Hammers: I’m afraid it’s you who’s mistaken here Blender is functioning properly and the next frame calculation is essential to calculating motion blur.
Motion blur is a photographic effect, and it’s caused by the fact that each frame you see is an individual photograph, exposed over a fixed length of time, governed by the camera’s shutter speed. While the shutter is open, any action that takes place will be exposed on to the film or sensor, so if something is moving left to right, all the infinitesimal steps in between will be captured to form a line.
Since a frame itself is a single entity, its point in time (F) can be considered to be in the middle of the time between when the shutter opens and closes, but what you really see is an average of what happens over that period of time, ranging from before to after.
That’s why Blender needs to know what’s happening in the previous and next frame, so it can accurately see where the pixel is coming from, where it’s going, and then interpolate what’s happening during that time period of a frame.
OK I stand corrected! I can understand what you’re saying here. I guess what throws me is that disparity between the frame I’ve created (thinking of this more as the end point of the exposure I suppose) and the blur effect which takes the frame as the midpoint of the exposure, which to me feels counterintuitive. Is there maybe a case for having an offset setting to compensate?
In RenderMan and i think in most renderers shuter opens in current frame and closes in current frame + exposing time. Most propreitary animation packages makes possible to get scene at sub frames to get nice non linear motion blur. But this method is not good for cmoposiing because of you have only one image with one frame
I’ve been looking at the dof posts and thought id attempt one and try and solve the bleeding problem. So here is my result and my quick simple setup. the final color ramp controlls antialiasing and the following map node finetunes the antialiasing. I’ve also used gaussian blur to blur away, and a normal blur to blur towards.
My only problem as maybe with every other dof node setup is that when the contrast between to Z values is to high the jaggedyness beggins. My setup will handle a relativly high level of contrast but still theres a limit. here is the test for your dof’s. render somthing against the sky and see if you get the jaggedynes. if you dont please tell me how you did it.
Ps: my setup aims to reduce bleeding as much as possible, not to eliminate it totally.
talking of DoF, is there any way of animating the focal length of a camera? I’ve tried looking in the edit buttons panel for the camera (thinking it might be key-able) and have resolved to try the nodes setup options instead
[edit] found out, have just discovered it is keyable…it’s a shame the ipo window properties and the F9 buttons property names don’t match!! [/edit]
use a time and math node (possibly a few other assanine side steps depending on the result you’re after) between the map value and color ramp nodes in the following image. plug the result into a DoF node with “No Z Buffer” enabled:
As long as youre not doing works for profit, you can always look among your games to find good sound effects. Personally I like the sounds from Doom 3, F.E.A.R. and S.T.A.L.K.E.R.
Blender is functioning properly and the next frame calculation is essential to calculating motion blur.
