Who you gonna call? Call Illoura, MPC, SPI, & Zero VFX

The original Slimer physical rig which now sits inside the halls of ILM, (Ghostbusters II).

Has anyone ever yelled out  ‘who you gonna call?’ without someone answering ‘Ghostbusters‘?  The iconic comedy from 1984 is such a classic that news of a reboot was met with internet outrage. Yet as the film found an audience, many felt it was great to see a new take and very different version with women in the lead roles.

While any film that works in the shadow of a classic is required to cover similar ground, there are quite different sequences in the new Ghostbusters, all of which were approached with the benefit of new digital vfx approaches.

The original Ghostbusters had special effects from  Richard Edlund, ex ILM, via Douglas Trumbull’s Entertainment Effects Group, which jointly became Boss Film Studios. The special effects at Boss (EEG) were done in just ten months with cloud tanks, mechanical animatronics and latex. ILM then lead the special effects on Ghostbusters 2 in 1989, along with Apogee, Stargate and VCE.

This time, under the VFX supervision of Pete Travers, rubber and wires has given way to volumetric deep comp pipelines and procedural Houdini photon beams from digitally replaced positron colliders. We spoke to the teams at Sony Pictures Imageworks, MPC, Zero VFX and Iloura about their ‘ungodly’ work.

The new digital Slimer

One new aspect of the latest film is the fight sequence in Times Square – which sees the Ghostbusters girls fight hand to hand with some fun new toys.

Stunt Viz and Sony Pictures Imageworks

The sequence with the Ghostbusters fighting one on one started out differently and much shorter. The origin of the Times Square fight sequence started with the stunt co-ordinator Walter Garcia. The veteran stuntman took it upon himself to produce a stunt viz sequence with the stunt women who would end up being the actual stunt doubles for the lead actresses. Garcia is extremely experienced having done complex fight choreography on X-men Apocalypse, Ant-Man, and most significantly with Ghostbuster’s director Paul Feig’s film Spy.

Working independently, Garcia produced an entire choreographed fight sequence complete with fairly impressive effects done in Adobe After Effects. An impressed Feig showed the stunt viz to visual effects supervisor Pete Travers who agreed, asking “wow, but where does this go in the movie?”  The sequence at that stage was not in the script at all.

Fieg ended up replacing a much smaller fight that had been scripted with some prohibition-era ghosts with the new much more dynamic hand to hand combat scene. In the new version of the script, the girls fight a complete array of ghosts from many different eras. It was “basically anyone who died in New York in the last few hundred years,” explained Daniel Kramer who headed the team at Sony Pictures Imageworks (SPI) that completed the fight.

The Times Square sequence grew to be almost half the total number of shots that SPI completed for the film. All the live action was all shot on greenscreen. As filmed, there were many more ghosts planned for the fight (see below) and thus there were many more actors in the background of each shot (all with LED tracking markers).

As it was decided in the edit to make the fight more personal, the SPI team had to remove all of these secondary background actors and roto the hero actresses from the background completely. This translated to everything but the actresses being digital in the final shots.

From the environment of Times Square to the smoke and of course the ghosts, everything was replaced. Given there is so much complex volumetrics surrounding the actors, SPI employed a deep compositing pipeline, embracing the depth compositing approach more completely than they had on any previous picture.

The original scene had many ghosts
The original scene had many ghosts…which completely changed in post.
The final shot - everything except the actresses are digital
The final shot. Everything except the actresses are digital (even their Proton packs).

Initial planning had SPI doing the ghost balloon sequence just prior to the new Times Square sequence, so the two scenes dovetailed in nicely in terms of pipeline and effects.

SPI’s Parade balloon sequence.

All the ghosts were complex and multilayered. They react to the environment and have their own emanations which manifested as an almost bioluminescence aura around them. In addition to this, the sequence had a large amount of smoke and the very strong light sources of the proton beams and contact lighting interacting between them. Each ghost had both an external presence or form as well as glowing internal skeletons or light sources that come and went during the shot. On top of all of this the team added additional sparks, dust, reflections, shadows and flares to make up a final shot.

The scenes were filmed twice, once with stunt women standing in for our heroes battling stuntmen on wire rigs and then filming was repeated again a couple of days later with the hero actresses. To make things even more complex, even then the portable ‘positron colliders’ or proton packs were just stand in prop packs on the day for safety reasons. All the proton packs were tracked to the actresses and added later in post.

SPI did not do all the ghosts in the sequence, as the “rock concert’ ghost seen in the fight sequences was the same asset as seen earlier in the film was provided by Iloura in Melbourne. Iloura had done all of the earlier Heavy Metal concert sequence.

Death Metal

Mayhem is a large and ominous all-CG ghost who inhabits a concert stadium and appears as part of a heavy metal stage show. The Iloura team handled a series of these more isolated sequences. The rock concert ghost needed to retain a ghostly, translucent look, but with subtle real-world detail such as skin texture and hair which needed to float and move as he flew through the air. Unlike other sequences there are long still shots of this ghost as he perches on the shoulders of Leslie Jones playing Patty Tolan. An internal glow was added to give Mayhem an inner ghostly glow on top of his skeleton and muscle system.


Other characters completed by Iloura include ghost rats and the Show Ghost featured in a sequence with Bill Murray’s character. Overall, the Iloura team completed 500 visual effects shots, led by Iloura VFX Supervisors Glenn Melenhorst and Andrew Hellen.

One of the key ghosts the Illoura team helped design was Gertrude, seen early on in the film when the team interact with the tour guide played by Zach Woods. In a pre-production test, potential VFX vendors were required to produce an animated ghost based on what they thought the characters should look like. Melenhorst said “the brief allowed us quite a bit of flexibility, and we developed some designs that evoked a creepy, ethereal look and feel, but still had enough charm and personality to work for the comedic aspects of the film.”



For Gertrude, a live-action/computer generated hybrid who ‘lives’ in the library, artists built a skeletal system, designed and added her clothes and lower body, and finished with a celestial aura. They then transformed her from ethereal beauty to skeletal ectoplasm-spewing-monster.


Another Iloura creation was the band of mirror ghouls, envisioned by the filmmakers as decomposed creatures recessed in a dark world who take on a suggestive form from afar, but becoming more legible as they approach the mirror surface and tried to escape. To achieve this vision, Iloura’s artists developed a number of looks for the translucent creatures from the recessed world, as well as more creepy, human-like forms that appear close to the surface. Cooler shading and lighting tones were utilized behind the glass, which shifted to warmer tones as the ghouls approach the surface; a technique that helped inform the audience of their other-worldly environment and their relationship to ours.

As the ghouls eventually smash through the mirrors and take on vaporous ethereal forms, CG warping and reflective bows were added to the mirror surfaces as well as practical and CG smashed glass to support the physicality of the event. This kind of seamless transition was key in this film as ghosts and ghouls frequently metamorphose from one form to another.

“Iloura played such a strong role in the design of our ghosts,” says Pete Travers, Sony Studios’ VFX Supervisor for Ghostbusters. “From the first test in pre-production which they hit out of the park, everything Iloura produced was not only of extremely high quality, but had such a strong aspect of creative design, in ways that we hadn’t asked or expressed. They went the extra mile, and it always paid off from our vantage point. Glenn and his team made it easy on us, defining the look of our ghostly beings like Gertrude, the Rock Concert Ghost and the Mirror Ghouls.”

New Toys

chipperPartly due to the Times Square fight sequence growing in size, another scene was added very late in production. There was a pickup shot done in LA where the girls test new weapons in an alleyway.  SPI, Iloura and Zero VFX all worked in that sequence. But, interestingly, the weapons such as the Ghost Chipper came from that original stunt viz video from fight choreographer Walter Garcia.

Aaron Sims Creative and designer Kyle Brown were involved in the prop designs of the new weapons. Brown took over the Ghost Chipper from the stunt viz.  It had a lot of iterations, including ideas of it being a sentry weapon that could unfold and stand on legs or tripod.

As an aside, the Proton pistol weapon shows up briefly in the weapon training montage slightly adapted from the original concept art. However, as real world toy companies get production artwork early, the actual toy gun one can buy from Amazon is modelled on this concept design (below right).


Aaron Sims Creative Company also provided concept art for the Times Square Ghosts.  Brown commented on Facebook that this was “a blast! It was totally blue sky, just some rough sketches trying to come up with different goofy ghost archetypes to populate the third act. The Flasher does show up for a split second, sans some more questionable features!”

Of all the great on-screen toys both new and old that are seen in the film, the hero prop of Ghostbusters must be the iconic Proton Packs first seen in the ‘don’t cross the beams’ 1984 version of the film. In the 2016 version, the Photon pack once again features prominently including in the Alley scene.

The Prop Packs
The Prop Packs

“The scene with the amped up proton pack really sticks out in my mind,” comments Zero VFX (LA), VFX supervisor Robert Nederhorst, who was extremely enthusiastic to work on the iconic photon packs. “The pack goes absolutely batshit crazy and blasts Melissa McCarthy all over the place, which is a really great moment,” he explains.

“One of our guys built a ‘goop system’ in just a couple of hours to make hot magma sludge drips down the melted bin,” Nederhorst relates. This was not actually a planned addition to the shot as the artists just thought it would look great. When Nederhorst saw it, he immediately loved it, but insisted that they go further, as the film was a comedy and subtly was not the order of the day.

“One of our Houdini guys just did that, and I saw it and said – that’s cool, do it a lot more, now go overboard, go nuts- this is a comedy, – so he added a lot more and made it red hot when it hits and drip to dark goop when it was done,” explains Nederhorst. “And that is now one of my favorite shots in our entire body of work for the film. It’s super cool.”


Zero VFX also got to focus on the Proton packs for another sequence when the new devices are first used in anger in the subway. On set for the subway, unlike in the Times Square sequence, the correct props were worn by the actresses but the end gun and the plasma rays were digital. The sequence is not only an action sequence but also very comedic as the first attempt to use the gun ‘falls short’ with a very limp ray, nothing like the iconic weapon of past films.

The beam was sometimes animated by hand and sometimes procedurally generated, but in both cases the beam was generated in Houdini. The prop gun was a large LED bar that had to be removed in post, but it provided the much needed contact lighting on the actresses. “Often times there’s a two-foot LED light beam attached to the barrel of the proton gun,” says Brian Drewes, co-founder of Zero VFX (Boston). “Removing it was a complicated challenge, especially when there was practical ectoplasm covering it. That’s where a lot of goop enhancements come to bear.”

A shocking replacement


The Ghostbusters come face to face with a subway ghost in a sequence that was initially intended to be achieved with enhancements to a live action actor in costume. As the sequence developed it was found to be easier and more flexible to keep the overall approach of the shot in terms of cinematography, but replace the ghost entirely and not enhance the actor (ex GG town troubadour, Dave Allen). To make this easier, most of the time the background was also replaced so that camera tracing was not an issue and rig removal did not need to be used to patch over the suspended actor’s extensive wire rig.

“The subway ghost was an awesome change for us,” explains Nederhorst. “It’s very early in the film, and represents the moment of discovery of the paranormal by Patty. It’s a great moment, which let us help sell the story for the filmmakers – that’s our entire vision in life at Zero VFX.”

In doing the subway, the team was presented with an additional challenge: to create a photorealistic character, something Zero VFX had not done before. “We had to really pull everything up by the bootstraps and build an entire character pipeline from nothing, while actively working on shots,” says Nederhorst.

“It was super-exciting to make this ghost, because it’s such a big moment in the film,” says Mike Warner, the CG and Animation Supervisor at  Zero VFX (LA). “We had to find the balance of the performance of our character with being dynamic, and meshing well with the other characters in the plate. We had a really great time exploring that performance and figuring out how to bring this guy to life.”

The ghost was fully keyframe animated but with visual reference to Allen’s acting choices on set combined with the added ‘untethered’ look of his arms and body movements. For the facial animation, the team wanted to distend the face so that Fred’s mouth would, for example, open far larger than it could in real life (it is basically double hinged).

The digital ghost could also do other things no actor could. For example, there was another shot when the ghost’s head turns almost all the way around 180 degrees to look at the Ghostbusters, adding to his creepiness.


The team nicknamed the ghost “Fred.” The brief was make Fred look cool, but to also play on the notion that he had been electrified, as he was the ghost of the first person to be put in the electric chair in New York. This meant adding things such as vapour or plasma.

“The production wanted a lot of electricity surrounding him,” explains Warner. We asked ourselves about how we could make it ethereal, and how we could make the photography and performance and character all look real together, with the smoke and electricity. We also thought a lot about who Fred was before he became a ghost, and built a lot of that visual identity into his character, and into the character of the effects…The ghost wasn’t just a vaporous apparition; he had a story we needed to tell. He was a bad dude, and we had to extend that story.”


Allen in costume on wire rig provided both the right eye lines and something for the actresses to play off of, but the live action photography could not capture the flowing nature of the ghostly costume. Had it been shot in a water tank, as was done with some of the original effects in the 80s, the director would have lost the interaction with the cast. Similarly, the team could have tried the old school technique of shooting Allen over cranked at a higher frame rate with a wind machine, but again this would loose the on set interaction that is so key to comedic timing. This restricted the on-set solution to wires and rigs and that simply did not produce a sufficiently threatening presence for Fred.

We love the original movie, of course! It’s one of my favorites and we’d play it from time to time to keep everyone fresh on the older techniques,” explains Nederhorst. One aspect that was very much inspired from the original ghosts of the 80’s was the luminescent quality and the semi-transparency of many of the characters.

“We wanted to maintain that and part of the lookdev for Fred was capturing that kind of translucent material: showing skin that isn’t skin, and bones under clothing that isn’t really there,” says Nederhorst. “We also explored how to create that electrical interference that Fred throws around the environment, and considered how it would light him up.”

These were all things introduced in the original film by characters such as the original library ghost. “We referenced those shots in terms of performance and timing,” relates Nederhorst. “We also looked at how the characters interacted with Bill Murray and the gang. How did the librarian ghost scare them in terms of timing? How did Slimer move? Those were questions we asked ourselves. It was super fun to look for those details, and to try to stay true to the original tone while moving it to the next step.”


The translucency was accomplished with a combination of lighting passes and compositing. The team knew that they had to see the skeleton from time to time, depending on the lighting interactions, so they rendered a ‘photographically real’ ghost with skin and everything fully textured. They also rendered out a skeleton with fully textured and displacement maps, affording them control over the level of translucency during the compositing process.

Fred’s beard was groomed in Houdini and rendered with the character to get proper shadowing and lighting with V-Ray. Zero VFX used nCloth for the prison garments and kept them in Maya as a base part of the character. They cached this through Alembic to send it to Houdini. nCloth informed the motion of all the Houdini-based effects that were generated. “It ended up being a pretty complex path system, going from Maya out to Houdini and then back into Maya for V-Ray for the lighting rendering,” explains Warner. “Thankfully we had a great comp team who put it all together!”


Trying to match the feeling of cloth and making it feel like Fred was wearing his grey prison suit was one thing, but the team also had to think about how to make a grey prison suit both photo-real and translucent.

“We went so far as to look right down into the details of cotton fabric, and when you do that you see little details like the piling that happens that gives a little bit of texture – we created what we called a ‘treads pass’ or ‘strings pass’ for some of the really close up shots. That gave us that extra bit of texture detail on that pyjama prison suit”.

Compositing supervisor Kurt Lawson built a set-up pipeline that was robust and allowed for maximum flexibility. The team could react quickly to feedback with more or less translucency in comp instead of needing to revisit the CG lighting and re-rendering everything.


“We did a great job of locating smart places to structuralize pathways for passes and images and built a bespoke toolset for compositors to stand up shots fairly quickly,” Drewes commented. “Efficiency is imperative, especially with the number of elements and passes going through each character. This system allowed us to get a version 1 rather quickly, mostly for internal use, but also for external use if necessary.”

Nederhorst also feels like this is a key reason for why Zero VFX manages to work effectively and have extremely small amounts of overtime. The company has an overall approach of quick turnaround and being willing to show a client a mock up or temp as quickly as possible and then to continue to iterate quickly. Unlike some companies, Zero VFX sets a premium on holding render times down and doing everything possible to allow the maximum amount of creative iterations, but only as a means to loop back to the client quickly. The company avoids making things unnecessarily complex and yet they encourage an iterative loop to a creative solution above all else.


The sequence ends with our heros escaping not only Fred but a speeding Subway train. Fred gets hit by the train, and with each carriage hit, he explode with ectoplasm. The train was set up on set with the crew pushing a dolly with a large light on it, to simulate the on coming train. Perhaps understandably the crew could not push a dolly as fast as a NY subway train so the Zero VFX team needed to fully remove this element and completely replace the train, the tunnel and then add in Fred trapped inside.

Who ya going to destroy? – New York!

New York once again suffers at the hands of a team of visual effects artists. But it is worth noting that there was a ‘time” theme connecting the events in the first film to the 2016 film that was scaled back in the final edit. This was also the case with Rowan North’s ( Neil Casey ~ Chris Hemsworth’s) disco dance sequence, partially seen in the feature film end credits. This sequence was fully produced by SPI and will now be offered in an extended DVD-Download version of the film.

MPC provided the largest and most expansive visual effects for the film’s final showdown.  We asked MPC VFX Supervisor Dave Seager (DS) about how the MPC team handled the large scale destruction and the various creative links to the special effects of the original films.

fxg: Can you discuss the building destruction at the end especially how much of the surrounding builds or city were digital?

DS: For the finale, we were called upon to recreate Times Square of the 1970s.  The Ghostbusters were filmed on a backlot green screen set that included streets, sidewalks, cars, and the first floor facade of the Mercado Hotel.  This meant that we needed recreate the majority of Times Square digitally using photos from the 1970s.  Additionally, we needed to add the fictional Mercado Hotel in the place of the Paramount Building.  The Mercado needed to be modelled to a high level of detail because the giant Rowan ghost grows to his monster size within the building and eventually breaks his way out.

For this effect we used MPC’s proprietary rigid body destruction tool Kali.  The challenge was that Rowan’s actions were intentionally exaggerated to give him a more cartoony feel since he evolves form a 2D character into 3D.  Therefore the physically based rigid body simulations of the destruction would not work well with the character’s non-physically accurate motion.  More specifically, Rowan would move rather fast for his size, causing the broken sections of the building to travel at extreme velocities.  This meant that we needed to compensate in the simulation to try and get more realistic speeds on the broken building pieces even in cases where Rowan was moving quickly.

MPC work was supervised by Dave Seager.


fxg:  There is a great homage to cloud tanks with the sky above the city, can you discuss that shot and how you achieved it?

DS: From the start of the project the director, Paul Feig, was interested in having the ghost cloud that forms above the Mercado Hotel looking similar to classic cloud tank effects.  We spent a great deal of time looking at classic films such as Close Encouters of the Third Kind and the 1984 version of Ghostbusters.  In the end we used Flowline to generate a highly customized vortex motion and did the final volumetric render in Renderman.


fxg: Could we get an idea of the layering of a ghost? Was there a recipe of visual layers that made up a typical ghost?

DS: The final look of the ghosts was created using a variety of rendered layers.

In many cases these layers were combined in different methods to achieve a level of variety to the looks of the ghosts.  These ghost looks varied from corporeal to ethereal, but they just used the same rendered elements in different ways.  Every ghost had an anatomical base that included the skeleton and some limited internal organs inside of a traditionally skinned character.

Often these ghosts would have a draped robe or tattered cloth that was simulated using conventional cloth solvers.  The final pieces to the ghosts include a combination of fluid and particle simulations that created the ghost trail as they flew through the world.

fxg:  Was this a deep pipeline? 

DS: As a standard, MPC uses a RenderMan deep compositing pipeline. We find that it gives us a great deal of flexibility when compositing effects such as smoke or other semi-transparent elements. Rigid body renders such as buildings, vehicles, or characters are also rendered with Deeps, but this is done primarily to help integrate them with FX renders.

fxg: Could you discuss the photon beams and the notion of matching to the 1984 style look please?

DS: The recreation of the proton streams was one of the most rewarding development efforts.  The look, the motion, and the sounds of the proton streams are pop culture icons.  All of the major elements of the proton streams from the 1984 film are present in our modern effect.  However, these elements were given an upgrade.

An example is that in the 1984 proton stream there is a strong red/orange glow within the core of the beam.  For our effect we opted for treating the glow as illuminated vapor coming from the beam.  We generated a fluid simulation of gas coming off of the beam and lit it to the scene.  This gave us some very interesting motion and texture as the proton stream whips around the rooms.  It gave the proton stream a more physically and photographically motivated look, while at the same time staying true to the original film.

fxg: How many shots did MPC do in all and did any particular challenge stand out? 

DS: MPC completed approximately 250 shots for Ghostbusters. One of our biggest challenges was trying to sell the scale of the final monster form of Rowan.  The design called for a very simplistic cartoon inspired body with very minimal texture detail.

Typically, one of the biggest tools in making something feel big is to give it very small geometric and textural details.  We were unable to add these types of small details, so we had to find other areas to emphasize his scale.

Firstly, we did a lot of work on his muscle and fat system in order to try and make it look like he had the appropriate amount of mass.  Secondly, we covered Rowan in a cloth as if he were a gigantic ghost in a sheet.  The simulation of this cloth proved very challenging because we found that most out of the box simulation solutions made him look as if he were 8 feet tall rather than hundreds of feet tall.  This required us to manipulate the simulation settings to try and make the cloth feel gigantic and heavy.  In the end, the wrinkle details and how they moved, proved to be our best tool in making him feel monstrous.