Motion Blur with Blender Internal

Hello again, im back with another question :slight_smile:

Im working on some music animations with blender currently and I came over a problem that I dont think has a solution, but I still wanted to ask.
What I want to do is to add a blur to a noise effect I use on a Image because the noise alone looks pretty boring.

I want to use motion blur as a effect, this kind of blur that many Music Animations have made with AE but my problem is that im still learning Cycles so im limited on Blender Internal for a longer time. Rendering takes huge huge time and everytime having to render a bit of the video to see if the motion blur works and looks well is too much.

I want a motion blur that i can set and eventuelly see directly if thats possible or at least have a way to set it properly without using extrem amounts of time for each adjustment. If possible I would love to set that the motion blur also only applies to a specific object or emitter/object that copies emitter (for example of a empty with noise effect)

If theres a way to do that or some infos of the best way coming close to my wanting that would be great! :smiley:

I attach a blend file with one of my works that include a shake built with noise so you know what I meant by that :slight_smile:
https://drive.google.com/file/d/0BzsPRM5i7rKnSlBvXzhMNHhYaEk/view?usp=sharing

Im thankful for all kinds of help!

Moved from “General Forums > Blender and CG Discussions” to “Support > Compositing and Post Processing”

Motion blur in BI is achieved by Blender rendering successive frames and mixing them to one. Cycles oversamples vector motion to estimate travel, while vector blur in the compositor uses vectors to do a directional blur. Cycles is best but you have to perform a render to see it. Try lowering the samples and the render dimensions via the % scale.

Since I’m too cheap to spend lots of money on hardware … :spin: … and since I do use BI nearly all of the time … I quite-frankly gave-up on Blender’s “motion blur” a long time ago. I simply could not stomach watching Blender grind through an animation frame-by-frame, calculating five different renders and then combining them, only to proceed to the next frame and therefore re(!)-calculate four of the five frames that it had calculated moments before. I simply did not have time for that. (Literally …)

Well, what I quickly found out is that “a little blur” is really all that you need, most of the time. I apply all such blurs during the compositing step. Quite often, “stupid ignorant ‘just-blur-the-damned-thing’ blurring” :wink: turns out to be satisfactory. Failing that, a vector blur should work quite nicely.

Your workflow needs to consist, in all stages except possibly the last, of “MultiLayer OpenEXR” outputs. (The final-cut files can omit the “layers.”) This gives you the data that you need to do these kinds of things in post-production.

thank you for that idea first! im quite a bit confused about this. is there maybe a good tutorial or can you explain to me what these multilayer open exr outputs are and how i do use them?

“MultiLayer OpenEXR” is one of the output-file types available in Blender. This format is a (Blender Foundation-originated, actually …) extension to a file type introduced by Industrial Light & Magic (yeah, them …) for use as intermediate-files.

OpenEXR stores data as high-resolution floating point numbers, and is “loss-less.” It stores one frame per file, so you put all related files in one directory. Yeah, the files are big. But, they are accurate.

MultiLayer OpenEXR extends this idea to allow multiple named “layers” of information to be stored in a single file. The idea was a good one, and it was subsequently embraced by the industry.

What this format buys you is that all of the various “RenderLayers” that can be produced by a Blender render will be captured in the file (set…), exactly as they were produced: the actual numbers.

When, in subsequent “down-stream” blend files, you reference the directory, all of the RenderLayers will individually be there, and they will be numerically identical to what the “up-stream” blend-file produced.

I use this file-format for all intermediate (“process”) files, up to the “final master,” which I produce as non-layered OpenEXR (because I still need OpenEXR, but don’t need the layers anymore … “'cuz I’m done”). Thus, throughout the entire process, I have never “thrown away information.” I’ve never stopped using floating-point. The color-space is strictly linear.

Additional blend-files to produce “deliverables” (.MOV files and what-have-you) are produced from this “final master,” this being the only time that any compression or encoding actually occurs. All deliverables needed are individually produced from the same master, using separate blend-files.

(If I find that some color-correction is needed and that I of course want it to be consistent across all deliverables, I will create a new blend-file which processes my “final master” to create a new set of OpenEXR files which I call, “adjusted final master.” The deliverables are then individually produced from this source.)

Yeah, “gobs of disk space,” but who cares: [external …] disk drives are cheap.

I informed myself about this a bit and thanks to the info I have an Idea of how this thing works. Now is the question. How do I use the OpenEXR files to create the blur in the best and fastest way?

You don’t use the file-format to “create the blur.” You use it as part of a multi-step rendering process to capture the intermediate data exactly as it was produced by up-stream stages. This can’t be achieved by rendering the intermediates to “image” files.

sorry for the long waiting time, I was on vacation. so I basically take the exr files and use them, i wil see if i find a tutorial for that somehow