View Full Version : VFXTalk.com Interviews Framestore-CFC Compositors & TD's for 'Troy'

July 19th, 2004, 04:44 AM
Could you please start off by introducing yourselves and giving us a little background information pertaining to your current job/role, and your background in the industry?

http://www.vfxtalk.com/feature/FCFC_Troy/AndrewWhitehurstLRG.jpgAndrew Whitehurst: Hi, I'm Andrew Whitehurst. I worked as a Senior Technical Director on 'Troy' at Framestore-CFC where I was mostly responsible for issues dealing with lighting and shading and running shots. Prior to Framestore-CFC I've worked on FX movies at other studios as well as commercials and some kids TV series having initially started out as a runner 6 years ago. I've always specialised in the lighting and technical direction side of things as I'm an appallingly bad character animator! I trained in fine art and then traditional film-making before finding I liked computer graphics.

http://www.vfxtalk.com/feature/FCFC_Troy/AndrewChapmanLRG.jpgAndrew Chapman: I'm from Australia, and after doing a Computing Science degree I started working as an R&D Developer for Animal Logic in Sydney. I moved to London to work for Framestore CFC a little over two years ago, and have been having an incredible time here. Recently I've been a Senior Technical Director working on Troy and before that the trailer and opening titles for Bond: Die Another Day.

http://www.vfxtalk.com/feature/FCFC_Troy/GavinToomeyLRG.jpgGavin Toomey: Hi, I was the 2d compositing supervisor on “Troy” at FCFC where I was responsible for our team of 8 compositors and 4 roto/paint artists . I've been working in the industry for 9 years starting as a runner for Framestore in 1995 following my graduation (studying film). After a few years of Matador paint/roto, I started designing shots with CFC's script driven compositing hardware. Apart from a small number of music promos and ads I've been lucky enough to work almost exclusively on feature films.

Approximately how many shots did FS-CFC complete for the film, and which sequences were you responsible for?

Andrew Whitehurst: I believe we did about 100+ shots divided between the armada, the Greek encampment and the beach battle. I worked on several armada and beach landing shots and also in developing our measured BRDF system for shading the sails. I guess I did about 15-20 shots in total.

Andrew Chapman: With a grand total of two, I think I worked on the fewest shots out of everyone on the team, but they were a couple of our trickiest – its not fair, I never get anything easy to do these days! Before running shots I was developing our crowd system, so I was kind of responsible for making sure the crowds worked well across all our shots too.

Gavin Toomey: 130 shots, As AndrewW has said we were responsible for a number of sequences including the armada shots and the beach battles/Greek encampment. I supervised all the 2d compositing work for
these shots and also composited a number of shots including the armada beach-landing and some of the flaming arrow filled battle scenes.




What was the most challenging shot to do, and why?

Andrew Whitehurst: I think the pull-out shot of the whole fleet was most challenging as it was one of the biggest shots in the film and it had to be delivered initially 6 months ahead of schedule for a teaser trailer. In fact once the trailer version was done we went back and completely re-did the shot again with improved tools for the film itself. The shot was good as it put the pipeline we developed to the test immediately and acted as a “proof of concept” that the direction we had taken for our pipeline was the correct one. Other complications in the shot were mainly tracking related. It's phenomenally hard to track the ocean as there is nothing static to get a firm track from. There were also issues with re-racking and varispeeding the footage which added extra hurdles. The lighting was pretty straight-forward, what was hard was that as we were running the shot we were continually improving the tools and so we were always finding that we had to upgrade to a newer version mid way through the shot development, but thanks to the great programmers we had this generally did not prove to be too stressful, though there were many very late nights as we approached deadline.

Andrew Chapman: My personal widow-maker was the shot referred to as the 'D-Day Shot'. This was a helicopter mounted camera flying down the beach as the ships roll in, panning up to reveal the armada and a beach full of thousands of soldiers setting up their encampments. There were a sprinkling of real extras on the beach (I think a few hundred) but they are lost amongst the crowd in the end. That shot used every animation clip, model and tool we had at our disposal, because it is a really tough job to fill a huge beach with almost idle CG people and have them look realistic. I think we pulled it off in the end – personally there's no way I can pick most of the CG guys from the real actors in that shot.

Gavin Toomey: One of the most challenging shots for the 2d team was the armada pull-out shot (used for the original trailer) which Andrew W. mentions. We had 2 reference ships in the plate and various tracking buoys. Even with these we still found it a challenge to get a track that locked all the CG ships into the plate, especially when we had applied a varispeed and stabilization to the original camera move. We graded the larger fg CG ships sitting in the scene using the 2 live action ships as invaluable reference by consistently matching the keylight/fill and the level of detail in the ships. We also matched live action water interaction and reflections to preserve the nuances of the live action vessels. As the ships pulled back to the horizon we needed to keep a realistic transition on the grading/density of these elements according to their particular depth in the scene. We also had to remove both excessive foam/water interaction from the propeller driven live action boats and speedboats/buoys scattering the plate - once they had served their purpose as vital tracking references.




When your team is faced with the challenge of creating a certain effects shot/scene...what are the typical work patterns you follow? And is the final result always as you originally envisioned it, or does the process often change and adapt as new ideas or challenges arise?

Andrew Whitehurst: We were continually improving our tools and methodologies right throughout the production. There would always be a new shader version or an improvement to the boat or crowd system so I think our methods for solving problems changed and adapted over the duration. The main problem we had with Troy was the amount of geometry we had to deal with at render-time. Each boat is very complicated and there are up to a 1000 on screen. We had to develop a very efficient method of only creating boats at precisely the time they were needed to be rendered and then immediately culling them from memory. In fact our RIB files that were written out by us contained practically no geometry at all, just a set of instructions on how to build a boat. Custom software was then called as the renderer encountered the area where the boat lay and a RIB archive for the boat was built on the fly. This kept the memory footprint as low as possible, though even then we were pushing the limits of our machines.

Andrew Chapman: The crews at Framestore CFC are quite eager to accept new techniques and technologies. I know a lot of companies that do job after job with the same tools and pipelines, but the guys here are always embracing new tools and methods to improve our workflows and the final results. Jobs tend to have a small R&D crew assigned initially to work out how things are going to be tackled. Once the problems are broken down a bit it becomes clear that certain things can be done with off-the-shelf or existing proprietary tools, and certain things are going to require some new tools to be written. Any new tools are either going to be developed by our R&D department or written by TD’s who are on the projects but also happen to be developers (like me!).

Gavin Toomey: A typical work pattern involves, initially, establishing a realistic schedule which enforces appropriate deadlines and encourages constructive decisions. While a certain flexibility should always exist it's vital that workflow pipelines and conventions are established in order to progress within the limited timeframe. While the process is always changing, I think you have to make a constant effort to envisage a final result while remaining aware that this goal can be in constant flux. It's always useful to have access to as many variables as possible and to maintain a certain amount of freedom but the proof is only visible as a result of clear decisions that achieve the agreed demands of a structured schedule.

What software was used for compositing and how many artists were involved on an average shot?

Andrew Whitehurst: Shake and some paint work in Matador.

Andrew Chapman: I just wanted to mention that we were also doing some look development and lighting within Shake for certain shots that were being difficult to light well in 3D. On average each frame we rendered from 3D contained up to 30 'arbitrary outs' (additional render passes generated at the same time as the main render image) and these could be used to reassemble or treat particular aspects of the look of our boats and people. With complex RenderMan shaders you're basically doing a lot of compositing within the shader, so its good to dump each phase of calculation out to a file so that the compositors have more flexibility. They can then either use the beauty images you have created, or make their own using the constituent parts.

Gavin Toomey: We used Shake for compositing and predominantly Commotion for roto/paint with some Matador just for painting. For some of the larger shots - such as the Armada pull back – The senior compositor on the shot had the assistance of 1 or 2 other compositors and 2 paint artists to look at addressing specific tasks that could be worked on in tandem. If possible we gave individual compositors responsibility for a series of similar shots with assistance from the roto team who also helped out with the lower-level compositing tasks where it was appropriate.

As Andrew Chapman mentions, because of the huge 3d renders that were sometimes involved, we received a large selection of 'arbitrary outs'. As a result we had a great deal of flexibility to adjust the lighting of our 3d elements in the comp without resorting to a re-render unless absolutely necessary for animation/tracking purposes. We generally reconstructed our beauty passes in order to tweak lighting of sails, hulls, shields and crew independently. We had to ensure that the compositors worked in parallel to maintain continuity of the armada “look” but were still flexible to tweak the lighting for particular shots.

What was the average time spent on a composite?

Gavin Toomey: A huge range really, depending on the type of shot. For the more routine green screen comps, the average was probably 2 to 3 days work but on the larger scale armada shots we could spend anything from 4 days to 5 or 6 weeks on a single shot. Some of the armada sequences filmed from the camera boat required heavy roto work for the rigging/bg characters. A single roto artist may spend 5 days cleaning up mattes, another 2 days removing tracking markers, and then the compositor another 5 days working the 3d elements into the scene. We found, as we established the “look” of a particular type of shot, or if we had good solid reference material shot on location, our compositing became a great deal more efficient.

On a project like Troy, how much RnD does Framestore-CFC do? Do you mainly make use of off-the-shelf tools or do you develop a lot of your own (in terms of both 2D and 3D)?

Andrew Whitehurst: We make a lot of our tools not to be unnecessarily complicated but just because a lot of off the shelf software doesn't do what we need it to.

Apart from the crowd system, which was Chapman's baby, we also did a lot of work on Image Based Lighting for this project as our supervisors Jon Thum and Ben Morris were able to shoot some great fish-eyes on set which we could use to get us 80% of the way to lighting a shot. We'd often find that a mixture of IBL and CG lighting worked best so were always trying to find the perfect balance. Oftentimes the HDR fish-eyes couldn't capture the intensity of the harsh direct sunlight so we'd use a CG light or artificially boost the HDR maps to get that hot sunlit look. What made lighting this film especially tough was that everything was in the harshest of sunlit conditions which is about as unforgiving as you can get when you're trying to match it. The whole armada sequence was shot on location and had very natural lighting which added an almost documentary feel to the shots. All of this lighting had to accomplished with custom shaders which we refined throughout. We also developed a system for measuring the reflectance of the sail fabric as none of the standard CG Bi-directional Reflectance Distribution Functions (BRDFs) really looked right, so we designed and built our own reflectometer and the tools needed to analyse the data we got out of it and then to convert that data into something that could get plugged into a shader.

There was a lot of pipeline R and D as I mentioned before because we were concerned about the load we were putting on the render boxes and we had to figure out a smarter way to only load geometry when it was needed. That was the crux of our whole pipeline, to keep it as lean as possible due to the amount of data we were throwing at our render boxes on every frame. A lot of effort was expended on getting that system efficient but still maintaining total control of the output result.

Andrew Chapman: Quite a lot of R&D is done, some of it in order to get the job done at all, some of it in order to get the job done quicker/smoother/smarter, and some of it done just to make all the separate pieces of a production pipeline fit together (I think of that as the 'glue' software). On Troy the major development blocks were the procedural boat generation system, the crowd system and the IBL lighting. There were also a lot of company-wide tools being developed, such as a graphical node based editor for complex render tasks involving many dependencies. The main reason for using custom software is not that the off-the-shelf packages don't do the job, but that they aren't flexible/customisable enough. Maya is a good example of an open package with a great API and everything accessible through MEL, attributes and connections that are visible to everyone if they go looking. If an existing tool isn't that open and flexible we will always want to write our own replacement, as we can't afford to get stuck in a bad situation with a 'black-box' solution that we have no control over.

Gavin Toomey: From a 2d perspective we relied on various plug-ins that were developed in-house for shake. One very useful tool enabled us to project textures in 3d workspace and another allowed us to import 3d tracks into the comp which proved extremely useful considering the 2d restrictions of Shake. We also relied on a number of standardised command-line tools that generally speeded up the 2d pipeline. Members of the 2d team also helped to prepare the Image Based Lighting images for the TD's and generally worked very closely alongside the 3d team during the look development stages.

We built Shake macros to recreate 3d beauty passes from the large selection of arb-out elements that were rendered but inevitably as the look developed we found these scripts were developing with every shot and we needed to keep certain variables quite flexible to tweak while also standardising our method sufficiently to share scripts amongst the team.




How did you R & D the flocking and crowd simulation software, and are they any details you can share with us on how it works?

Andrew Chapman: Without giving too much away or I'll get into trouble... It is a plug-in solution for both Maya and RenderMan, which has a single node for each character agent in the scene. The agents are drawn by default as OpenGL skeletons, which is the lightest possible representation that will still show you where he is and what he is doing. I'm a big proponent of having everything visible in your scene, with lots of interaction and feedback. Traditionally custom software was often run as command line utilities where you don't see the result until you do a render. To me that is just a huge gamble and often a waste of time – the more feedback and iterations people see of their work the better it will be in the end.

Within Maya you can create groups of agents and have them stick to the terrain and be influenced by flow fields, which are provided by an auxiliary plug-in written by Samy Ben-Rabah. This allows you to plot curves, arrows and points on your terrain for the agents to walk with, toward or away from. The agents play back animation clips that were produced from motion capture, often quite heavily blended and finessed by our animators. The system also has the ability to do some basic clip blending of its own, including modifying the playback speed, random offsets, etc.

At render time, a very minimal description of each agent (e.g. position, soldier type, animation clips, etc) goes into the RIB file as a RenderMan procedural call. Each procedural call has a bounding box, and RenderMan only calls it if it is within the frame and visible. The procedural knows about the animation clips and the geometry files, and can re-create fully textured and shaded soldiers on the fly, which always match up perfectly with the characters in the Maya plug-in. Doing the geometry generation procedurally like this was very efficient – a film res frame of a beach full of people would render in a matter of minutes (without props or boats in the scene). The RenderMan procedural was written by Hal Bertram.

When starting work on a really meaty shot, do you approach it as just another shot, or do you prefer to really understand the context and emotional state of it first?

Andrew Whitehurst: All the emotional and contextual content is worked out during previz. That's one of the main reasons for doing it. It gives the director the opportunity to see how our proposed shot works within his or her temp cut and to make changes before we get too locked into a certain way of doing the shot. Generally our shots were contained within a sequence and we had before and after shots on the Avid to check that everything went together OK. Once the plates are scanned and graded it's really a matter of trying not only to match the real lighting but also to enhance it a bit on the CG. Sometimes we took a little artistic license with the lighting to enhance the dramatic effect we were trying to get, but you're always walking a tightrope when you do that because if you go too far the CG no longer matches the plate and obviously that's a no-no.

Andrew Chapman: You should always know the context of your shot within the film, as it can affect how you do things without you realising or planning for it. On Troy for example this meant knowing the overall pitch and strategy of the battle so that all the people had the correct density, direction and intensity to flow from one shot to the next. If you just look at each shot in isolation you'll be forever tweaking things to match up with the shot next door, and it is a never ending process.

Gavin Toomey: As Andrew W. mentions, the dramatic context of the visual effect is usually established at the previs stage and we will be kept up to date with the avid edit as the shot footage is assembled. Often the development of the sequence edit will affect the vfx we create and vice-versa as the shots are realised in the context of the cut. It's often in the subtle detail that you can add to the emotional energy of a shot. The vfx elements should compliment the shot footage, not dominate. Without the physical restrictions of the film recording medium, any digital manipulation of a shot can completely reinterpret the way it will be psychologically perceived by an audience. We rely on updating the Avid cut to view our vfx in context of the sequence and, if possible, the soundtrack.


HIDDEN FLEET - Wireframe Plate

HIDDEN FLEET - Finished Plate

How do you tackle a sequence where you need to grade the shots to match those around it? If this is done by other people, how would you do it?

Andrew Whitehurst: We always did our CG lighting to a technical grade at least (a technical grade removes any colour shifts introduced by the developing and also sorts out a decent base exposure). Often we'd get a grade which Roger Pratt, the cinematographer had approved which was more of a creative grade and that gave us more of a feel for the final look. Even so there was more grading done after we'd supplied our final shots which bleached out a lot of the shots and desaturated them a fair bit. I think that was done across the board as people from other FX houses were saying the same thing when they saw their shots in the cinema for the first time. So really even when you've supplied your final you still don't know that's how it'll look when the film gets to the cinema.

Andrew Chapman: This is one aspect where I'm really looking forward to the increasing use of digital technologies, from filming right through to projection in the cinemas. It is heartbreaking to finesse the look of a shot until it is just perfect, then go to the cinema and see it crushed, colour shifted or bobbing and weaving through a cheap multiplex projector. But there is just something beautiful about the look of things captured on film though, and until that can be reproduced we still need to be able to deal with all these imprecise chemical and mechanical processes. In any case, if possible you really want to get a consistent tech grade as early as possible, or you're always fighting to get things sitting in nicely for all these individual shots when you could get it all working approximately once, then spend your time on other things.

Gavin Toomey: Our standard practice is to perform a technical grade on all the background plates in order that they maintain as much continuity as possible. We do this as soon as possible after receiving the scans to give the TD's and compers some hint as to the final grade. This grade is often changed as the post-production develops but it gives us a more tangible reference to work with. By the very nature of working with film you can never be certain that all prints will have matching grades so we have to allow for a certain amount of this sway in the presentation of our composites. In order to avoid these problems we aim to preserve as much colour range as possible.

What was unique about working on this picture that sets it aside from other film work?

Andrew Whitehurst: I've never worked on a film as epic as this before and the scale of everything in it was overwhelming. I'm also a big fan of Ancient Greek history and mythology so it was great to be able to work on a movie which is based on a subject I already had a huge passion for. This film also had the least stylised lighting of any movie I've worked on; it's almost documentary-like in places. This made our challenge much greater as there was little room for anything other than total naturalism in the lighting, which is always more difficult than a more “fantastic” look.

Andrew Chapman: Sheer volume. At each stage of the process we were struggling to push more geometry, animation and textures through the rendering pipeline. Through lots of procedural tools we were able to get it all through reasonably easily, but there's no way you could have done this job in a manual way – hand animating the crowds, or rendering a few boats at a time from a fleet because that is all that would be manageable within Maya.

Gavin Toomey: Working on a film of such epic proportions was a unique experience. While the narrative takes place in an environment of natural, dirty elements, the scale of the vfx work has the potential to look hyper real and exaggerated. The unique challenge for the team was to maintain an “unspectacular” persistence to the illusion despite the sheer spectacle of the events.



What rendering pipeline does FS-CFC use?

Andrew Whitehurst: Maya for modelling and animation, Liquid and MtoR for export to PRMan which is our renderer. We use Shake to composite. There are also plenty of custom scripts, plug-ins and programs which run within that framework to give us the effects we want. Houdini was used for some effects work.

Andrew Chapman: Yep, everything on the film side renders through Pixar's PRMan, except some Houdini stuff which is a mix of PRMan and its built-in renderer Mantra. The Maya renderer and Mental Ray are used quite a lot in the Commercials and Long Form departments. The thing with both Maya and RenderMan is that by default they may not give you as good/fast a result as some of the alternatives, but they are infinitely more flexible. Nobody in high-end work uses these tools straight out of the box, every production has myriad custom shaders, plug-ins and scripts. It is their ability to be customised which makes these tools so powerful. We have our own render farm system, and our own tools for piping data out of Maya into RenderMan. The RenderMan shaders tend to be mostly hand-crafted, though people will still sometimes use Slim to build shader networks.

Gavin Toomey: From a 2d perspective we ran our Shake renders on a farm of 50 dual proc machines through Pixars Alfred system. We had a team of render wranglers to ensure that our overnight renders were healthy.

What resolution were you working at?

Andrew Whitehurst: 2K, it was shot super 35 so we worked to a 1:1.66 safe area.

Andrew Chapman: Generally everything is done at 2K. As soon as you do the numbers and give a production back a quote with prices of dealing with 2K versus 4K they usually decide pretty quickly! To be honest there isn't a lot of difference in reality – as soon as the camera is moving even slightly the difference in detail is usually lost. When the cost of storage comes down, and the speed of the computers shifting the data around get faster (particularly in compositing) then it will become a more viable option. Also, you can sometimes do your 3D renders at 4K and down sample to 2K for compositing, which can give you nicer results when you're having aliasing problems, though we didn't need to take that step on Troy.

Gavin Toomey: As Andrew W has said, 2k, working to 1:66 safety. If we were working on a plate which required a great deal of transformation and could risk a softening in the process, we would scan these elements at 4K or over to retain as much information as possible.

Do you have onset high-speed compositing artists or is all your post work done in house?

Gavin Toomey: For Troy, All the compositing work was in-house. For other projects we have had operators on set to perform high-speed reference comps where appropriate.

Which fx shot did you feel ended up looking better than you had ever anticipated, and why?

Andrew Whitehurst: There's quite a few shots I like that we worked on. I'm still really pleased with the way the big pull-out shot of the fleet looks, especially as I'm only too aware of the heartache that went into making it. This was the shot that I thought might be a step too far, especially as we tackled it so early in the production. It was the most physically demanding shot to work on given the hours and late nights those of us on it had to devote to it, and it was only as we approached the final deadline that I began to see that it was working and eventually that it was working well. I had quite a few sleepless nights worrying about that shot!

There's also a smaller shot of boats coming into the beach at dusk which I really liked doing the lighting on. The plate was a beautiful sunset with a hazy/misty quality to it so the sunlight radiated throughout which made it technically a lot harder to light but aesthetically I really loved it. It was one of the few shots that wasn't in bright noon conditions and that gave us the opportunity to try something a little more gentle and subtle in the lighting.

Andrew Chapman: In most things you work on some things go well, and a few things are a terrible headache and just don't look good no matter what you do with them. One of our shots that was a bit of a worry for a long time, as it was a very different lighting situation to the others. Someone took a different approach and it suddenly looked great and fitted in well with all the others. The lesson to be learnt there is not to struggle on with things yourself when it is just not happening – the benefit of a fresh pair of eyes is priceless. Another shot I was amazed with when it came together is the first few seconds of the fleet pull-back shot Whitehurst mentioned above. The first few CG ships that come into frame are completely indistinguishable from the two real boats. After a point I think the shot breaks, not from anything we could have done, but just conceptually. No matter how good your CG is, you can still break the point of plausibility where everybody thinks “that HAS to be CG”.

Gavin Toomey: Viewing the complete film, I was very pleased with the armada pull back shot which was definitely the result of strong teamwork. While the initial trailer version involved a steep learning curve for everyone, good communication encouraged a very positive environment that balanced the workflow across the team constructively.



Coffee, red bull, or something else?

Andrew Whitehurst: I'm very English so it's Earl Grey tea about 20 times a day for me!

Andrew Chapman: I recently discovered the amazing properties of Red Bull during the Glastonbury Music Festival, so I'm looking forward to putting it to good use on our current project ;)

Gavin Toomey: Sensible hours, good food and red wine.

Thanks again for your time guys!! Top effort!! I hope we will speak again soon.
Cheers !! – Paul Moran :easystree

Framestore-CFC http://www.framestore-cfc.com/

July 19th, 2004, 06:46 AM
Thanks guys for that awsome insight!! This review has more pretty pics than the latest edition of Cinefex!!! Great to get a different POV's on the same topics.

Insperational stuff and congrats to Paul on getting this together!

July 19th, 2004, 12:23 PM
wowww !!!
good questions...great answers.

thanks everyone.


July 19th, 2004, 12:38 PM
Nice one. I worked on 'Troy' at MPC and it's good to read what was going on, and how, at other places.

Four roto/paint artists at Framestore-CFC? I was one of 24 at MPC!

So, Cinesite's contribution next?

July 19th, 2004, 07:07 PM
in BEACH ENCAMPMENT who asked for more sky?
Director or was it needed to fix Render scale?

July 19th, 2004, 11:06 PM
great information guys .thanks alot....

July 20th, 2004, 12:31 AM
Allways loved CFC style. When I was in London early this year I went to see their buildings as if they were part of the sightseeing of London! Great job.

July 20th, 2004, 06:03 AM
Excellent interview guys, much appreciated...great looking work!

September 21st, 2004, 11:50 AM
Why am just seeing this now two months after it was posted?? Man, I must be so out of it.. Great interview guys.. Loved to hear how it was working on that film.

September 21st, 2004, 06:07 PM
Why am just seeing this now two months after it was posted?? Man, I must be so out of it.. Great interview guys.. Loved to hear how it was working on that film.

LOL ...yup it was a purler interview series...just need to start doing a few more :)

October 10th, 2004, 11:24 PM
hey guys!! awesome to read about this stuff ;) haha

remember me? i was sitting next to u guys always high on caffeine and lots of redbull (yeah, andrew, it works!)...

hope everything is going well at fcfc still... if u ever come down to san francisco, gimme a shout! :D

afonso (:

Adeel Zafar
June 23rd, 2005, 07:51 AM
loved to wathc and read all opinions regarding their efforts on TROY !
wish best of lluck to CFC Framestore

Keep going!


July 21st, 2005, 01:10 AM
Hey, you guys are good !!
Keep it up !
If you guys ever happen to be in India, do let me know