Pretty much every fall we produce a few ads for Intel in time for the holiday pc game buying spree and we're proud to announce the completion of a new ad for them revolving around Farcry 2.
Check the video out in high quality, where leaders from rival clans in Far Cry 2 bump into each other while shopping for automatic weapons and fret about Intel powered computers that are making them paranoid and miserable.
Thanks to the Andrew Payton and his team at McCann Erickson New York who were the creative leads on this project, to Ubisoft for their support and to the guys at Luminous Arts who did an amazing job on the 3D assets, animation and much more: George, Greg and Alex Lambrakis, Erin Margolis, and the rest of the gang.
The event was a class act from start to finish and we were honored to be nominated alongside such incredible works of machinima that are truly moving the genre forward. Here's a link to all the nominees.
Kudos must be given to Phil Rice (who most of you probably know as Overman or Sol Bartz in Second Life) and Ricky Grove (also known as GToon Jun) as well as the awesome team who put the Expo together. These guys did an incredible job and conducted an event in world with a great vibe that had minimal technical issues (which is a HUGE achievement in Second Life!)
Congratulations to all who were screened, our fellow nominees, and our fellow jury award recipients. We look forward to next year's expo!
We recently used the new lip-syncing feature in Second Life and discovered some important issues, as well as the tricks it took to work them out. We used prerecorded audio that was supplied by the client and played back in game by the puppeteer, who simultaneously triggered animation specifically for that take. We recorded video along with audio using fraps, so the audio would act as a scratch track in post, which is when we would match it to the clean, original audio track.
In general, lip-syncing works at the client side, which means your computer handles the actual syncing, not the server. This is good news if you have a nice, robust machine (which is always best for any machinima), and means less server side lag. It's bad news, though, because the audio has to travel up to the server and back down, therefore causing other lip-syncing issues due to lag.
One issue in our case was that some of the dialogue matched fine, others not at all. One adjustment we made was to the level of the audio that was played by the puppeteer. Too low, and certain words didn’t make the avatar mouth move; too loud, and we got a serous echo—which meant extra lip-syncing that just didn't match in post.
Another adjustment for lip-syncing that didn't quite match was the "LipSyncOohAahRate," which is in the "Debug Settings" adjustments. ("Debug Settings" is in the "Advanced" menu.)
By default it's set to a value of 18.00. For our specific audio, adjusting the settings around 19 to 19.5 seemed to help. (This will probably vary based on your avatar and the dialogue.)
At times we had an issue with echoing that was heard only on the client side. It ruined some lip synched words and often created an extra word or two at the end, which resulted in extra mouth movements when a character’s dialogue should have stopped.
We also found that the speed of audio dialogue had changed by the time it was played back on the client side. In other words, when I matched the first word of the clean dialogue audio in my edit with the scratch track audio (recorded by fraps along with the footage), they would slowly lose sync. The playback audio that created the lip-syncing was a little slower and therefore stretched in duration. This meant I had to cut up the clean audio in post and shift it around to match the lip-syncing.
These are a few of the problems/solutions we have so far, would luv to hear from other users as good quality lip syncing, though hard work and frustrating at times, can only help our various projects.
Something else we posted in our previous lives. People liked it so we're posting again...updated with some important information.
One of our favorite things about making Second Life Machinima is the control you have over camera movement. We all know the alt+click, ctrl+alt+shift+click etc. But with the advent of the Joystick Flycam a little over a year ago the possibilities with this virtual camera are endless.
Important safety tip...With the new version, be sure and set your camera up by going to Edit -> Preferences, then click on the Joystick Setup button and enable your cam in the input and camera section. Sounds obvious, but if you're not used to doing that, it can leave you scratching your head when it doesn't just work.
We also disable "3D cursor" to allow the Joystick to flow and not snap back into place.
One particular shot that most camera operators know and love to use is called a dolly-zoom, push-pull, “the Hitchcock shot” or a bunch of other names (including “zolly” which cracks us up.) To over simplify, it’s done by doing a push or dolly forward while zooming out at the same time (or a pull while zooming in – whatever floats your boat.) If you want more detail on it, this wikipedia article explains it pretty well.
NOTE In the above example, the cam is NOT set up for a dynamic FOV change.
In Machinima, the only way to get this shot was to use a green screen. But in Second Life, you can use the zoom function in the flycam to dynamically change the FOV and mimic the shot. What’s more - it’s actually quite simple to set-up.
All you really need to do match your zoom mapping to an axis. Generally, we like to use the z axis, but you can play with it and see what you’re comfortable with.
In this example, we matched the z-axis and the zoom mapping with a value of 1.
Here’s a tip – if you still want do a normal push but have the freedom to do a zoom-dolly – raise the dead zone of the zoom to something higher than the default. This allows the zoom not to kick in until you reach the threshold you want. Generally, a value of 0.56 – 0.60 is about right to give you the flexibility, but again, you’re best off playing with the values and seeing what works for you as a camera operator.
Brenda Seabrooke’sCEMETERY STREET is set on Limbo Key and is filled with spooky suspense – including a creepy graveyard, a fortune-telling mother, satanic rituals, daring detective work and the utter agony of first love!!!
When 13-year-old Jeff discovers a 400 year old vampire in his bathtub his life becomes overwhelmingly complicated as he tries to juggle feeding of the voracious vampire, fending off an ancient evil nemesis, his suspicious mother and worst of all, his nosy girl neighbor! The story weaves from the present to the tragic events in Victorian New Orleans to create a mysterious and entertaining spell.
About a year ago, we had the chance to work with Eric Dziurzynski in the vMTV engine on a series called Eden’s Garden, and it was a total blast.
Our own Kerria Seabrooke lent her voice and acting talents to the character of Daisy and I was fortunate enough to do sound design and write music for the series.
It’s up for a Mackie this year and we couldn’t be happier for Eric and everyone else we worked on it with. Check it out if you get a chance!
In an earlier blog entry I alluded to using Apple’s Compressor software to encode your Fraps footage to a more Final Cut Pro-friendly codec and size for editing. This is an important step of the machinima production process, as using raw Fraps footage or other non-FCP native codecs eliminate many of the software’s real time features. The constant need to render while editing – or almost as bad, having the canvas drop frames or resolution while playing - disrupts the flow of an edit and can make the process become a chore, diluting the satisfaction of watching your piece come together in the edit suite.
Choose Your Codec Wisely
If you have access to Final Cut Studio 2 then the codec that you should be editing with is a no-brainer: Apple’s ProRes 422 codec is only available to those that purchase this software suite, and is a high-quality, relatively lossless codec that, in Apple’s words, has “Quality indistinguishable from the most pristine sources. [It] Maintains superb quality even after multiple encoding/decoding generations.” Aside from the quality, though, what is important with ProRes 422 is that its data rates allow for real-time, multistream editing without the need for RAIDs and other expensive storage solutions that are required for a typical uncompressed video workflow.
And if you don’t own FC Studio 2? If you have the drive speed (a RAID, most likely) an uncompressed codec will deliver the best results (note that I said an uncompressed codec, of which there are many flavors – selecting “none” to encode to is rarely a good choice for editing - “none” is not a codec!) For those of you who aren’t shelling out big bucks for high-end storage, you may want to consider encoding to something like Photo-Jpeg (nearly lossless when set to 100% quality, but taxing on drives and thus not an ideal “real time” option for FCP) or DVCPro50/DVCProHD (a popular choice for video work but one that I’ve never gotten great results in with machinima). Animation set to 100% is also a lossless codec and great for archiving, but not so great for editing. H.264 is a popular codec, but is NOT an editing codec, but rather a delivery codec – don’t edit with H.264-encoded footage!
For the purposes of this article we’re going to stick with ProRes 422 and assume that if you haven’t already upgraded to the newest edition of FCP, that you will soon.
Don't Ignore Frame Size
FCP is fairly resolution-independent and able to work with sequences of various sizes. Like using the wrong codec to edit with, non-standard sizes can require rendering at almost every turn, which makes for a frustrating edit. For this reason I find it best to use a “native” sequence size when working with standard def footage. Nearly all of my SD work is cut using a sequence size of 720x480, as it compresses proportionately to the square pixel 4x3 web sizes that I use most often. (We won’t get into square vs. rectangular pixels here, as it is a lengthy topic, but check out this link for some great info on that subject.
As mentioned in my prior article, it is best to should at resolutions larger than 800x600 for the Perian Quicktime component to be able to properly decode your footage once you have it on your Mac. There are two schools of thought here: shoot at 802x602 and trim the “extra” pixels off using Compressor as part of the ProRes encoding process (a little more complicated, but makes for smaller file sizes) or shoot at 1024x768, which is proportionate to the 720x480 size that we are shooting for. Again, even though those two resolutions don’t seem like they are the same ratios, remember that 1024x768 is using square pixels, the only option with Fraps, and 720x480 is using rectangular pixels, which are native to FCP. Here I will walk you through the settings to use to compress 1024x768 to 720x480, but if you have questions on the 802x602 process feel free to drop me a line.
Setting up a Preset in Compressor
Now lets walk through the steps necessary to enable Compressor to process your footage:
First, import your footage into Compressor. Next, you’ll need to tweak one of the presets to the settings that we need. Do this by duplicating the preset called “Apple ProRes 422” that is found in the Settings Tab > Formats > Quicktime. The new preset called “Apple ProRes 422 copy” will appear in the Custom folder of your Settings tab. Double click the new preset to open its settings window and then we can go through the buttons at the top of the settings window.
The first button gives you a summary of the settings, so we can skip that. The second button brings up the Encoder Settings. With this button selected, click the Video Settings button. This will bring up the standard QT compression dialog box. By default the compression type will be set to Apple ProRes 422 HQ. There are actually two flavors of the ProRes codec – both are 10-but, but HQ has a slightly higher data rate. It’s your call whether you’d like to use HQ or not, but to save file sizes I typically work with the non-HQ flavor unless I know that the piece is going to broadcast or to be heavily composited or effected, for example.
Keep frame rate at “Current” (assuming that you are shooting at 30/29.97 fps), set key frames to Automatic, make sure “Interlaced” is unchecked, and keep “Enable 4:4:4 chroma filtering” checked.
Change “Pass-through” for Audio Settings to “Disabled,” as we aren’t dealing with audio here. Streaming should be set to “None.”
The next button is for Frame Controls. By default it will be turned off and the whole panel will be ghosted, but we want to tweak these settings as well, so click the button next to “Frame Controls – Off”, and then turn Frame Controls to “On.” This will un-ghost the rest of the panel. Most of these settings are probably OK set to their defaults, but since there are so many variables that could arise in shooting (especially when this is being read by an audience that could be using different methods to acquire their source footage), let’s make sure we are all on the same page: to do so, use these settings:
Resize Filter – Better
Output fields – Progressive
Deinterlace – Better
Adaptive Details – checked
Anti-alias and Details Level – sliders all the way to the LEFT
Rate Conversion – Better
Set Duration to 100% of source (this setting will vary if you have not shot at 30/29.97 fps – drop me a line if this is the case for the lengthy explanation)
The next button is for Filters. We won’t adjust anything for the purposes of this conversion.
Next the Geometry button. All of the “Crop-to” fields (left, right, top, bottom) should be set to zero (0) – this is where we’d be cropping pixels off if we had shot at 802x602. Next to “Frame Size” where it currently says “100% of Source,” adjust the popup to “720x480.” Pixel aspect should stay at default for size (1.1250) and all of the “Padding” fields should remain at zero (0).
The final button is for Actions, which we won’t address here as it doesn’t have a bearing on how the footage will be compressed.
Save the new preset, and rename it if you’d like. Once all of your footage is imported into Compressor your can select all of the clips in the Batch window and drag the preset over one of them – it will apply it to all the clips in the batch, then Submit, and when the dust settles you’ll have a folder full of gorgeous ProRes-encoded 720x480 clips that should play in real time in your FCP sequence, making for an enjoyable edit session.
We just found out that Tiny Nation has been nominated for an award, Best Technical Achievement from the 2008 Machinima Film Festival. Awesome news, gratz to the team for their hard work and to the other nominees as well. If you're in NYC Sat Nov 1 do stop by the festival as it's looking to be great this year, held at Eyebeam, 540 W. 21st Street, (between 10th and 11th Ave), starts at 11am.
We're happy to announce the world premiere of our new short film - TINY NATION!
Follow the heroic adventures of a band of motley mammals.
They're cute. They're small. They're trying to take over the world.
Together they raise the question – WHAT'S IN YOUR BACKYARD?
Look for more - coming soon!
I'd like to take a moment to give special thanks and acknowledge the amazing contributions of Damien Fate, Sean Daniels, Jason Crisman and Michael Johns and of course Paul Jannicola, Frank Dellario and Tom Donnelly. Many thanks also to Razzi Rockett, Stacie Ball and Calamity Hathaway.
