AcidGFX Logo
Download a sample DVD
Pilgrim Stuff/R2 Review
This is my review of Pilgrim R2 Pro 3d as posted on VJ Central here

The best things come to those who wait - isn't that what "they" say? Pilgrim R2 has been a *long* time in the coming, but as it turns out, for all the right reasons. I've been tinkering with R2 on and off since the beta-testing started last year. I've found myself swaying between R1 and R2 for day-to-day stuff - R1 because of its familiarity, R2 because of what it can do. It's only in the last couple of months since the latest version of R2 that I've actually *really* started to dig a bit deeper as to what R2 is capable of and only in the last couple of weeks have I used it for the first time on a paying job...and…it rocks.

R1 got itself a solid following of competent, dedicated users from a lot of different visual backgrounds. The authors have listened long and hard to their users' (endless) requests, complaints and suggestions and for the most part they've nearly all been implemented (or improved upon).

The end result is actually quite hard to define since it's more of a realtime 2d/3d video processing/ compositing platform you can adapt to suit your needs rather than "A vj package which does a/b/c only". For those who *still* demand more, there's even freeframe filter support and a filter SDK so you can write your own plugins in Visual Studio 6 or above.

R2 Comes in three different flavours: Mixer, Pro and Pro 3d, everything in this document refers to the Pro 3d version.

There's no way I can feasibly sum up a year+ worth of updates, forum discussions etc in a few paragraphs so I'll try and run through the most notable improvements.

The interface is much improved over R1's - the layout is still roughly the same but there's none of the randomly vanishing buttons with which some R1 users were all too familiar. The interface generally seems pretty stable and there have been some well thought out extra addons to make more detailed live work easier.

Midi seems largely unchanged - it worked before and still does. Output resolutions have been increased to include HD resolutions too for good measure. There's also now a built-in backup system for projects - I've not had to rely on it 100% yet but it has proven useful on a couple of occasions already so gets the thumbs up so far.

The capture device manager seems to also have been improved in R2. It's not something I made much use of in R1 since my hardware wasn't that happy with the setup but it seems to work pretty well now and allow syou to use live capture feeds as textures on 3d objects or indeed anywhere else you would use any other type of clip.

Whilst I don't really do enough to comment on their implementation, Quicktime and Flash files are now also supported as import formats.

Reflections and water surfaces

The 2d system
The Main part of the UI is the layer grid which comprises a number of "channels" - basically compositing layers like you'd have in a video package, but arranged vertically. You can assign transfer modes to each of these channels (e.g. additive or multiply) and also trigger which layers are currently active - thus giving you the output composition.

Layers are the basic building block of Pilgrim projects and a layer can be either an image,  video clip, 3d scene capture feed. Once you get beyond the basic mechanics of triggering layers the fun begins. You then have 4 separate effects chains you can apply to any one layer and this is where the processing work actually happens.

To keep the processing structure managable yet flexible the authors have introduced a node-tree structure remeniscent of shake's node editor or maya's "hypershade" editor. This lets you combine the basic compositing building blocks (i.e. colour & channel proceessing, keying, distortion etc) as you see fit. Each node has its own set of animatable parameters and if you do delve into the SDK, this is where your creations would end up (i.e. as custom nodes). You have access to all four built in texture channels from the node editor so rather than just being "an effect" applied to a video clip, you can create proper video compositions in real time. It's here that the hard work put in by the authors really starts to show its strengths.

This is way harder to explain than actually do so whilst might sound complex, in actual it fact means you can quickly build video processing structures limited only by your imagination and available processing power

Layer effects node-based structure

To get you going there's a full selection of pre-built node structures (e.g. "glow", "timelag" etc) which you can apply to layers without having to understand the workings, and you can start to tinker with them in your own time.

Just for good measure they've also replaced the old fade-in/fade-out for layers with a transitions system which uses the exact same node structures as the effects - so basically you can build your own transitions too if the mood takes you, or just use the new selection of built in ones.

3d system
Pilgrim implements pretty much the full range of DX9 3d toys - basically making available to you the same sort of visual effects as you're likely to see in any decent 3d game. You can import 3ds and direct x models and textures and apply deformations and other processing in real time as well as generating other types of 3d object (tunnels and morph objects). The model processing is once again handled by a node tree interface like the filters and transitions.

There was access to a number of these mesh processing features in R1 but the control and interaction of them was nowhere as near a refined structure as is presented by R2.

The particle systems have had a makeover too with a simplified interface and more animatable parameters. It just seems a bit easier to get up and running with now too which has got to be a good thing J

It may sound geeky but my all time favourite new feature in R2 and what made it for me is the shader system, in particular the addition of various camera tricks to enable real-time reflections, refractions and shadows on 3d surfaces. Now the concept of acheiving reflections using extra cameras and environment mapping won't be unfamiliar to anyone who has used a 3d package - but to have these tools which work in real time on a £700 laptop is awesome. You can set up objects to generate flat mirrors or cube environment maps, and assign those maps as textures onto most of the available shader types.

Enough glowiness for all

There's a full set of DX8 and DX9 compatible shaders, all the basics for mapping textures every which way, and then a number of "special" shader types, such as the flat mirror/cube reflection/refraction types and a few other interesting ones like subsurface scattering and a water surface (awesome!). The different shader types can be mixed and matched and layered as high as your gpu will take. Many of the dx9 shader types also suppport bump/normal-mapping.

In practise this allows you to build 3d scenes where each surface has exactly the properties you need from it (i.e. receive shadows, reflect surroundings etc). If you find you've still got gpu cycles spare you can even splash out on realtime depth of field effects.

If  you feel like delving deeper there is also access to the shading scripts used, which means you can also write your own shaders.

Procedural particle shaders

According to the authors R2 is developed/tested on a 1.7ghz athlon - I've found basic projects run fine on that sort of spec of machine as long as your 3d card is up to scratch but if you want to really open it up you might need something a bit more powerful.

R2 seems to be about flexibility - any one task you can probably accomplish a number of different ways depending on your requirements and setup. For instance if your 3d card and hard disks aren't quite as fast as you'd like but you've got a beefy cpu you could use a codec like divx (comparatively cpu hungry but minimal disk usage) and move much of the video processing to the cpu to leave your card resources free for other tasks - the inverse also applies - with a suitably kickass 3d card you can probably get away with quite a modest cpu - hence the 1.7ghz test machine I suspect. Either way it's nice to have the options.

There's no backwards compatibility to R1 layers/projects - however, the authors did the right thing here. What they had planned by way of upgrades needed a clean slate and whilst  it's a bit of a pain not to be able to use any of your old R1 stuff, once you get into R2 it all becomes irrelevant anyhow ;-)

It still crashes randomly but nowhere near as much as it used to and this does seem to be largely dependent on which video codecs you use among other things. What's also important is that normally crashes occur during building of projects rather than playback. This was the same with R1 - it would crash from time to time whilst building but only ever once did it live live (where it would normally run for 5-10hrs uninterrupted). The authors have also implemented a bug-tracking system which has noticably helped eliminate gremlins.

The recording/non-live renderer is still buggy imo - based on my own testing and forum posts from other users, anything which reacts to sound or realtime input will probably not render correctly (including mid-render setting changes). It *is* possible to use the renderer to good effect but I've only had good results with layer combinations which require no external input to function. That said - if your layers fulfil those criteria it works fine and the image quality is superb (as you would expect), but it's still a bit hit and miss.

Although the interface has been thoroughly revised and there have been countless improvements overall including some very nice touches like the "column flooding" and the quick settings for layers, I still have some quite specific (minor) gripes with it.

The tab ordering of data entry boxes still seems to be completely random - this makes entering data a mouse+keyboard operation instead of just being able to tab to the next data-entry box with the keyboard as you would in max/maya/photoshop etc. It may seem trivial but it's a usability issue you encounter on most dialogue-boxes which means it gets tedious pretty quickly if you spend a lot of time in the program.

The scene navigator lists all the 3d-entities and provides your access top the animatable parameters. However when you select a different layer it disappears, and although there are shortcut keys to bring it back, the times I find I try to use them e.g. having just entered a value and perhaps want to animate that value, the keyboard shortcuts don't work - they only seem to work  when the mouse focus is on the main layer grid. The various windows also don't really scale very well which can make for a cluttered UI, especially on a small screen (e.g. 1024x768 laptop screen).

Please note
Re the downsides above - given the choice I'd rather the authors had spent their time on the *engine* (which they did) rather than having a pretty interface but the basics don't work - so please be aware this is just me being picky. Also I seem to be the only one on the forums who moans about the interface so please take my "cons" with a pinch of salt - they are just my opinions ;)


R2 isn't an upgrade, it's an evolutionary leap. If you just want to trigger clips then please move along, there's nothing to see here, otherwise - it's well and truly the dog's danglies displayed as a node-tree. Download the demo and give it a go! (tip - do the tutorials!)

There's more screenshots of my recent R2 tinkerings here and a video featuring much of the same content here (7min, 45mb, Divx5)

Our Best Sellers