i find that using vector blur first of defocus node it appers a resultif i invert the order of precess, first i defocus and after i apply a vector blur the result is different: what is the best way?
There is no sure method in your case. Since both operations distort imagespace in own different way. Show your scene - it will be easier to find a cheat way to achieve desired result.
But I’ve once heard a funny way to get fast motion blur and defocus in raytraced render (it was old Vray if I’m not wrong).
- Everything was rendered WITHOUT bluring in vray into 32-bit EXR.
- Then CAMERA MAPPED back onto original geometry and rendered again with shadeless shader with something rendermanish. Both types of blur used on rendertime.
Same concept can be transformed and used in BI\Blender too I guess.
I suggest that you choose one or the other, but not both. Preferr the cheapest and fastest way to obtain a visually acceptable result … given that the viewer will not be studying it too closely.
As Michael Douglas’ character said in A Chorus Line: “Don’t draw my eye!”
The human eye sees things “by exception.” Aeons of history taught us that “the tiger in the jungle, who might be about to eat you, is different.” We might not consciously know what drew our eye, but, b-z-zzzzing!, it just did. If that happens, the shot is dead. If you don’t let that happen, the viewer will see either what he expects to be there, or what he recently saw there (skipping over what is known or expected, looking for that other tiger in the bushes).
I’ve also heard people refer to the “Thomas Kinkade Effect” (and to others who point out that about 80% of American households have something inside their walls that was made by that company). A slightly hazy, slightly glowing overprint or underprint, used sparingly or not-so-sparingly within a frame, can provide a very cheap but also very effective suggestion of the desired visual effect, at a fraction of the computational cost. For example, take a rotating wheel without motion or vector blur and lay it over a moderately blurred and considerably desaturated copy of itself … in a two-dimensional compositing operation which uses as its only input the rendered sharp wheel. Try it, perhaps in a situation where that wheel isn’t the front-and-center “subject of” the shot. The slight fuzz is unnoticeable but it acts as a sort of anti-aliasing, and you can tweak it cheaply until it looks “good enough” … having rendered it once.
Is it “real?” No. But, neither were the Western towns in however-many successful movies. Is it “cheating?” You bet! Does it work? That’s up to you.