Andrew Somers
Forum Replies Created
-
If it is being used for render only, you don’t even need that – just install the render engine on that machine – it is a free install included with every license, and you have have as many as you like.
-
Yes, anytime you can use a practical effect to help sell something do! This is especially true when you want something in the scene to light up its immediate surroundings.
Doing this effectively in post requires a bit of work “such as creating 3D light catchers).
-
I usually create things of my own when I can, rather than use stock plug ins.
Work in 32 bit mode, linear color space – play with masked solids that have the blur and the exposure plug in on them, and over-expose, blur it out, use ADD transfer mode, etc etc.
Experiment!!
If you have a marble with an LED point of light, that will give you a great tracking point.
A 3D track, with object tracking, could also be very useful, as then you can add in “shadow catchers” where certain key objects, and in particular the ground/walls are, and these can be affected by a point light source that you have connected to the object motion track of the ball ,and things that interact with the ball like hid hand, etc.
TIP: make shadow catchers for shadows “neutral” and light catchers for highlights either neutral or close to the color of the object they represent. Shadow catcher “shadows” are added to the comp using the MULTIPLY mode., and light catcher highlights are added to the comp using the ADD mode.
(I find its easier to control separate shadow catchers for shadows vs light, instead of trying to use the same catcher for both).
-
Govind said: “I’d like to add a rolling shutter / jello effect too. Odd request I know,”
It’s not an odd request, I’m doing something similar for a feature, where the director wants the camera to look “very low end” and have a jello effect added to it.
Time displacement won’t give a real rolling shutter effect, as it will just chop the image up into time slices, and be very jagged.
What you want is to track the image motion, then extract the delta with an expression, and use that to drive a transform plug-in.
I am planning on a tutorial on this subject in a couple months when I’m done with this film.
-
Looking at the video example you posted in HD, it looks like a simple large clear marble was used, and the lighting effect was added afterwords. I don’t see any indication that there was any internal light or LED, etc.
The reason I mentioned using a practical (LED or other light source) as something you might want to try is that it will help integrate the effect into the shot.
-
Why a green ball” If you were thinking “green screen ball” that may or may not be a good idea.
COnsider instead how you want the obs to look. Solid glow? Or transparent? If held in the hand do you want to be able to see thorugh the balls?
Then *DON’T” use solid green balls.
For the record (and contrary to common misconception), there is no “magic button” you can click to make green objects disappear.
If you want the balls to end up with a solid glow, you may want the balls to be absolute black (not just painted flat black, but flocked black so that are as non-reflective as possible). With flocked black, you should be able to make a “MULT matte” and then add glow effects to it in ADD transfer mode while working in 32 bit
If you want the balls transparent, then I’d suggest using a highly transparent ball (hollow plastic sphere may work best).
You may want transparent or translucent balls to be illuminated on set, in shot – LEDs inside as a seed, to add real lighting, then augment with additive glow effects in post.
Whenever possible, get the *in shot* look as close to what the final should be, then augment from there.
-
There are a few ways to handle this.
Since coins are flat discs, one way might be to use Mocha to track each coin with a mask, and then use those masks to select the coins on a dupe layer of the footage that had blur and exposure (boost) on it as effects. Adjust the feather of the masks for the amount of “glow”, as an example.
-
The first question to ask yourself is “why?”.
Network rendering for After Effects may not be useful to you, or it may be mandatory. But having it just to have it is not productive. (Sure, it may be fun to brag about the size of your renderfarm at the local users group meeting… but do you really need it?)
Consider:
1) Today, the biggest bottleneck in most renders is disk speed. If you don’t have super fast disks, and a super fast datapath, most renders will not benefit measurably from distributed (network) rendering.2) If you *already* have a renderfarm for other CPU intensive tasks (like 3D raytracing), then adding functionality for AE on that farm may be useful.
3) A real renderfarm for AE of several quad-core machines will need, at the very least, fibre-channel cards in each machine, a server, and a SAN RAID. Ethernet – even gigabit ethernet – is too slow for file serving media to multiple machines.
4) Weight the costs – the cost of setting up a bunch of quad core machines, with fibre channel and a SAN etc are quite high – while you can buy a stand alone 12 core machine and fill it with 48 GB of RAM and eSATA drives in a cheap raid that will give you similar performance at an overall lower cost.
5) The tutorial video that Walter links to above does show a “workable” solution that may be of benefit in some cases.
Here are three examples of when and when not to use a renderfarm:
Example one: Lone wolf, no farm.
If you are an independent artist working by yourself, you have little or no need for a farm. The costs to set up, and the time spend to manage it, will not pay off in terms of speed. Get the aforementioned 12 core system full of RAM and inexpensive eSATA raids instead – that’s a far better investment.
You can find a number of scripts that will allow for background rendering when you work in a foreground instance of AE.
https://aescripts.com/bg-renderer/
Now, if BG rendering is too much of a performance hit for you, then as an add on, add an 8 or 12 core machine (just one) that you can copy projects to for background render and leave it at that.
Example two: Wolf pack, with farm.
If instead you have a *department* of artists all working together on the related projects, then you need fast common storage for everyone – and as long as you have fast common storage, then adding render nodes begins to make real sense.
This will allow artists to quickly “collect files (project only)” for watch folder render, and move on to the next project.
I’d estimate the cost/benefit curve of this method begins to prove itself with about three or four (or more) artists working simultaneously on the same project.
The downside here is that each machine in this setup needs about $2000 in hardware and software for fibre channel and the SAN client, etc. And your SAN server will be too mission critical that it cannot (well, should not) be used for rendering. And the RAID itself won’t be cheap, either.
Example three: Middle ground, a couple stray wolves and a chicken coop.
In the middle ground, with two, perhaps three artists, it’s probably better to have simple NAS storage and a simple ethernet server for the main files, and have each artist copy those files locally for work while working, then copy the files and project to independent standalone machines for background render, using “Collect Files (ALL)”
It is important that artists do NOT try and work by just linking to files over the network, but always copy files locally for work purposes.
For consistency (needed if you have to resurrect an old project), mandate that each working project be in a folder, and in that folder also have a folder called “media” where ALL media for the project be stored. This makes it was to delete the media from the local drive when the project is done, and *also* means that to reconstruct, all you need do is repopulate the “media” folder with the media for that project.
A typical two-artist setup would be to have three machines total (assume these are MacPros for the sake of argument):
1) Machine A: Artist one’s machine – an octocore with 24 GB RAM, boot drive, and a 3 or 4 drive raid 0.
2) Machine B: Artist two’s machine – an octocore with 24 GB RAM, boot drive, and a 3 or 4 drive raid 0.
3) Machine C: The server and BG render machine – a 12 core machine with 48 GB RAM, and a RAID 5 connected via eSATA and a PCIe raid controller.
https://eshop.macsales.com/item/NewerTech/MXPCIE6GRS/
Machine A is directly connected to Machine C’s Ethernet Port 1, and Machine B is directly connected to Machine C’s Ethernet Port 2.
Machine C acts both as file server and render machine.
The object here is that you spend all your money on a really fast machine with really fast drives rendering in the BG, and really fast machines with really fast drives to do work on in the foreground.
And you spend *no* money on exotic network topologies (that cost $2000+ per machine).
-
As I recall it was added in AE CS3.
Yes, it has more than a few “quirks” especially STROKE – if you want real control, then the traditional methods like solid/mask/precomp etc. may work better, but for the down and dirty stuff (and in particular temp work), it’s nice to have.
Inner/Outer glow I find most useful.
-
Michael: You’re not rendering with OpenGL on, are you?
DING!!
To add: Check your default output settings and create some that you know definitively have OpenGL off for render (for instance, “Best” should have OpenGL off as default).
I’d also like to add that if using OpenGL for fast previews, it is a very good idea to turn it off for PREVIEWS before render, and to check your comp with it OFF, as your “normal render” my render differently than the OpenGL previews would indicate. This is ESPECIALLY true when working in Linear – if you are working in LINEARIZED colorspace, then you must ensure that OpenGL is off for any color related work/previewing (black levels in particular will be totally wrong, at least in CS5)