The opening shot of the sequence (not quite the opening shot of the film) isn’t going to be particularly difficult - at least, not when compared to some of the other shots.

You can either watch it on it’s own here, or as part of the whole sequence here

This shot will be a purely 2D shot. Backplate will be a crane on the roof of a building (or locked off… I need to find out what kind of budget we’re going to have for equipment), and the two foregrounsd characters will be shot against green-screen. Because of the amount of motion blur that they are going to have, we could almost get away with doing all of the horizontal motion and motion-blur in post. Keying people with lots of motion-blur off DV does not sound like my idea of fun at all. We will probably shoot both ways and then see how the key looks.

Best-case scenario is a lovely craning shot of the city and two easily keyable people with natural motion-blur.

Worst-case scenario is having to use a locked-off shot (or, god-forbid, a still) of the background and using manually cut-out/painted stills of the character and whacking the crap out of them with motion-blur so nobody realises that they aren’t actually moving.

The TimeX node is one that people are either terrified by or ignorant of how it works. If you’re not afraid of it, then you shouldn’t be usingit. I’ve seen numerous scripts completely collapse around someone’s ears (often my own…)

It’s a bit like any other extreme sport - if you have respect for what it can do to you and your script, you should be safe.

There are two main issues that the TimeX node presents - unexpectedly changing the time, and keyframes.

• Changing the Time

  To understand what’s happening here, you’ve got to understand how Shake works internally. Shake is a bottom-up system. When a node needs an image, it says to its inputs “You need to give me an image at time X. I need THIS part of the image, and I want you to give me THESE channels”. These values work their way up the tree - each node will either pass them on or modify them in some way (a Reorder node would change the channels, a Transform node would change what part of the image is needed, and a TimeX node would change what time is wanted). Nodes with more than one input will pass it on to whichever inputs it needs, until the request reaches an input node (SFileIn, RotoShape, etc). The input node will then do its thing, reading the file, or generating an image from scratch, and will then pass this image back down.

  Every time a node gets an image returned to it, it will do whatever it needs to do to that image, before passing it on to whoever called it. For a single node at a certain time, there is only one output. If there are two outputs requesting different things (different channels, for example), then it will ask its inputs for everything it needs, and then pass on whatever needs to be passed on.

  I’m sure you can see the problem with this method… If one input asks for an output at time X, and another one asks the same node from an output at time Y, then what does it do? It can only give one image output (which is fine for different channels, but not for different times). I have no idea how it actually chooses, but it always seems to go with whichever one has the TimeX node on (if there are two TimeX nodes, it seems to go with whichever one was created first).

  If you wanted to do some kind of fake motion-blur, you might think that you’d be able to do something like this:
  
  with “time-1″ in the “newTime” parameter of the TimeX node. This, however, won’t work - all you will get is the whole thing at “time-1″. Try it yourself - if you don’t have any footage around, use a Text node with “%F” as the text (this shows the frame number).

  The solution to this is this:
  
  Even though the two SFileIn nodes are identical copies of each other, because they are still different nodes, they can cope with passing out different images.

  The essential thing to remember about wiring in a TimeX node is to keep the node-tree above the TimeX node completely seperate from the node tree below. The only way anything should be able to get from above it to below it is through the TimeX node.

• Keyframes

  If you have a node with on-screen-controls, the on-screen controls will always show what the node should be outputting at the current frame, and will ignore any TimeX node that may be affecting it. When you move them, though, you’ll actually be changing the parameters at the time that is affected by the TimeX node. What this means is that the on-screen controls will seem to be acting very strangely. Sometimes you’ll grab something and the image will change, but the on-screen controls will be stick in place. Whichever way it goes, it is sure to confuse you the first few times. Once you realise what’s going on, you’ll start to pick up on what is acting strangely. I don’t really have a ’solution’ to this one - just make sure you’re aware of what the TimeX node could be doing to your script.

All that said, the TimeX node can be really useful. I find I use it mostly when I need to do something like showing only every other frame, for 2 frames each. (”(floor((time-1)/2)*2)+1″ will give you frames 1, 1, 3, 3, 5, 5, 7, 7, 9, 9, etc, giving a stop-motion-like effect.)

Let me know if there’s anything you don’t understand in here, or if you think I’ve got something horribly wrong!

Crap. So much for regular updates…..

I spent saturday working through the second half of the opening scene’s previz (pretty easy to put together, as a lot of the shots could be completely static). This is our VFX cut - I’m also sending off the individual shots to Dom so that he can cut them together how he wants to.

We decided to start off the previz without thinking too much about how we were going to do each shot. We’re going to be sitting down with Dom later this week to go through the cut, dicsuss any changes he wants us to make, and think about the practicalities of each shot.

Before you watch this previz, do be aware that there are a number of shots that are still planned to be re-worked - RO_101_030 and RO_101_170 are ones that springs to mind at the moment.

Anyway, without further ado, here you go. It’s not well lit, and it’s not well animated, but it gives a reasonable impression of what kind of shots we’re looking at.

I’m going to start on a daily (well, vaguely daily) breakdown of the shots, and implementation thoughts on each one.

EDIT: New image at the bottom….

It’s been a bit of a slow week, but we’re just about to start animating the previz. For the previz, I did a very arbitrary layout of various buildings (who am I kidding? They’re stretched grey cubes…) with the main building (okay, that’s a cube with a wall around the top) in the middle.

So that you don’t all get very bored with the apparent lack of updates, I thought I’d give you some shots of what the previz city was looking like. You’ll never actually see the whole thing in the film itself - only bits and pieces, but, for the previz, I wanted to put together a model that I could use throughout.

The whole set:

(the main building that a lot of the action will take place on is just to the left of the much taller building that’s right in the middle. If you look closely at the main building, you’ll see that it’s got a bit of a ledge around it.)

A view through the buildings:

This is much more the kind of shot for the first few shots of the film - it’ll be very fast moving (yay for motion blur!) and will focus on the main characters.

The main building:

This is where the second part of the opening will take place. The standing block is 2m tall (6′7″) which is approximately the height of the characters (okay, so I picked it as it was a nice round number to judge the scale against. Incidentally, the scale I’ve gone for for the whole of this is 1 unit = 10cm - so the characters are approximately 20 units tall.

The view from the main building:

Just to show what kind of view we’ll be seeing from the roof top.

The characters from the previz:

These are the two characters we’ll be using in the previz. The gold guy is the “good guy”, and the red guy is the “bad guy”.

Hopefully I’ll have some actual animations to post soon. I’m not intending to spend huge amounts of time perfecting each shot, as these are much more to figure out the edit and how we’re going to shoot them. I have no intention of trying to impress anyone with the previz.

And I’ll be bringing you more Shake Tips soon!

I had a meeting today with Dom (director) and Tim, who’s helping out with the pre-viz.

Dom had done some great storyboards using photos of posable characters, and we spent most of the meeting talking about exactly what he was looking for and how we would potentially be able to film them. It’s going to be interesting (if anyone knows of a greenscreen studio with a wire rig in or near london that would be available relatively cheaply, please do let me know….!)

Tim and I are going to be starting on the pre-viz this weekend. It looks like there are going to be 2 or 3 very hard shots in the opening sequence, and the rest will just be hard.

I’ll get some of the storyboards scanned in this weekend….

I know it’s been a little quiet this week - I’d like to apologise for that - things have been changing around a bit at work for me recently, and I’m also moving house this weekend (and so will probably be internet-less at home for a bit… I’ll still be keeping things updated on here, though)

Pre-viz on Roy’s Odyssey starts next week. There will be two of us working on it (I’m not mentioning anyone else’s name on here without their permission… usually it’s because I haven’t asked, though), and we’ve got a meeting with Dom (director) and the writer next week.

We’re going to be pre-vizing the whole of the opening scene, mainly so that we can plan how to shoot the flying sequence, but also to help the director decide how to visualize the scene. I don’t know yet if I’ll be posting whole quicktimes of the scene, but I’ll certainly show stills of the kind of shots we’ll be looking at, and, if you’re lucky, you may well get a preview of the scene….

More info as it comes!

I know I’ve already linked to HD For Indies once this month, but I followed a link to another post over there that was incredibly informative, especially as I’ve been thinking about the different HD formats recently….

Thanks to worldofmaya for posting on VFXTalk.com with this link:

HD For Indies: Format Overview: what the various tape formats really record

In the article, Mike runs through all of the available formats, what they capture, what they record to tape, what compression is used and what visual quality is lost. If you’re thinking of shooting something on HD of any kind, then well worth a read.

Oh, and the shake tip will be coming at some point this week….

Well, the shoot yesterday went well, except for a couple of small issues….

The plan was to go and get two shots (things changed since my last post on this), that we could use to develop the look of the lasers from the two characters in the opening scene. We decided to go for dynamic shots of each character, focusing on a medium/wide shot that was close enough to get the details of where the lasers came from, but wide enough to see how the lasers looked travelling over a distance. The shoot consisted of the director (Dom) and me (and yes, there are pictures….!)

I took my shiny new tracking markers, my digital camera, 5mm graph paper and coloured pens (thanks Jeff), a tape measure, and a vague plan.

First hitch of the day: I turned my camera on, and the battery was dead. When I got home, I realised that, while the battery had been in the charger, the charger hadn’t been plugged in. Bugger. As a result, all images in this post come from the footage that we shot.

We decided to do both shots with and without tracking markers - in the end, we only captured the ones with the markers, but it was the kind of set that could probably be tracked just as well without markers. On the actual shoot itself, I’m going to hold off using markers if I feel they’re not needed.

Anyway, how about the shots themselves?

Click here for the full res image (Stretched from 1440×1080 to 1920×1080 to give a pixel aspect ratio of 1.0)
(That’s Dom, the director, in that shot)
Play quicktime here (7.8Mb) - Will be switched out for a smaller Sorensen one this evening)

Click here for the full res image (Stretched from 1440×1080 to 1920×1080 to give a pixel aspect ratio of 1.0)
(And that’s me… ahem…)
Play quicktime here (7.2Mb) - Will be switched out for a smaller Sorensen one this evening)

If anyone’s interested, the shoot took place on the roof of a flat in Camberwell, SE London. We don’t know yet if this will be the one that’ll be used in the shoot itself (I think they’re planning for a more central look, but they’ve got to find a location first!)

Once we got back, I captured the footage onto my computer (Using the trial of Cineform Connect HD), and exported to an image sequence using Virtual Dub.

I also managed to get my camera working, so set about taking some reference photos (for modelling for 3D roto):

I’ll keep things updated as the shots progress….

And you’re going to have to wait until tomorrow for this week’s Shake tip.

I discovered something not so good today when shooting this concept footage…. (More details of how that went later…)

The Sony HVR-Z1E, which is what we’re using on Roy’s Odyssey, only allows you to do PAL, NTSC or 1080i…. I was sure (must have dreamt it) that you could do 720p on it, which is what I was going to persuade the production team to use….

So we’re going to be working interlaced all the way, it seems!

Ah well - I’m capturing the footage now, and I’ll post again once I’ve got it all ready!

I got home tonight to find a parcel of tracking markers waiting for me - just in time for their first outing this weekend:

Once we’ve done this shoot this weekend, I should have a lot more for you - hopefully this week’s lull won’t continue for too long.