. This really hasn't been an issue for us because we always break scenes out into multiple layers and each layer gets its own vector blur processing. I imagine if we ever ran into a problem, we'd just work around it by separating the glass elements.


11-13-2014, 11:27 AM
greenlaw - just for checking - you guys are still using Fusion there?

Also is there a step by step tutorial for vector blur - between Lightwave - Fusion (i am so interested)

and - what is the best way imitating same effect with After effects?

Can you - and other fusion users - review this new version of Fusion in terms of LW friendly point of view.. Thanks a bunch .. and it is so good to have - pro- users in this forum..

11-13-2014, 06:43 PM
For Fusion it's pretty straightforward if you're using exrTrader. The only thing to watch out for is that you give each embedded channel the name Fusion recognizes for that specific channel. For example, for motion vectors you need to call the 'X' channel 'velX' and the 'Y' channel 'velY'. If you use the default names, 'X' and 'Y', Fusion will not load the channels automatically and you will need to manually enable them. (Which can get tiresome quickly.) Once you set this up, make sure to save your settings as a preset.

The list of 'Fusion friendly' channel names can be found in the Fusion docs and I think also in the exrTrader docs.

To apply the vector data, all you need to do is add a Vector Blur node after the image. If the channel is enabed in the EXR file, it will be used automoatically. If you want to put the Vector Blur farther downstream, you may need to re-route the Vectors to the node because it's possible that another operation can wipe out your vector channel. To bypass this, just pipe in the original image to the second input on the Vector Blur. The replaces the 'corrupted' vector channel with the original channel data.

Many of Lightwave's channels are useful in Fusion. I mostly use the motion vectors for post motion blur effects but I sometimes like to use Normals to tweak or alter the lighting. Sometimes I'll use Object ID, Material ID and Depth in Fusion, but because Lightwave doesn't offer Coverage, it's use is a bit limited. (Unlike Vue for example. Because you can get Coverage from a Vue render, any O/M selections made with the Fusion eyedropper tools get automatically anti-aliased.)

I'm not sure how to do all this in After Effects. Coincidentally, I started looking this up just this week. Some of the info I'm finding seems dated but even the more recent info seems more complicated than how it works in Fusion. From what I gather, you need to extract the channels from the .exr, which kinda defeats the purpose of embedded the data in the first place.

I don't know if that's the only way to work with the auxiliary data in AE though. Admittedly, I'm fairly new to using AE, so if anybody knows of some good tutorials to do the equivalents of the above in After Effects, please share the links. Thanks! :)


11-13-2014, 06:47 PM
BTW, I haven't used the new Fusion. I still use 6.4 because I had let our subscriptions lapse over a year ago.

I'm curious to learn what it would take to upgrade to the new 'Black Magic' release of Fusion 7 since they seem to have dropped the whole subscription thing.


11-15-2014, 05:06 PM
thanks for awesome info... one last thing do you use Octane? and how about its EXR exporter for fusion?