Tweet
Hey Guys,
I thought I'd post this really quick tutorial about passing velocity data from vray to nuke. This setup is done in maya but also works for 3ds max.
Often there are questions on the web on how to properly output vectors for 2d motion blur and often people have strange results. I will try to cover the setup as well as explain why some of the problems occur and what to do to avoid them.
I have created a scene where two cubes are moving, one in Positive X and one in Positive Y. The reason I did it this way is that I know what to expect in motion vector result, one should have more or less pure X (red) color and one should have Y (green).
At first I have just simply added the velocity render element to my scene and rendered it.
If you measure the value of the Y vector, you can see 1.0 which means the cube is indeed moving properly on Y.
For reference I have rendered an image with true 3d motion blur out of vray. The settings for proper motion blur should be duration of 0.5 and interval center of 0.0
Lets bring this into nuke and setup the vector.
What we need to do is copy the R and G from velocity pass to forward u and forward v of the rgb beauty using copy node.
For motion vector, set the uv channels as forward, multiply to 0.5, method to forward and check use rgba.alpha
The alpha is necessary for the proper edging of the motion blur.
After this setup however there is no motion blur. This is problem number one. Lets go back to maya for the explanation.
Since the default value in velocity is set to clamp, any information above 1 will be clipped. So for that one must use a max velocity value to offset the multiplier inside nuke to get accurate motion blur length. We wont do that. Instead we will uncheck clamp velocity and render again. You can see now that the vector Y has 28.526
When we bring the new motion vector to nuke there is another problem. Since vray computes vectors against gray background that color contributes to the motion vector offset. For this to work properly, we need to apply a grade node and set the black point to be 0.5, note that we also must remove "black clamp" since motion vector contains both positive and negative data.
If we measure the value of the Y and compare it to the value from vray in maya the Y vector is the same, meaning we do not need to multiply the amount in the vector blur, we set the vector blur to 0.5 which is 180 deg of the shatter, and let the 3d vector provide us proper data.
You can see the result now in comparison to 3d motion blur is nearly the same.
I thought I'd post this really quick tutorial about passing velocity data from vray to nuke. This setup is done in maya but also works for 3ds max.
Often there are questions on the web on how to properly output vectors for 2d motion blur and often people have strange results. I will try to cover the setup as well as explain why some of the problems occur and what to do to avoid them.
I have created a scene where two cubes are moving, one in Positive X and one in Positive Y. The reason I did it this way is that I know what to expect in motion vector result, one should have more or less pure X (red) color and one should have Y (green).
At first I have just simply added the velocity render element to my scene and rendered it.
If you measure the value of the Y vector, you can see 1.0 which means the cube is indeed moving properly on Y.
For reference I have rendered an image with true 3d motion blur out of vray. The settings for proper motion blur should be duration of 0.5 and interval center of 0.0
Lets bring this into nuke and setup the vector.
What we need to do is copy the R and G from velocity pass to forward u and forward v of the rgb beauty using copy node.
For motion vector, set the uv channels as forward, multiply to 0.5, method to forward and check use rgba.alpha
The alpha is necessary for the proper edging of the motion blur.
After this setup however there is no motion blur. This is problem number one. Lets go back to maya for the explanation.
Since the default value in velocity is set to clamp, any information above 1 will be clipped. So for that one must use a max velocity value to offset the multiplier inside nuke to get accurate motion blur length. We wont do that. Instead we will uncheck clamp velocity and render again. You can see now that the vector Y has 28.526
When we bring the new motion vector to nuke there is another problem. Since vray computes vectors against gray background that color contributes to the motion vector offset. For this to work properly, we need to apply a grade node and set the black point to be 0.5, note that we also must remove "black clamp" since motion vector contains both positive and negative data.
If we measure the value of the Y and compare it to the value from vray in maya the Y vector is the same, meaning we do not need to multiply the amount in the vector blur, we set the vector blur to 0.5 which is 180 deg of the shatter, and let the 3d vector provide us proper data.
You can see the result now in comparison to 3d motion blur is nearly the same.
Comment