The Pixel Crush

-------------------------------------------|Digital Animation & Game Criticism|-------------------------------------------

Friday 5 October 2012

Directorial Legacy

Not long ago I was asked for some advice on making student films in the final year of university.
I thought I'd reproduce it here, with some considered editing. Here are the 8 arbitrarily chosen truths of student animated shorts.

  • First off some background rendering knowledge, I wrote a tutorial on the linear lighting workflow and specifically using final gather to light your scenes, and about setting up for it, with exposures, colourspace, and stuff. A good grounding for rendering. Also This is a brilliant blog for mental ray rendering in general, a great resource for different techniques and tricks, and there are plenty of places to find out about linear workflow elsewhere on the internet.
  • Start creating artwork soon, we were always short of design stuff, and the more you have the sooner you can start modelling, trick your artists into doing turn arounds if you can ;) Reference is so important, something I'm learning even now. If you have photographs or concepts to work to your art will shine shine shine. I'm not saying make everything photoreal, but faithfully recreating specific aspects of reality is vital- this lighting here, that material there, these shapes here, and these colours there.
  • Make sure you know your stuff and have thought about the world, the characters, the practicalities of making the film, and anticipate the questions, then you'll be covered. Your team deserves a director that knows their story and can make decisions quickly based on a world that can hold up to scrutiny. You're going to be making hundreds of decisions, and fast, so if you have nothing to inform them you will struggle to keep up.
  • No one seems to be able to avoid this but nail down the story as soon as possible, without compromising too much. People both above you and on your team will keep you making changes 'til the 11th hour and listen to what they have to say, but ignore the bits that don't feel right, or jeopardise the production. If you agree to things you don't feel comfortable with I guarantee you'll regret them later.
  • If you're directing make sure its all working at a storyboard and animatic stage, and make sure your team is on board with it then, and NOT mutinying and demanding changes later. We kind of bodged our layout phase so I cant really advise much here, by this time the animatic was out of my hands so interpreting camera angles and timing was tricky, if you can, keep the storyboard artist and layout artist the same, someone with a cinematic eye to keep things consistent, if its done well these layout scenes can become the templates for every shot, incrementally becoming more complete with each pass.
  • Finding a good file structure is essential and difficult. Breaking things down into folders of scenes and shots works, and organise props and characters separately. Nail this early. Make sure the team knows this early, and make sure they are fucking neat. Clean out scenes after they've been modelled etc, no spare shaders, textures, geometry etc, use the optimise scene tool, delete duplicate shading networks, merge texture files, cleanup meshes. Just the model alone. preferably unwrapped ;)
  • Make sure your team talks to each other when they have a problem or they'll come straight to you and you won't be able to spend time fixing everything. Texture at a high resolution, you can always downsize later, you can never upsize. We thought we could get away with some small textures and then ended up having giant pixels everywhere,texture at 4k for large props, just to be safe, 2k for small props. Be efficient with UV space, while its not essential because memory is rarely an issue it makes texuring and shading much faster to only have one texture map to refresh. Don't use tiffs, theyre enormous, pngs are good, jpegs are only acceptable when very uncompressed. No ngons (polygons with more than 4 corners), try and unwrap neatly, one, maybe two textures per model. rather than a texture for each bit of the model. I recommend roadkill, free, effective, slightly buggy. If you don't clean up things like ngons mudbox and other software will reject the mesh outright.
  • Learn nuke, it'll serve you well in the industry later. but if you don't feel confident or comfortable with it (like I didn't) after effects is still fantastic, just try and avoid the default stuff and customise the look as much as possible. While we're on look, don't just stick lamberts on everything, shaders make your look (aside from colour and form). They make it look clayey, or glossy, or fuzzy, or whatever. They define the overall feel the objects in your world have. For Kernel the key lookdev research was about translucency. See through surfaces of the green house, thin leaves absorbing light, and Leonard's skin doing the same. All those things took aaaaaaaaages of tweaking and experimenting to 'perfection'. Use the mia_material_x_passes and put fresnel on everything, even if its just a little bit.

With a good story, the most hard work you've ever done, and good people, you can make an awesome fillm, and you will never have been prouder in your life.



Here's a little nugget I found on my phone the other day, its reference for this shot.
I love my really unhelpful directing. "Now do this, and this, now this. Done." Functional I suppose.

Pixel Propaganda

A cool looking tool for designing game mechanics, quite in depth though.

Some fancy realtime(ish) rendering techniques using Maya's viewport 2.0.

A great look at the making of indie game Limbo, and its unusual development process and team.

The new Blender Foundation film incorporates live action and, as always, is extremely impressive considering its open source software.

No comments:

Post a Comment

Let The Discussion Begin: