Product: Lustre Digital Grading System
Version: 2.5 (beta)*
Manufacturer: Autodesk Media & Entertinment(formerly Discreet)
Price: Low-mid six figures (various configurations offered)
Performance & stability: n/a*
Features: 8/10
Expected return on investment: 7/10
Support: n/a*
Longevity: 10/10
Flexibility: 7/10

Pros: Incorporates conforming, dust-busting and some effects alongside a very powerful and extensive colour-correction engine
Cons: Some careless omissions in the feature-set, too expensive to buy on speculation, suffers from being too integrated
Bottom line: The best commercial digital intermediate solution available, and it’s only going to get better

Lustre 2.5

*This review is based on demonstrations, evaluation of documented features and interviews with the developers rather than first-hand experience, so no data is available for reliability or performance.

Discreet has long been known in the post-production industry for creating high-end turn-key systems, such as its Inferno and Flame products for effects work, as well as for its online editing tools, Smoke and Fire.  More recently, it has targeted the desktop market, with its Combustion and 3DS MAX lines.  Discreet seems now to have its sights actively locked onto the whole digital intermediate pipeline, with its new Lustre product taking the helm.

I say new, but this product is evolving quite rapidly.  At the time of writing, version 2.5 is available to buy, and version 1.0, released in 2003, was in turn based upon the grading system developed by Colorfront, which was itself used to grade the Lord of the Rings trilogy.  Over the course of Lustre’s development, a number of improvements have been made, but the most recent incarnation of the product brings the most interesting advancements, most likely because this time around Discreet have incorporated feedback from colourists and other users of the software. Rather than go through the new features and improvements, I’ll assume that you’re unfamiliar with the product line, and examine it as a whole. In fact version 2.5 has probably sold more new copies than upgrades to existing licenses, reflecting the move in the industry away from turn-key systems such as Thomson’s Specter.

Lustre is designed to run on any Linux or Windows-based PC, but obviously the higher the host machine’s spec, the greater the performance, allowing real-time playback of film at 2k resolution with the right configuration.  The interface is reminiscent of Discreet’s other systems, such as Inferno, so operators with that sort of background will probably feel most comfortable with it.  There are separate sections for editing (conforming), image manipulation (resizing and repositioning, as well as a dustbuster), colour grading, effects, and output.  In fact, Lustre is equipped to handle almost all of the processes in a typical digital intermediate pipeline.

There are plenty of options for bringing material into Lustre to work with. Lustre is resolution-independent, and provides options to load image files (including digital video formats such as Windows AVI) from a local or network disk, or capture directly from video (including HD). Material can be auto-conformed and auto-captured using an EDL (cutlists and keycode are not supported). And for facilities already using other discreet systems, it can access Stone partitions directly (including support for soft clips).  However, once the material is conformed in, the options for checking or editing the sequence are somewhat limited. For starters, there is a "confidence check" feature to automatically compare a conformed sequence against a digitised offline video, but there is no option to sync Lustre’s playback with a video deck to accomplish the same thing more accurately. Another glaring omission is the lack of multiple timelines, such as found in most other editing systems, which is useful when conforming visual effects or other material separately from the scanned footage. In fact, Lustre could take a few leaves out of Fire‘s book in terms of editing. The editing system in Lustre is clearly a hang-over from its Colorfront incarnation, and it suffers for it. Many systems developers I’ve spoken to share the view that DI systems should not infringe upon the editing domain, and that DI systems should keep editing capabilities to a minimum. I have no doubt that this is a view propagated by colourists, who have no desire to learn a lot of complicated editing tools that they will never use. That’s fine, but Lustre is clearly designed to be an all-in-one solution, and as such, needs to be used by a number of other operators, such as editing or conforming assistants, who will use the editing functions almost exclusively. Many of them will have experience of NLE systems such as Fire, or those from Avid. On the plus side, it has a functional re-conforming engine, which is always a good thing.

Lustre includes a rudimentary, automated dust-busting module that is basic to the point of being useless: a much better option would have been to include a paint module instead to allow manual dust-busting and touch-ups. There is also a "text plate" option which allows keyed text to be superimposed over the image (which unfortunately requires a special control file to be written by hand to specify in and out points), which in practice could be used to superimpose any still image, such as a bug or watermark. Image repositioning is handled well though, with controls to adjust the size, position, and rotation of shots, independently from the grading layer if need-be.

But the real area where Lustre shines is colour grading. Lustre caters for both a linear (for video) and logarithmic (for film) colour architecture, and these are reflected in the available tools. For example, when working with video, there are controls to adjust gamma and gain, while film grading will provide you with controls for brightness and contrast. There is support for monitor calibration, as well as for 2D and 3D LUTs. Grading is accomplished using a 3-tiered approach, separating primary, selective (secondary) and curve-based grading (for changing hue vs. hue, lightness vs. hue, saturation vs. hue, and lightness vs. saturation). These are easy enough to control with a mouse, although the "Lustre Master System" package also comes with a "control surface", which allows for more tactile colour manipulation that experienced colourists may be more familiar with. The selective section allows grading using shapes such as circles or drawn-shapes, as well as a line shape which is useful for a number of different effects. There is also a decent keyer module as well as the ability to import saved mattes. All in all, Lustre boasts 12 selective channels, but in practice multiple shapes can be placed on each channel, so it is highly unlikely that this level will be reached in most situations. The included feature-tracker seems to work well; so well in fact, that it is possible to draw a selective shape around an object on the scene, and the tracker will track every point on the shape, effectively bringing automated rotoscoping capabilities to the system. There is a "bin" on the screen for saving grades (which are then visible as thumbnails), but I can’t help wondering if this could be improved somehow, perhaps allowing a grade "library" to be incorporated, like the materials library in 3DS MAX, or even the particles library in Combustion. There are also options for automatic match-grading, as well as the ability to do split-screens to compare grades.

Lustre has integrated the inclusion of certain effects, such as noise and grain reduction, as well as film process emulation and a defocus effect, into the pipeline, allowing them to be applied to selectives, effectively allowing these effects to be used as grading tools. However, after speaking to someone at GenArts, which make Sapphire Sparks for Discreet’s other systems, it seems that the architecture is not yet completely open to allow third-party effects just yet. I can only imagine what might be possible if Lustre is ever given the capability to integrate After Effects-compatible plugins into it the way that Combustion can.

As with most of Discreet’s systems, most of the parameters for grading, positioning and effects are fully animatable, so any number of creative effects are possible. There are options for retiming (speed-changing) footage and applying optical effects such as dissolves, but they are not really useful for anything beyond basic effects. The dissolve module is horrible to work with, and the retimer does not provide graph-based re-timing or allow you to change the retiming method used.

In terms of output, Lustre can render to digital image sequences (although there is not yet any support for HDR images), or output directly to SD or HD video. Rendering can be accomplished on dedicated rendering systems or run as a background process.

If the system suffers greatly from anything right now, it’s that it is too closed up. When using all of the built-in modules, it is tremendously efficient. However, it can be frustrating when needing to work with external systems, such as to create a speed-change effect, or to dustbust, as such changes need to be manually updated in Lustre’s single timeline. Filmlight’s Baselight on the other hand, provides a vertical timeline interface, with different operators, comments, and different versions of sots can be stacked on top of each other, which to my mind is much more suitable for DI productions that have a larger team, with multiple colourists, digital paint artists, editors and assistants. But if Lustre were to borrow more from Combustion and Smoke, it would be far superior to everything else on the market.

Lustre is one of the most expensive solutions on the market (different scaled packages are available), but for facilities with a large enough catalogue of work, it is probably a wise investment. It may well be that the most exciting features are the ones to come. One of the best things about Discreet as a company is that they constantly develop their product lines. Flame was impressive when it first came out, and the more recent Flame 9 is equally impressive. And eventually, producers may even steer clear of facilities that lack Lustre.

More information can be found at Autodesk’s website

The reviwer’s opinions are his own, and not affiliated with any third-party.

UPDATE: Discreet has announced that release 2.6 will address some of the issues with the editorial side. More details here.