Motion Vector Maps Question!

Hello, fellow B3D fans!

I had some questions about Understanding Motion Vector Maps.

• How can I best understand the Gradient values for Motion Vector Maps?
Or is that a study in itself to understand Motion Vector Maps?

Here are some examples:
5e4789470d56c297e8a8bf6f99dba45e
images


a30731017bc30be42dc88b8565da517a0314b01b

Thank you all in advance!

RGB is converted to a XYZ matrix resulting in a vector for each pixel and frame. Them those values are used to generate motion blur.

Hello !

It can go a bit technical and on top of that blender is doing it it’s own way, and I’m not even sure there is a standard across application for that.
It’s not rocket science either, but it’s not as simple as say an AO map to understand.

So for each pixel, the color represent the direction where it’s going, Red represent X, Green Y.
And it’s in screenspace, meaning that’s how much a pixel moved in the screen/final image.

So if you sample a pixel that is R = -10, G = 0, that means that the pixel moved 10 pixels to the right and 0 pixels up.

Now you might have noticed that there is also information in the B channel, and also in the A channel.
This is because blender store the frame before, and the frame after.

So if you render the image 10, for each pixel of the image , R/G give you how much the pixel moved from image 10 to 9. and B/A gives you how much the pixel moved from image 10 to 11.
That’s what is happening in the vector pass.

Now in the vector blur node, the Z pass is added and taken into account, some other stuff is happening and the logic is a bit over my head for that one.

Hope that helps !

Thank you so much, @anon39274998 and @sozap!

And clear explanations, Sozap.

Then I’ll continue to research exactly what it is about Z-Passes in the Motion Vector Map!

It’s used in the vector blur node , not in the vector map pass :smiley: sorry I may have made things a bit confusing.

In blender the vector pass needs to be plugged into a vector blur node, alongside the Depth / Z pass.
That’s what I was talking about.

Thank you so much, all of you!

A little off-tipic.
Does anyone know how I can also render this Setup separately GI?
Because I seem to remember that at one time GI could be selected or did that merge into Depth Pass or into a Mist Pass or into a Vector Pass?

Or is that no longer possible to create a separate GI?


_(upload://gxrGG2LCFIr5VZuUK3r14WAAu2G.jpeg)

GI is basically Diffind pass, but it might not be what you’re expecting

So what is/are the differences, as a Pass outcome? :smile: :speak_no_evil: :stuck_out_tongue_winking_eye:

And can it be correct that the Passes in this Setup is only rendered when you render an EXR, PNG or Tiff Sequences, and not with the H:264 MP4?

Because with MP4, I don’t see anything happening in my Passes Folders!

Well, it’s indeed a bad idea to try to export these passes in MP4, EXR is much better !

And to answer your question, ask yourself what a GI pass is supposed to be and to do ?
What is the problem you’re trying to solve here with it ?

In general it’s best to leave it as it is unless you really have a need.
Tweaking the passes can be useful in some edge cases where you want to say , add more reflection ( glossy dir / ind / col ) to your render without re-rendering. Or changing color of objects by tweaking the passes in comp.

It’s quite powerful, but first it’s quite heavy, and secondly it’s very easy to screw things up, by rendering the passes to mp4 for instance :smiley:

My advice is to tweak the full render and avoid exporting passes unless it’s really the solution to the problem you’re trying to solve. And if something looks wrong, best is to fix it in 3D rather than in comp.

GI Pass should show all bounces.

Super good tips, thanks!

I want to start using those passes for Breakdowns for in my Showreel and Portfolio.
So that everyone can see the Passes and the Polished Renders, side by side.

And I have now rendered it in PNG even though I know exactly what is the limit of PNGs versus EXRs.

Disadvantage of EXR is, you can’t open it everywhere. Like in Premiere, After Effects, DaVinci Resolve and Houdini, or is it possible?

And is it recommended to choose OpenEXR in Blender or just EXR?

I get it! :smile:

It’s possible:

In Blender you have 2 options to choose from. OpenEXR and OpenEXR multilayer. OpenEXR is just your render in a much higher bit depth and without artifacts like you can see on JPG files. If you want to export the passes you need to choose the OpenEXR multilayer option.

You can see a good introduction to them here:

Be aware of the size of those files. They are HUGE, compared to MP4 files. So, don’t expect the same. In the last link I sent you, they talk about compression as well.

Thank a lot for the article about EXR, I’ll give DWAA a shot !

Why was this marked as solved by me?

That went by accident, sorry.