More HDV Camera Tests
Posted: January 31, 2006 at 1:35 pm, by Isaac
DV.com recently ran an article written by video guru Adam Wilt about a lab test of the main four 1/3″ CCD HD cameras – Canon XL H1, JVC GY-HD100U, Panasonic AG-HVX200, and the Sony HVR-Z1U. Unlike 24′s production trial, this test involves a side-by-side setup with lots of operators, observers, models, charts, multiple lighting , with everything routed through the same banks of recording, monitoring, and diagnostic tools.
For comparison’s sake, a Panasonic HDC27F Varicam and a Sony HDW-F900/3 CineAlta were also set up and tied into all the testing gear. Obviously, with so many cameras and so many independently adjustable features on each camera, there’s only so much that can be done in a single day, so the tests were really only focused on sharpness, clarity, latitude, and general quality of the CCD array and image processing of each camera.
Here’s the brief conclusion: the four little cameras are pretty similar in most respects, and are getting pretty close to the two big cameras in most respects. However, the 1/3″ CCDs tend to have about half of the crispness of the 2/3″ CCDs. Of the four, the Canon (1080) was the obvious winner in terms of raw, clear, unaliased image detail, with the JVC (720) second, and the Sony (1080) and Panasonic (1080) tied for third. In terms of overall (and more subjective) image quality, each camera had its own strengths and weaknesses. The Sony had the lowest noise. The Canon had more image control. The Panasonic handled highlights and skin tones well, but not quite as well as the very noisy JVC.
There’s more info in the article, as well as frame grabs and video clips. To read it you’ll need to sign up (free) for a DV.com account. This article alone makes it worth the trouble. However, I must offer a brief disclaimer. Bear in mind that the results of these experiments don’t necessarily reveal the best all-around camera for the independent filmmaker. There weren’t any motion tests, codec tests, stress tests, lens test, or power tests and none of the cameras were taken outside. Yet. Keep an eye out for more results.
HDV on the Set of 24
Posted: January 27, 2006 at 1:22 pm, by Isaac
Ok, this is a couple of days old, but it’s required reading for anyone who wants to know what the HDV cameras are like in a real production environment – in this case prime-time TV drama. Showreel magazine has an article about DPs Rodney Charters and Taylor Wigton experimenting with some recent HDV cameras on the set of 24. Showreel’s articles are usually members-only, but this one is free.
24 is an ideal testing ground because it has an experienced staff, a solid production track record, and already uses a great deal of video for security camera and news footage. Also, a lot of the drama takes place on a single set with highly controllable lighting, which is good since these cameras tend to have very limited latitude. At the moment the show still uses 35mm film for the exteriors, but the DPs have been using the Sony Z1 and the JVC HD100 on the set for stage one of their tests. Stage two will involve the HVX200 P2 and the Canon XL H1. I’m looking forward to seeing what stage three will be…
Also included in these tests are the requisite acceories for these tiny cameras; matte boxes, fig rigs, and most importantly, adapters for the 35mm lenses already being used on the show. This is important for a several reasons, firstly so all the footage will match, and secondly so that the focus pullers are able to have total control over a high quality lens. 24 is generally a very free-form, action-packed, hand-held show, so rock solid remote focus options were a must.
Keep an eye out for articles on the next batch of cameras, and more info on the dailies, editing, and post processes involved.
Premiere Pro 2 & After Effects 7 Released
Posted: January 19, 2006 at 1:07 pm, by Isaac
Ok, updates to the site have been a little scarce lately. At the moment I’m out of the country… long story… complicated project… However, since I tend to recommend Adobe products, it would be remiss of me not to take a break to mention that Adobe released Premiere Pro 2 and After Effects 7 a couple of days ago. This isn’t really a surprise since there were many rumors concerning this, but now that it finally is officially out, we can see what’s new.
Premiere Pro 2 has a better interface, HDV and DVCPRO HD/P2 support, more color corrections tools and effects, more DVD encoding and exporting options, and better integration with other Adobe products. It also seems to have had some serious retooling under the hood, with a new video engine, supporting resolution up to 4k and color depth up to 16-bit, and utilizing the GPU for some rendering tasks. Sounds very neat indeed.
After Effects 7 also has a new rendering engine that makes use of OpenGL, provides plenty of speed increases, and can render in 32-bit space. You can make use of this with the many new input/output options. There’s also a new TimeWarp feature, with works like Twixtor or ReTimer by using per-pixel warping to create interpolated frames for smoother slow motion ramps or adjusting the amount of motion blur in a shot. Of course, it also has the revamped, dockable interface that Premiere does, and the same back-and-forth integration options.
That integration is part of the new Dynamic Link feature, which cleverly allows Adobe apps to share the same system resources. For example, if you have the same video clip loaded into After Effects, Premiere, and Encore, rather than loading the same clip into three separate RAM partitions for each program, there is a single shared RAM allocation that all the applications use, so that all of them simply see the same clip. In this way, you can load the same large project into all the programs without increasing the load on your system too much.
It also means that, since all the programs are sharing the same resource files, they might as well share their output files too. So, with the new system, you can, for example, drag unrendered After Effects compositions into Premiere or Encore, and see and edit them in real time as After Effects generates the preview in the background. In order to find out more about something that is more than the sum of its parts, check out the new Production Studio, which includes Illustrator CS2, Photoshop CS2, After Effects 7.0, Encore DVD 2.0, Premiere Pro 2.0, and Audition 2.0, and the Adobe Bridge tool.
CES 06 Overview
Posted: January 9, 2006 at 1:05 pm, by Isaac
Ok, time for a wrap up of this year’s International Consumer Electronics Show. In the old days, CES was where retailers and developers went to see the latest clocks, lamps, refrigerators and so forth. Now it’s a place where everybody goes to see and introduce the newest DVD players, plasma screens, computers, portable media players, HD gear, video game consoles, cell phones, camcorders, and home theater systems. Almost every product or service at CES this year was related to media or visual entertainment in some way.
Scott Kirstner is calling this the Year of Video at CES. It’s not NAB yet, but the emphasis on media production, consumption, and distribution was overwhelming. The big electronics companies and big studios teamed up, and major actors like Tom Hanks, Morgan Freeman, Robin Williams, and Tom Cruise were on hand to introduce products and speakers. The FCC was there to talk about issues with the new media technologies, and pretty much everyone was thinking about television and movies.
Engadget had the fastest, fullest coverage, with a full team live-blogging lots of the keynotes and providing plenty of pics and mini write-ups of all the new stuff. You can also check out ArsTechnica for more in-depth articles on such things as Intel’s keynote, which pushed the digital entertainment platform Viiv and its subsequent tv and film distribution potential, Microsoft’s keynote, which touted Windows Media Center and its tv and film management ability, and Google’s keynote, which introduced the Google video store and its partners. Seeing a theme here?
Let’s focus on Google for a moment. It looks a lot like Apple’s new video store; a distribution contract with a major network (CBS) and record company (Sony BMG), and a couple of dollars per download. They also have all upcoming NBA games, classic cartoons, and a few indie offerings, but by far the greatest selling point is the ability to upload and sell your own videos. No mega-million-dollar studio contract is required to be a vendor in the Google Video Store, just upload your clip — be it a short, how-to, documentary, or feature — and choose a price for it. That’s all that will be required. No real info on its DRM solution yet, just that it will have one.
Obviously, there were a lot more things that went on there, but the Google news is, I think, the most interesting for those trying to make and market independent films. There are plenty of other articles about all aspects of the show, and now that the event is over you can expect to find plenty of editorials about the tech and breakdowns of each keynote. And this week we’ve got Apple vs. Adobe at Macworld SF. Fun fun.
Displacement Maps in 3D Animation
Posted: January 3, 2006 at 12:58 pm, by Isaac
While digging through an old backup hard drive the other day (resolution 1: manage files better) I found a 3d test that I had completely forgotten about. Six months ago, I had a couple of days free, and spent them experimenting with a few different way of creating and using displacement maps for complex organic modeling. I call this post “Remember the Allosaur” (resolution 2: fewer puns).
A displacement map is a texture or procedural map or that can be used to distort or displace the geometry of a 3d model. Generally, this is used to animate flapping flags, rippling water, or swaying vegetation, but with the right tools, it can be a big help in creating complex organic models with simple geometry. I started out with a very simple Allosaurus head and neck, which took about 20 minutes to build, and had all the detail of an early 90s video game character.
This blocky head could be smoothed out by subdividing these original 480 polygons into thousands more, rounding off the hard edges, but I needed a way to control how the new geometry would be generated. I had to paint a displacement map that would add these details. This texture map would be a simple greyscale image wrapped around the allosaur describing what the new high-density mesh should look. The white areas would bulge out, and the black areas would crimp in.
In the image below (click to enlarge), you can see the blocky base model with a simple displacement map applied. Most of the map is close to 50% grey, because the model is pretty close to what I want, but you can see white smudges around the eyesocket, base of the skull, and most notably along the vertebrae of the neck. As I turn up the level of subdivision, the new shape of the model becomes more detailed. At nine subdivisions, there are nearly 75,000 polygons, and fine wrinkles and folds are now obvious around the nose, jaw and eyelids.
To paint my displacement map, I used ZBrush 2
, which allows the artist to see the changes to the topography of his model while he paints directly on it. It’s a fantastic new tool for organic modelers, and it was great fun to use. I could spin my character around under the virtual light while sculpting bone structure and musculature directly into the geometry with broad strokes of my Wacom tablet. It’s almost as effortless to use as you could imagine; except that it isn’t.
The creative part is very simple and intuitive, but ZBrush suffers from being the first program to do what it does, and so I had no idea how to use it. The two days I spent building the allosaur were mostly taken up with trying to figure out how to import Lightwave models and UV coordinates into Zbrush, and then how to get a working 16-bit displacement map out. Not to mention trying to get my head around how to use a program that can accomplish any 2d function in 3d space – and almost every 3d function in 2d space. It’s every bit as complicated as you could imagine.
But actually painting the map itself was a breeze, and after rethinking all three dimensions several times, ZBrush started to feel a little like it was defying all the laws of the universe. Above is a detail from the neck area, much emphasized to you can more easily see the individual scales. In the actual map, the skin details are tiny variations of grey on top of the much more dramatic changes of the underlying bone and muscle details (the broader black and white strokes on top of the scales and wrinkles).
This is just a real quick throw-away modeling test, but if I was going to use it as real character for animation, I’d need to rethink the process in order to separate the motions of bones, muscles, and skin. The “proper” way to simulate this is to actually have bones and muscles built, with a layer of skin sliding over the top, but this is a pain to calculate, and there are very few off-the-shelf solutions for this. It can be cheaper and quicker to use “muscle bones” that fake the stretching and bunching of real musculature, and morph targets for extreme muscular distortions and fat jiggles, a method that would work nicely with a few layers of displacement maps.
The animation package messiah will even allow you to attach one displacement map to the polygons (for skin details) and another to the underlying rig so that muscle bulges and skin textures and move independently while still affecting each other. Regardless of which package you’re using, the possibilities for using weight maps, procedurals, expressions, shaders, and morph targets to affect your geometry by altering displacement maps are limitless. Even this very quick-and-dirty experiment, with only a low amount of subdivision and no other texture maps, has a very controllable level of fine detail.
Of course, that I’ve done is nothing ground-breaking or special – after all, I built this six months ago. This is currently a much-used method for detailing organic models, and even creating characters almost from scratch. It requires more polygons when rendering, but it gives you full access to a lower-rez version for animations. That means less geometry for dynamics calculations to worry about, better rigging control, a more convenient way to make changes, an easier initial modeling task, and, at the end of the day, more detail.
Posted: January 1, 2006 at 12:52 pm, by Isaac
I’d like to introduce the readers to the greatest show on television. This is of course predominantly a film website, but even veteran filmmakers could do worse than work a few seasons of Top Gear. I don’t actually watch much television, but I don’t have to in order to be certain that the BBC has created the top show.
And I’m not even a car guy. That is, I wasn’t until I started watching this program. It’s pretty standard format for a motor program, with the three main hosts test driving various cars around the countryside or on their very own track (formerly an airfield) and then returning to discuss their results together with a live audience on the show’s set (formerly one of the airfield’s hangers). If you are a car guy, and I mean, a REAL car guy, you’ll get a kick out of the rather special wheels that they get given to test. If you are a media person, you’ll be blown away by the show itself.
The editing: perfect. All camera work: perfect. Sound design: perfect. Music: nearly always flawless. Post processing: rarely ever overdone. Directing: consistently awe-inspiring. It is a thing of great beauty, much like the vehicles often featured. In fact, there are very few top-dollar car commercials shot and edited as well as most of the segments on this show. Whether they’re speed testing the latest multi-engine supercar or grinding a custom-built Baja truck down cliffs, the end result combines substance with style.
Almost as exemplary as the lighting and lenses and the superb driving (both on the part of the hosts, experts, and camera car team), is the writing (both the scripted narration and off-the-cuff reviewer reactions). The creators of this program know how to tell a compelling story, even if they’re only discussing the differences between a 2wd and a 4wd Porsche chassis. Generally though, they take on more interesting stories by setting up races between cars, or cars and various other forms of transport. Recommended study material for anyone currently working in any area of video production.
I’m not sure how long Google Video files stay fresh, but at the moment you can watch the Bugatti Veyron episode (warning: mild language and extreme British humour), and if it vanishes you can always search for more clips. You might look out for the team using a pneumatic cannon to play darts with old junkers, racing rally cars against bobsleds, testing aerodynamics by driving behind a 747‘s engines, and so forth.