About My Blog

This blog will follow the creation of my Visual Effects work and eventually my Visual Effects Show-Reel.

Wednesday, 22 October 2014

Horrors Of The Skies VFX Project- Making Of Shot 1

Hi Guys,

I have been working hard on the project for the past week, Wanted to do another update, The next update after this will be once the Project is Finished.

Right now I am working on Shot 2&3 but Shot 1 is Final, Grade might be a WIP. This shot consisted of the Hero plane (The Sopwith Camel), getting its wing shot out by a German BiPlane casuing an explosion and the engine to catch fire. I used the Animation from the Previs as a starting point, I used the High Res Sopwith Asset which I rigged to control the flaps and add flapping to the wings through FFD's.
The Destruction of the wing is done with Rayfire, However I did not want a bog standard RBD Sim, I wanted it to look like it was breaking from the stress from the explosion and Bend then Snap. The wing was manually cut into Face Fragments, then using Activation by Geo to get the progressive destruction of the wing over 10 frames, This looks alot more realistic then it just exploding. I then used a Noise Modifier to bend the fragments as they break so the flap about as they fly off, so they look more like wings breaking rather than concrete or rock, I also used a few FFDs to have the wing bend as an flap to again make it look less like rock and more realistic.

The Fume FX sims were really good fun, from the outset I know if a Biplane had its wing shot out it would not explode, but I needed it to for this shot to not only make it more entertaining but also more engaging to show the plane is falling out the sky and cannot recover.

Alot of my work on WarThunder For Gamescom 2014 came in useful here, especially doing the engine fire, this sim here is no where near as Hires/quality as the one I did for the trailer because I dont have the time on this short 1 month deadline, plus on WarThunder I was doing FX, on this I am doing everything. The Wing Explosion I wanted it to be traily, but not too much, in the end I used Colour caching to get the diffrent colored smoke trails in the same grid as the main explosion. The Engine Fire was in a seprate grid, using wind to fake its velocity as the Grid was parented to the plane to save not having a huge Fume Grid.

For the Clouds I used Afterburn, however these got feathed quite a lot in comp as I used Cloud Pictures on Cards as it was a lot quicker, ideally I would of wanted to spend the time getting the AfterBurn clouds working well. Comp Was Done in AE.


Thanks For Looking

Saturday, 11 October 2014

Project HOTS- Week 02

Hi Guys,

So Week 2 has been extremely busy and its as I write this that Production on Shot 01 is in full swing, I have less that 1 month now to finish and hope to be moving onto Shot 2 tonight.

I planned to start production on Monday this week, Late Sunday night I was doing some clouds RnD when I pushed my PC to far and crashed it (Which is normal) when I restarted I noticed parts of my Hard Drive had gone corrupt, I'd had DNS Failure even when trying to flush it, which took some of my software Licenses down with it. So after all of Monday in extreme anger and anxiousness I decided to do a Clean Re-install so I could get back working ASAP, I spent 14 Hours on Tuesday re-installing everything so I could start production on Wednesday. Already that was two wasted days.

Luckily I had been doing some More RnD over the weekend which was mainly based on doing Clouds and some early Fire tests for the Engine Fire of the Sopwith Camel Plane also some Flak Scene setups. I have some RnD down here for you to see, just a note Project HOTS stands for Horrors Of The Skies.

For Clouds I started thinking my Cloud setup was going to be good enough, but after aactuallytesting it in production it was clear it was too slow and still needed a lot of work. But Here are some tests using my Cloud setup, its based using PFlow and Standard Max Shaders.

Facing Particles and Spheres were my Inital Ideas.

I turned to Sitni Sati's Afterburn which I have never used before and heres some of the RnD I did, I had under 1 day to learn it because I had to get started on Production. Its really fast and pretty easy to setup, its all about looking at reference and messing with the settings! I think its the best tool for clouds in Max right now, Fume FX would be amazing but I dont have time to do Cloud Sims for big cloudscapes plus the long render times.

So now its time to work on Shot 1 which you can see above here ^^.

Lighting And Rendering For Shot 01


                                                   Early Fire RnD

Sorry for this brief post, but am very busy, Shot 1 is supposed to be the easiest shot and I have under 4 weeks left!! But its Fun and Exciting Times!!

Thursday, 2 October 2014

VFX Project- HOTS- Update 01

The Animatic Of the Final Shot (Plane Crash In North Sea)

Hi Guys!

So right now I'm working on a VFX Sequence about WW1 Dogfights, its a kinda freelance project for my old project, its 7 Shots- 30 Seconds- Under 2 Months Deadline- And its for Project with the aim to be as close to photoreal quality as possible.
Catch is, theres only one artist doing the whole gig. That's Me!
Its day 3 right now and already I've done the shot list (Lots of drafts also!), The Previs/Animatic (See Above) and also Model Prep of the hero plane the Sopwith Camel and full texture painting and UVing for the asset! Been a one man band I really mean only one man, I have no other artists working on this, I started with a brief and a Lowres Biplane asset I made in college last year.
The Sequence is no doubt FX heavy which makes it even more complex, but even more awesome, all 7 Shots are FX heavy, I have about 1 week for pre production then less than 1 week per shot, the Final shot involving the plane crash is going to be extremely complex and will take the majority of production time. I feel at an advantage been an FX Artist because I am able to do FX, and because of that have grounded skills in Lighting, Rendering, Animation & Compositing, so this project will be pushing the bar on them skills and really pushing the bar on my FX Skills.
Because of just finishing some FX jobs, I've now been able to expand computing power, also upgrade to V-Ray 3.0 and be able to use an Alembic Pipeline on this project. For the destruction FX I will be heavily relying on nCloth in Maya so Alembic is going to help there, also RealFlow for the Final Shot (Fluid Simulation/ Big White Water Splash!) now supports Alembic. And Ofcourse V-Ray 3.0s Alembic Proxy is amazing and will be used heavily not only for assets from scene to scene, but also for Rendering RealFlow Particles for the big Final Shot! I'm also looking into maybe using Houdini for some of the Fluids, I plan to do a RnD to see if the pipeline is quick enough from Houdini to Max, Via Krakatoa Magma, Frost, and V-Ray Alembic. Via PRT and Alembic ROPs
I fully prepped my Lowres Sopwith Camel (BiPlane) Asset into a Hires Asset, and re UV'd and fully Textured it within 8 hours! I learnt texturing before I went to college  and use it now and again but not much now as an FX Artist. I used Mudbox, 3DS Max, Maya & Photoshop for this, I've not used Maya for a while!
UV'ing and Model Prep In Maya & Mudbox

 Final Asset In Maya With All Maps Painted Within 6 Hours
Final Rig Done In 3DS Max
Some Time Lapse Screen Captures of the Texturing and Rigging Process:
I don't think Modelling and Texturing is for me, but I had to do it and make it the best I could with my limited knowledge in these areas! Can't Wait to Destroy this Asset!
As you will see below I have my own cloud solution for the project which is very fast and again I created it lone wolf!! In the coming days I will be doing Look Dev for the Planes and environments, Preparing some Preset FX scenes to use for Foundations in Production which starts next week, and doing lots more research and starting some if I have time FX RnD.  I'm taking constant Notes, Screenshots and ScreenCaptures of my working so I can share how I solved problems and show how I got on with the sequence. I plan to do updates for the project once a week for now, and then more as the Deadline on November the 9th Closes in (Project is for Remembrance Day).

I feel pretty crazy when I think about doing the whole thing alone, but I'm up for a challenge!!

I hope you will enjoy watching the project been created over the coming weeks!


Thanks For Looking!



Sunday, 28 September 2014

Cloud System RnD- 3DS Max, No Plugins


Hi Guys,

Just wanted to make this small post, as a side project I really wanted a quick way to get good looking clouds in 3DS Max, without any Plugins. I came up with this which is still a WIP right now.
Its using all Max Shaders and PFlow only, However since I have just upgraded to V-Ray 3.0 I wanted to test the speed with the Progressive Renderer. I made little demo video showing the un edited render speed also. These are not meant for hero clouds, for that I would go down the FumeFX route but for clouds that are gonna be heavily blurred in comp anyway they work very well. I still need to work on the shaders to get more scattering/Translucency to them.

Thanks For Looking!

Sunday, 21 September 2014

Kaiju Breach FX Project Shot

 This Project Shot has been in the works since June, there's been 3 iterations of, this is the third! I started working on this before I worked for Platige/Juice on War Thunder, However it was really low res as I was just testing at this stage. Then when I finished on WarThunder I started working on the second iteration and even did the first renders, but it was not looking good enough, so I decided once I got more free time I would start from scratch and make it the best I could, I don't settle for second best!
I wanted to create my most ambitious FX Fluid Shot based on a Kaiju Breaching from the water, Godzilla/Pacific Rim Style. Igor Zanic also did a shot like this last year. Also now with RealFlow RenderKit for 3DS Max 2014 (RFRK) I can now render Mist, which adds so much realism to Fluid Simulations (more below on mist). This is also my first ever project using Houdini (Houdini was not used for final Fluid Sims). I was aiming to get as Photo-Real as Possible, to create a believable FX Shot.
First Iteration, Very Low Res, Originally From Early June (2014)
This shot had quite a few problems to overcome, I'll do my best to explain them! 
The Kaiju is Modelled and Rigged by Sean Buck & Robert Campbell, I've had this rig since last year and wanted to wait until I was good enough to do a Kaiju Breach Shot. The animation I did in Maya, its really simple, just the Kaiju breaching through the surface, not jumping though, kinda whish I had him jumping and splashing back down.
The Kaiju Mesh itself has a few problems that caused Flickering Particles as the Kaiju passes through the surface, this was due to bad iso surfacing, LuisM from Next Limit released a really good tutorial on working with different iso surfaces, I did some RnD using Shell Mode however the results were not as good, so after running several low res sims under different cell size's I decided to stick with Solid inside as the flickering was only there for a few frames and on the end of the sim, so I decided to shorten the sim as a compromise. I found out its due to the skinning in Maya, so If I use this again I would need to look into re skinning the rig. The texture for this I quickly painted in Mudbox.
Flashing Particles On Breach
In RealFlow I setup the Hybrido Core Fluid and after a previous tests I really wanted to get an Epic Splash, To get the really nice push and displacement of the core fluid theres no way of relying on the Kaiju's collision only, so I used Bounded Noise and Vorticity Boost, the Vorticity Boost makes a huge difference and is one of the most valuable settings in RealFlow Hybrido. I used 25.0 which made a huge difference, the Bounded Noise was animated to be high value as the Kaiju Breaches and then fade out over the rest of the sim, these combined work really well for a realistic splash in the Core Fluid. I really liked the look of the Splashes done by ILM for Battleship when the Stingers first breach out the water, Ofcouse there sims are much more detailed and complex!
Left is with no Bounded Noise and no Vorticity Boost, Right Is With animated Noise, and Vorticity Boost of 25.0
Next Problem was again related to the bad Kaiju geo, because I was using a KVolume Daemon as most cases you don't want the Fluid Colliding with the container sides, But because of the bad Iso Surface it was causing Fluid to slowly escape, so with the help of Particle Sampling and (Later on in Max and Comp Hiding the edges) I got over this problem.
See Particles Escaping, Overcome with Particle Sampling (New In RF2014)
The Final Core Fluid Simm contained around 30 Million Particles, Took 12-13 Hours on my PC (I7 8 Thread, 16GB Ram, GTX660), My PC is quite slow in terms of FX work, Soon to change! Next was onto the Secondary Sims, Splash was first. I Originally tested using FLIP For Splash, but I changed to Ballistic (Dumb) because I wanted a more dense splash sim, The splash sim uses a stronger Gravity Daemon than the Core Fluid to avoid the splash flying off and looking fake. For the Foam I used the new Bubbles Emitter in RF2014, its really nice and within a bit of tweaking it created really nice foam.
Ballistic Splash Sim
 FLIP Based Bubbles/Foam Sim
Next for the Mist sim I tried using the splash to drive the mist, however it got to dense and started to look like smoke, also the sim times per frame were awful, so I setup another splash emitter that was really low res only siming around 30-40,000 Particles, this was then used to drive the mist sim with High Dissipation. Mist adds so much realism as it simulates when high speed water droplets (Splash/Spary) Particles break up in the air. The Mist sim turned out nice, however I wanted to try something I saw on the Odforce Forums, so I exported the mist to Houdini via VDB, then emitted more mist and re exported it as VDB. Its my first time using Houdini in a shot, And I know its only a small use, but it really helped the final shot. Also Houdini was used as quick way to preview sims because RealFlows playblasts are very slow compared to exporting to Houdini and running a FlipBook.
Final Fluid Sim Passes 
Back in 3DS Max, I loaded my Core Fluid PRT Files via Krakatoa and used Frost to mesh them, the Final Fluid Mesh was very hires, I suffred lots of complete PC Crashes. I used Zhu/Bridson and squeezed as much detail into the mesh as possible. The PRT Loader was using a Magma Modifier to Colour by Velocity, For the first time I used Extra Tex Map using a Vertex Colour Map to export the velocity pass aswell as a Refraction pass. From this I could control how much whitewater I wanted in comp (Thanks to Adam Trowers for help on this). Its worth the extra work to get the White Water at high velocity, I think when CG Fluid sims fail is when the water shaders don't work, if you look at a big splash in real life you see the water moving at high Velocity and Vorticity turns white due to bubbles under the surface, so using Krakatoa and Frost you can simulate this.
 Frost Mesh Testing
Showing How Much Realism Velocity Shading Adds
There were a few edge errors due to the deleting particles, however messing with the Blend Radius sorted this out. (Also looks worse in the viewport due to only displaying half res).
Frost Mesh Artifacts, Sorted at Render Time
I Rendered through V-Ray to get the Velocity (ExtraTex) Pass and Refraction Pass. The Kaiju was also rendred in this beauty pass so all the Refractions and Reflections were accurate. The Fluid Mesh waves were made using Noise and HOT, and copied onto an Extended Ocean Plane. The Splash and Foam passes were rendered in Krakatoa and the Mist Was rendered using RFRK through Mental Ray using AutoVolume. The Frost Mesh was NOT used as a Matte Object only the Kaiju. I used Boxes with a Matte Material on for the Frost Mesh Pass in front of the edges to fake the Open Boundaries Mesh (See Above Screenshot).
Final Render Times were done on my Nights after working on my new FX Job I don't know the exact times, but I know the Fluid Mesh Pass/ Kaiju Beauty pass took around 25 Mins Per Frame.
There was a lot of Compositing for this to blend all the sim render's together and combine the Fluid Mesh to the Extended Ocean. To Do this I used a few animated masks and the paint erase tool to soften the edges.

Blending the Fluid Mesh to the Extended Ocean
Compositing the velocity Pass to create the white water is very simple, I have the Refraction Pass of the Fluid mesh in a separate comp and then I add the ExtraTex/Velocity Pass ontop and lower the opacity depending on how white I want the white water. Then I blend this comp into the main Fluid mesh using Screen Blending Mode. I had to manually mask the velocity pass with a lot of feather to slowly fade the velocity pass in after the flickering of the initial breach, but also because it was effecting the whole surface of the fluid mesh, so manually masking the splash area only is the way to go! This is pretty much invisible due to the Splash, Foam and Mist passes hiding it. Same for the Splash Pass, Masking the flickering out where the Kaiju mesh is and lots of feather, but by then the Velocity pass is hiding that, so theres lots of masks going on to hide the errors!
 Colour Correction On The Splash Pass
Velocity+ Refraction Pass to create White Water Shading In Comp
A lot more CC later, one really nice thing I did was add some noise/grain to the mist render to kinda create the look of particles rather than looking to much like smoke, it also helped bring out the detail.

I added some Camera Shake and Zooms to add more Movement to the Final Shot

For the Final shot I made a short breakdown showing the Render and Sim Passes, Theres a lot to be improved on my Next Fluid sim shot, but its a big step since my last (The UFO Crash), A lot of the problems were due to the bad geo/Skinning of the Kaiju, that's what caused the flickering, so it caused for lots of masking and feathering the passes together in comp, they saved the day!
I really like how the Final Sim looks, the Mist Really adds a lot to it, It was nice to use Houdini and VDB for my first time. Also the ExtraTex Map Velocity shading really helped the Final Shot! Back to doing World War FX now. More Info Soon!



Thanks For Looking, I learnt so much from this project, My Next Fluid Sim FX Shot will be even more ambitious!

Monday, 8 September 2014

My VFX Future Plans

I just wanted to make this post firstly, to Thank Everyone for all the positive feedback, Emails and Comments from my War Thunder Breakdown.

I feel so lucky that my first project was working on one of my "Dream" Projects! Since I uploaded my VFX Breakdown its had over 10,500 views. I've also had quite a few emails and messages from a lot of my VFX Idols, Thanks so much guys!
 Now I have got a good body of work to go on my Reel rather than it been full of personal projects. I've got more possible FX jobs lined up, however in the long term I really want to be hired and work at a studio.My blogs been quite quiet. instead of posting here a lot, like I used to, I'm spending much more time on each shot/RnD I do, to get it right. As well as working on freelance FX work I plan to work on a VFX sequence.
The reason why is at the minute I am just doing lots of RnD and uploading it, and its nice but I want to attract more clients (While I'm Freelance) but mainly to make my showreel shine, with a sequence plus a breakdown for the sequence. So I plan to do a 30 Seconds Max VFX Seqence based on a very cool idea! The aim of the sequence is for me to demonstrate my FX skills, but also been able to Light and Render And Composite the shots so they are realistic. I want to use as many technical tools as I can but making sure I can get the sequence looking to a Game Cinematic Trailer / near photoreal quality. However I will not be doing Modelling or Textruing!
I've also Started Learning Houdini's FLIP. So I hope to use it on the project, its defiantly not a replacement for RealFlow for now, its a very technical program, its easy to forget about your shot and get tangled up in node networks! I never thought I'd get to be able to use it, but now I am able to jump in and do fluid sims and export them for meshing and even export them to Max and use my RealFlow Meshing/Rendering Pipeline.
Saying all that, I have the biggest RealFlow sim I have ever done Rendering. So stay tuned for a making of article of it!
RealFlow Kaiju Breach Project, Coming Soon:



Friday, 15 August 2014

WarThunder GamesCom Trailer- FX Work Making Of

All Screenshots © Gaijin Entertainment
All Working Screenshots © Niall Horn VFX  

Hi Guys, So since June I have been working with the guys at Juice, Who are part of Platige Image on a WarThunder Game Cinematic trailer that has been released at GamesCom this year. Let me start of by saying what a pleasure it was to work with such a talented team on such a short turnaround of just over 2 months. This has been my first "Proper" freelance job, and working on a Cinematic Trailer for my favourite game was a dream come true. As you will see on my blog I've done some Tests in the past on WW2 themed shots (and I will continue to as I improve my FX and VFX Skills). Before I continue just want to say big thanks to my Supervisor: Adam Tunikowski and My Lead TD/FX Supervisor: Błażej Kowalski.

I Uploaded my FumeFX Summer Test/RnD Reel in Early June and a few days later got an email from Juice/Platige asking if I would want to work on the WarThunder Trailer. OfCourse I said Yes. I've learnt so much about working on a Studio Pipeline, Working with other really talented artists and in an FX Team, Learning a lot about how Platige/Juice setup there scenes, Also learning more about FumeFX and having dallies over Skype. Its been so much fun and a dream experience, My First job and I got to work on my dream project. I also got acess to use a second computer in Poland at Juice so I could leave renders and sims on the Farm overnight while I worked on my PC. This helped so much, Thanks Blazej! Render Farms are amazing!
First of all my Final FX Works on the trailer are:

* All the Flak Simulations, Flak Lookdev and RnD.
* End Shot Hero Bomber Engine Smoke, I also did my own Engine Fire sim and lots of RnD also.
* Part of the Bomber Crash Explosion/Fireball, Lots of RnD For the Explosion and Shockwaves.


FX- Making Of Process

Early in the project I was set to work on the FX for the Tank Crashing into the Jeep shot, However this whole shot got given to the amazing team at Bottleship VFX in Bulgaria lead by the amazing Hristo Velv!
Bomber Explosion
 Final Bomber Explosion in the Trailer 
So my first FX element to work on was the explosion of the Bomber (Before Time Reverses). I worked on this from Mid June to the start of July. For the explosion I was using FumeFX and PFlow to drive the sim. Very early on it was established that the client (Gaijin) and the Director wanted a fiery explosion that freezes in time as the bomber debris heads towards the camera and then time goes in reverse.
Bomber Explosion WIP Comp Screenshot 

My First thoughts were to have the burn rate really fast on the emission/s and then animate the burn rate and timescale down to very close to 0 over the duration of the shot, however this does cause a few problems. First is that you start getting weird patterns in the fire when the burn rate is keyframed for to long and there is no longer any fuel/particles emitting, so the fire started to look ugly and almost blobby towards the end of each sim. To try and combat this I made over 25 Fire Shaders testing them all out. Another problem was, which I told the guys at Juice, I said when you have firey explosions going into environments, they may look good on black backgrounds, but when they have to be cc'd to match an environment you get problems with the lighting but also with the fire edges in comp. At this point I was looking at reference of the explosions RealtimeUK did for the World Of Warships Trailer at E3 2012, very fiery explosions with little/no smoke.
At this time I had done quite a few iterations, lots of Fume Previews and some Lowres-Mid Res Renders, I was still having problems with the fire lasting for so long. After speaking with the TD and Supervisor, I decided to go down a different route of using not so little burn rate but use lots of Multiple Scattering which I rarely rely on, which worked well with my particle setup because I have two PFlow emitters emitting fire on the impact Frame, then about 5 frames later two separate smoke only emitters on the sides which then rely on the high multiple scattering to illuminate them, thus giving the effect of fire burning at a slower burn rate. The fire itself does last for over 100 frames, but the gradual falloff to the smoke that's created thanks to the scattering works really well. And that's what was used in the Final Sim. At Juice there was another FX artist working on this also, So mine and his explosion renders were mixed by the Compositor.
Its a lot harder to get an CG/Fume explosion looking good when the camera is pointing downwards areal instead of the camera facing the side. Also the explosion had to crawl close to camera whereas on a side view the fire would be crawling/rolling out of frame.
I also did a thin Smoke/Shockwave you see across the ground in the field. This was done using a object source of a Tourus with a noise Modifier on in FumeFX.
 Then I had KeyFramed High turbulence to get this^
Working Many Late Nights/Mornings. This Selfie took at 3:30am While Working on the Explosion. Cappuccinos +  Lots of Sugar!!
Flak Explosions
 My Flak Sims were a Key FX Element in the Final Shot
 After I handed over my Render/Scene Files to Juice I started working on the end shot along with some other FX Artists, I had to rolls FX wise on this shot, 1 Was doing all the Flak Explosions and the other the Engine Fire (See Below).
In my own projects I had been doing some basic Flak RnD but for the trailer I wanted to do better than anything I'd done before same goes for the Flak Sims. I started by looking at a lot of reference, Based on the Animatic and conversations with my TD We were going for the "Spikey" Looking Flak and not the Proximity Fuzed Flak which was developed later on in WW2. I referenced and researched the Flak that had been done in the RealtimeUK Trailers by Graham Collier and did very basic annotations of How I thought I could get that look. Also I looked at the Flak sims the Pixomondo did for there work in Red Tails (2012).
I really liked the look of these due to having the "Core" Trails which mushroom on the edges and then various different coloured Smoke trails that radiate out with different thicknesses. So my Fume FX Setups were using a Quick 2 Frame PFlow System to emit a quick Fireball which then burnt out really quickly. Then I have the Black smoke trails which mushroom on the edges (Like Real Flak Explosions Do) by using a PFlow Spawn setup and then a high Velocity Multiplier in Fume and high inheritance and Divergence values in the PFlow System.
Then For the Longer and Thinner Trails they were using the same system, just a lot less inheritance and Divergence, and longer lifespan so they "Shoot Out" More. In Fume I used Colour Caching to have different coloured smokes for each emitter. The Fire and Mushroom Trails would have black smoke, then the thinner trails would be white and grey smoke. I had a quick blast of turbulence at the start of particle emission for about 5 frames. A Key thing I did was using an animated the timescale so the Flak bursts and shoots out trails then by the 20th Frame we slow the sim right down and it really does look like Flak Then.
 Early Flak Tests
  Early Flak Fire Tests
 The Plan for the trailer was I would set up several Flak Scenes and send them over to Juice so they can merge the FumeFX grids into the FX scene, which was what they did at the end, but When they wanted even more Flak some of it was renders hand tracked by the Compositors into the shot. When we started testing this we realised that the flak needed to be more Turbulent so I used a lot of Turbulence throughout the Flak sims to get nicer motion, also did not have such a big drop in the animated timescale, Also I took out the longer trails and just kept the short grey trails in.
More Turbulent Flak
Flak Test Render and Comp
Overall for my Flak Sims I wrote over 5,000 Words of notes all about making Flak Sims! The Team at Juice Lit my sims using there setup and rendred them on the Render Farm. Credit goes to the amazing Compositors at Juice who made my Flak sims fit well into the shot.

B-17 Engine Explosion
Out of all the FX this is what I spent the most of my time on. Theres a few B-17s that get shot down in the final shot however I worked on the one that is closest to camera before the camera swings around.
Final Engine Fire In The Trailer
End Shot Concept Art- By Juice
 I worked on the engine explosion once I was happy the Flak sims were getting there. The big problem in the scene was how far in 3D Space the Bombers travelled it was impossible to use one long Fume Grid for the Engine Fire. I looked into using N-Sim in Fume but in the end went down the route of having the FumeFX Grid attached to the B-17s Wing, and then faking the velocity using Directional Wind and some Turbulence.
FumeFX Grid Attached to the B-17

The Grid However still had to be very long because as you see in the Final Shot the camera has to swing past the end of the grid. When the weeks and days were counting down I looked into doing Volumetric shaded particles using Krakatoa Voxel rendering and using PRT Volume. However its really hard to get the right result and is no competition for FumeFX.
Now it was clear I was going to have to make the grid smaller but not to short on the length. I also had to use really high substeps to avoid getting bad patterns in the smoke. All of this made the sim times very long, so when i was doing the hires sims towards the end of production I was leaving these sims on the the RenderFarm node overnight which sometimes led into two full days for one sim. Another Problem I had was the smoke was to fat, where as the Director and TD were looking for a lot thinner smoke like you see in the Red Tails Opening Sequence.
B-17 Engine Fire- By ILM- Red Tails

After some testing I decided to use a Cylinder to kinda Shape the smoke and keep it from becoming so "fat and puffy". To stop the collision along the edges of the cylinder, I tried to use animated noise on it to get rid of the uniform edges, however due to needing to have a high res cylinder collision so I could have animated noise, the sims were taking even longer. So I had to use really high vales of Turbulence and Vorticity to keep the turbulence throughout the sim.
Using the Cylinder worked quite well
With this success I was able to shorten the fume grid and after some testing I was able to hide the end of the grid as the camera pans past. Originally planned was some debris to fly off the engine and fly over the camera, however we did not go down this route in the end. I also created a sparks system based on the FumeFX smoke but that was not used by the compositors is the final shot.
 My Engine Fire/Smoke

 Early Slap Comp of the Engine Fire/Smoke

In the Final Shot there are Four Engine Fire/Explosion sims including mine, the others were done by the talented FX team from Juice/Platige. My Final Engine Sim got mixed with another Artists Fire sim.
Once Again, Just want to say what an Amazing Experience its been, It has been the most fun 2 months in my Life. Working with Platige/Juice has taught me so much about working with a studio and to a real life client. I've made a few new Friends and connections, and hope to be working with the talented teams again.
Heres the Final Trailer:

Update! Breakdown Now Online: WarThunder FX Breakdown


Thanks So Much For Reading This!