Manipulating the pupil of an eye →

Dumpduck post on forums.thefoundry.co.uk:

Hi, I'm working with footage of an eye and I'm trying to dilate the pupil ever so slightly. The challenge for me is getting the iris to look as it is contracting accordingly. By rendering out a normal shader from Maya (one channel from top to bottom, one from right to left and one with facing ratio) and using it with an IDistort (same method as this guy fakes refraction in Nuke: vimeo.com, I've managed to push the pixels around the pupil. It works to some extent, but doesn't seem like a good way of achieving this.

Good problem solving.

Nuke User Group Meeting with Nuke 7 →

Via gnomonschool.com:

Join RFX, The Foundry, and Gnomon for the Fall Nuke User Group Meeting and learn about the new features in Nuke 7. This event will include presentations and demos of Nuke 7 from Deke Kincaid of The Foundry, as well as workflow demonstrations from Gnomon’s own Martin Hall. If you’re interested in learning more about the next generation of 2D/3D compositing, this event is not to be missed!

See you there.

7.0

Nuke 7

The Foundry:

Set to be our biggest NUKE release to date, NUKE 7.0 comes with a host of polished and perfected tools to streamline your day-to-day workflow, as well as new additions to complement it's highly acclaimed feature set.

Looking forward to the new roto and paint. Good to see a updated 2d tracker. Basics, I'm liking this. The filters and Kronos will be great on the GPU.

The new 3d system stuff always powerful in the right hands.

Here is a link to the beta

OTS

I had a client over my shoulder all day today. I'd like to share some take aways.

  • Know your script. Don't fumble about.

  • Only make suggestions when you are not getting the feedback you need. Sometimes that jump starts the clients ideas.

  • Never make suggestions that cant be done. No jokes.

  • Stay with what you know.

  • Be ready to translate plain English into a group of nodes.

Last a tip of the hat to all the commercial artists, client based compositors, colorist. It exhausting.

I didn't f%* up.

Never let anyone tell you how to comp. Composite with purpose. Have a reason for everything you do, and the facts to back it up.

Defend your techniques.

If someone doesn't understand, explain yourself in a matter that gets the point across.

Most of what I do, make as many mistakes as I can in the shortest amount of time.

I love my job. Wouldn't change a thing.

Comp

When I'm asked: "What do you do for a living".

I get a lot of blank stares when I respond "Compositor". So most of the time the next job title i try is Visual Effects. I then get "oh ok".

I was an AD[Assistant Director]for awhile as that was the only thing that made sense to my family. Which is a strange idea altogether. I would make a horrible AD. The scary thing is Compositing is even more obscure.

If you search wikkipedia for Compositing, you get a page that gives the basic rundown. Putting the mathematics right on top of the page seems to be a bad idea. Scares people away. Scolldown and you will get into more of the nitty gritty. The main thought to take away from this page is that Compositing is not a new sport.

Compositing [Comp as I will refer to it] at its simpleist is problem solving. Layer A needs to be placed over layer B. Simple right, wrong. Not simple. Theres math, theres color. Most comps start off very simple but very fast turn into a huge web of color values being pushed and pulled to make a cohesive image that makes sense. Most of the time you work either background to foreground. Make sure all elements work based on color, sharpness and. atmosphere Its when you run out of buttons to push is when you have run out how good you are. As a compositor you have no choice but to be successful. You're the last stop. The Maya artists are done and the coordinators are waiting for your render. Make it perfect. Dont laugh I have seen those notes for shots. My mentor has "Make more Magical" come across his desk.

Understand Lenses, color and have ninja powers on your compositing software and all will be good.

Getting Good

Went to lunch last week with a old friend from Pactitle. We were a Shake house; this was before Nuke was for sale to the public.

Walking back from lunch he asked a question

"What are the important things to be good at with Nuke".

Tom, you asked a very complex question. Some would argue the channel system, some would say the mock 3D world. Others would say just straight 2d comp. Its not the RotoPaint node that's for sure. Here are some ideas and links. Enjoy.

  • Layers and Channels

    So much flexibility can be found in layers and channels. Not to mention cleaner more readable scripts. Which is always a good thing. Having your mattes and other images down stream from you is great. Here is a link to the nukepedia article about the subject.

  • RotoPaint

    Most of this is learned from Josh Johnson. He was and should still be the lead Roto/Paint artist at Digital Domain. Shuffle a white alpha before you start painting if you need to use the onion skin feature. This is not a fail safe. but it helps. Get good at paint in Nuke it will get you out of some bad rock and a hard place type of situations.

  • Camera Tracker

    If you have access to NukeX then get good the Camera Tracker. At least as good as you are with the 2d tracker. So much data you can get for free if you have a camera for the shot. Pointsto3d is amazing node. Aleast makes up for the shit paint in Nuke.

  • Python

    Yeah I know programing languages. Boring! It can make your life easier. Even if you just copy and paste other peoples Python[OPP] it can help. Python for Artists