=Lord of the Rings Junkie;1096080@PapaSmurf: Ok, but with the same reasoning - why have a subsurf modifier when you can just tweak vertices manually? I believe someone else here said something along the lines of “Take jahka’s new particle system. I’m sure you can get the exact same results with the old system, but not without 10x more work and effort.”
I agree with what you’re saying in principle about keeping nodes simple, I just think this particular functionality is common enough to merit its own node.
To make it useful better add functionality to give it a reason to exist, like let the user affect the wiggle using a texture and such.
I don’t like the attitude saying screw state of the art systems if we can do it the old way. If PapeSmurf wanted he could just make his art by coding it in assembly.
I cannot find a way to add randomness to numbers automatically. But could it be possible with some math node magic? If not, why isn’t there some sort of python node available? That would probably solve it, right?
I think of something similar to the single line python expression that’s available for the driver/driven relationship in the IPO window. It could be as an option in the math node?!?
And it wouldn’t only be for camera shakes
I’m sorry if I am hard to understand, I’m tired and probably shouldn’t have made this post >_<
I must agree with t3d. Or why not expand it further and have nodes that make it possible to load whole python-files to use in the composition. Or maybe some logic-gate-nodes like those in Synthmaker, I almost managed to make a pong-game in that program without using a single line of code. (:
Proper controls for the time-node would also be useful. Or why not have the option of loading a IPO-curve? That would be immensely useful.
I guess a random-node would also be useful in other things than just a wiggle-effect
I can’t attach images, but you just click around in two time nodes. subtract half from each, multiply by some number like 20, and feed one to X and the other to Y of a translate node, which will shift the image 10 pixels each way. for the image, scale it up absolutely (or relatively) to cover the edge.
I’ve used Fusion’s Camera Shake node and it does make things much easier than animating translation settings.
Mainly because it’s not just random noise, but it’s filtered, and uses other things too like randomised sine waves. You can also affect rotation as well, but the main advantage is being able to tweak (and animate) all those settings interactively, from the same place in the UI while the comp is playing back. It’s a plus in terms of workflow, and capabilities, compared to your average DIY approach.
Having said that, it is (or will be soon) possible to animate node settings in 2.5, and you can use fcurve modifiers like noise on the animation curves, however it’s still not nearly as intuitive or convenient to use as a purpose built node.
On the other hand, if it were possible for one clever soul to be able to roll the manual DIY approach up into an easily distributable node group, packaged with thoughtfully exposed parameters, that was still fast and efficient, that would be the best of both worlds for these sorts of things
What if you had an object set as a softbody and just parent the camera to the softbody? With the right settings I’m sure you could get a nice wiggle. That seems like a more practical approach, at the moment. A random node would be extremely useful as well. It would be nicer if you could just add some sine wave noise to the camera’s IPO curve. Maybe just a python script, right?
since the scale node wants either pixels that are in addition to original image size, or percentage overall, but the translate node wants pixels for x&y, a clever node group I have doesn’t work in the general case with images of various sizes. For example, a 10% scale of a 640 image gives me 64 pixels, but there is no way for me to get that 64 into the translate, since we don’t have an “image info” node that puts out the resolution of the image…so I guess i’m not clever enough…
What it merits is a “noise” node. Then you can use that in conjunction with time, sine, translate, and rotation nodes to create a camera shake effect.
However, I’m not entirely against more macro-sized nodes as long as they are fairly widely useful.
Regarding your comparison to the subsurf modifier: the difference in difficulty/tedium between an automatic system for subsurfing and having to do it (accurately) by hand is enormous. If one were to put numbers to it, using the modifer is a difficulty of 1, and doing it accurately by hand is a difficulty of 10,000.
Creating your own camera-shake node network, on the other hand, is only somewhat more difficult/tedious than using a pre-existing one. Pre-existing: 5 (for tweaking settings). Create your own network: 20. And once you’ve created the network, you can use it over and over again thanks to node groups. So for subsequent uses the difficulty is negligable.
we need a random option in the Math node (min value, max value)
but just to translate the render image randomly would look weird, we need too hook up the radom number with a vector of some sort. so we have a direction. that keeps changing … this is hard stuf to do
Im in ae alot at my work. and after effects
repeat edge pixels === fit (in blender)
it basiclly does the same, add a distort --> Lens distortion node, and use fit. or hoover over fit to get the decsription.
+1 from me.
A noise node with some decent controls would be very useful for several applications.
(says the guy who has no idea how to code such a beast)
