Blender stories – learning to light

the beauty and intrigue of working in the world of 3D lies in being able to completely control every element of every scene.

unlike stills and film where your life can become anxious and deeply stressful when the weather won’t play along, in 3D, you are the weather, and the lighting, and the colour of every element, together with its placement, and your actors are never temperamental or unreliable, and you have infinite fuel and explosives, not to speak of the infinite options available to you in lighting and lenses, and even including anamorphic.

truth be told, i have an addictive personality and right now i am hooked.

so let’s dive into it.

freshly brought in as an OBJ, all i’ve done was add a sun, and 4 x point lights in the fire fighting converted shipping container

one of the less thrilling aspects of the 3D world is that software houses prefer not to make their filetypes interchangeable, or shareable from one package (like Maya) to another (like Blender). as previously mentioned i have experience working in Google Sketchup (before it became Trimble Sketchup) and used it to design and engineer actual fire protection systems. i therefore have a library of old files to choose from with pre-built sites. After a few hits and misses i figured out the right settings when exporting OBJ files from Sketchup, to (1) stop the imported model from showing up with blotchy materials superimposed on one another, and (2), the right way up.

initially i went for a volumetric lighting look with 1 fairly low sun

up until this project all of the lighting i had ever used in a scene was either 2 or more suns, or big overhead area or spot lights, and the occasional HDRI. so, having decided also to commit to volumetric lighting for this project, i then decided to get cute with the lighting…

lighting and believability

with careful attention to proper texturing, it is possible to create highly realistic and believable scenery such that your work can be an almost seamless cgi and real-world experience. however that’s not what i’m currently capable of, nor is it what I want to achieve

after the building

i started working on the vehicle lights. i had to build small boxes filled with smoke to create the headlights and tail lights of the truck. there is no shelf of ready-made screw-thread or bayonet or fluorescent light fittings to choose from. Everything has to be built from scratch.

a lot of this is trial and error

being completely self-taught has its advantages and disadvantages. one of the disadvantages is that i have to learn from scratch how to add realism. here i’ve added in some spotlights on the fuel tanks and dialled up the volumetric mistiness in the world

with the remainder of the structures done,

i also added in some red fire alarm lights, but truth be told, building little boxes of smoke and dropping a point light is probably not the way to go. while the light being cast is good, the look of the “light” itself leaves a lot to be desired.

not that easy to see in this image

but i managed to create an interesting cast-light on the floor, with light shining through a window and open door, which i cut out of the “building”.

now for the fire

first create a fire emitter.

give it a shader that makes it look like an actual fire

it’s always amazing to me how you can use a smoke and fire shader in Blender to make your fire look like one big ball of smoke, or take the same volume and simulation and make it look like clear-burning polar solvent, just by fiddling with shaders.

ladies and gentlemen,

a burning truck.

and a firefighter trying to get wet stuff on the hot stuff

 

pesky physics

i built a particle emitter and then had to build unseen geometry to divert the flow of particles. interestingly, with particle velocity between 20-23m/sec i discovered the best realism for foam flow through a 1.5” hose, which is roughly the same as the real-world foam. Getting my particles to spray in a decent pattern was harder than I thought.

getting better

but not quite there yet. i also had to add physics to the ground and the truck so that the foam particles would stick and run

looking for flame in frame

now i set about finding the most dramatic frame in the fire simulation, for a quick render

looking for flame in frame

now i set about finding the most dramatic frame in the fire simulation, for a quick render

there have been so many times when i have started out on a project with a vision, but not expected to get anything even remotely decent looking. i know my limitations and am acutely aware of my lack of experience.

but that doesn’t stop me.

i use a brute-force approach to life and learning.

it’s ok.
i forgive you for being judgemental about my little “play-play” make-believe world of fire…
i can also see the lack of realism and toy-like look of the place
but it’s the angle – your point of view – that matters

and for the bubbles themselves, dialling in jusssst enough translucency into each bubble…but not too much. and just like in the real world, when you take a photo on your DSLR and the lighting and colours look beautiful on the LCD but look dull and lifeless on the big screen, so too a small-scale, low quality render looks nothing like high quality 4k output. this means you have to constantly generate cropped output at 4k size and medium quality to guide you accurately toward conclusion. …but, even a cropped render takes minutes when you’re operating 2019 software on a 2013 machine.

…oh and did i mention, it was only in month 3 that i learned how to select a portion of my camera view, and render only that cropped portion? Up until that point i had to save keyframes of how the camera was supposed to be, and then zoom all the way in to the detail i was working on and scale the output all the way down, then when satisfied with the detail, move the keyframes back into place again. (don’t worry if that was all greek to you…)

we don’t need no water let dat muddatrucker burn
burn muddatrucker
burn

…what this means in the real-world is that it takes a tremendous amount of resolve and sacrifice to put in the hours needed to iteratively learn. when every small tweak takes hours to review, you must have a firm vision to work towards, and/or be willing to evolve within the creative limitations offered by the technology at hand.
there have been nights where i’ve set a timer for every 2 hours to wake up and run one physics “bake” after another for a series of fire sprinklers in a structure. to bake physics is first to, for example, create a hose pipe or tap with water drops that fall out of it and drop on the ground, secondly, to tell the water:
-how big its drops are,
-how heavy they are,
-how malleable and plastic-like they are – or
-how viscous the overall fluid is,
-how much they are influenced by forces like gravity and/or turbulence and/or wind, etc,
thirdly, to tell the ground:
-how solid it is,
-how permeable it is,
-how sticky it is,
-how much friction it has,
-etc
and lastly, to then let the computer run the simulation for the length of time specified, in the background, and calculate all the results of the geometry and the physics into big and daunting looking spreadsheets that are now “baked” and stored in a directory, and can become as big as 40gb for a scene.
the reason that you bake them is so that your computer doesn’t have to manually calculate each frame while you move back and forward in the timeline, as you work.
bakes can take time.

baking baking baking for the truck
for the truck
it’s ok guys. adjust that hose all you want. you have all the time in the world

in a scene where a lot of geometry is present, such as a truck, getting your foam particles to randomly stick and/or bounce, but not look too neat and mathematical, or too plasticine-like – finding that delicate balance takes painstaking attention to detail, and a bucket load of time.

on a recent project i had foam cannons emitting 240,000 particles each x 9 of them, together with a high resolution fire, and a lot of trucks. hence, every 2 hours i was up to bake another cannon…

therefore these 5 months have been a super-human effort for me, but one i have gladly embraced because i am well and truly besotted with my vocation now.

my commitment is total, but so too is my machine’s.
my computer can do nothing else while it is calculating physics or when it is engaged with a render of a visualisation.

onward and upward – with some camera motion i can fake the fireman running at the fire
also, note the increase in the number of foam particles
initially i create the physics in low numbers to test their movement and realism, but then, as i get closer to finalising, i increase the particle count and realism
also, note the luminescence of the “foam”
it’s really starting to look good at this point
i have all the time in the world to explore whatever angle i can imagine
my fire will never stop burning
my talent will never need a toilet break
the truck never needs to be moved
i never need to turn off the water

by now i think you’re starting to understand why this world can be so consuming to a creative personality with an already addictive nature – oh, and did i mention, somewhat of a control freak?…

but all along, these lights had been bugging me…
believe it or not, this is 2 lights each.
one light is deep yellow and faces back and shines on the already yellow geometry of the pretend lighbulb
the other light faces forward and shines down on the cab, with an off-white-toward-yellow shade
the front-facing light is then iteratively tweaked for brightness value to make sure it’s not overpowering
this felt soft enough in the end

it’s a truism of life that it is those little weeds we leave behind, those “i’ll get to it tomorrow” seemingly trivial problems, (or in my case) details, which grow and fester and keep you awake at night.
and maybe it’s a statement on the nature of my reality that the biggest problem i am consumed with is modelling in those lights that i know i should have done in the first place, a statement for which i am eternally grateful within the current paradigm of things…

the emotional trauma of self-discipline involved in weathering the attrition of watching a computer repeatedly, slowwwwwwwwwwwwwwwwwwwwly rendering out your test image
that trauma

…that willingness to undergo the repeated life-on-hold waiting as you iteratively work toward selling the realism. that’s not laziness. that’s anything but laziness

it’s probably also not sane…

But there’s also just about nothing else as rewarding…

i feel like i should add a title to my name.

technical illustrator for the fire engineering industry
that sounds about right…

that same curiosity drives me to explore unique perspectives and angles. this is partly why i love drone technology but only the proper, high-end stuff that has cable-like, buttery-smooth gimbal motion. the DJI’s and GoPro’s don’t impress me with their jerky, clunky, one-movement-at-a-time, camera motion.

yes yes, i know, those brake lights…
but look at the motion and look at that angle
*seriously though – these brake lights are going to freak me out now till I’ve fixed them…, which is not going to be today…sigh…
by this time you may have noticed i had abandoned the volumetric lighting.
at low quality renders and even with Intel’s beautiful a.i. denoising, there was too much blotchiness
not like this blotchiness. this is a low-quality render

by this stage i had learned how to introduce invisible geometry to make the flames “bend” around the tanker and wheels. i also added a second foam “cannon”, and got the physics of the “foam” particles to interact with the flames.

always time for tweaking “foam” appearance.

after ramping up and doubling the effective “foam” particle count, i discovered that the emission shader i had added, with which to help slightly illuminate each “bubble”of my “foam”, was causing a buildup and eventual overflow of too-much light into my scene, as the light-emitting “bubbles” were stacking up on top of each other…

one frame of foam and flame

rendering out a full HD (nevermind 4k) video at a decent quality (not like these low quality and low res images) is just not possible at this stage. On my computer it will take 45 minutes per frame for the (slightly over 10 seconds at 24 frames per second, or fps) 250 frames, or 7.8 days…

…which is why (and to circle back to adding a title to my name) i’m not quite ready to add “technical animator for the fire engineering industry” to my titles, just yet…
but watch this space.

and oh fuck, the geek in me just realised what day it is.

i’m completely in love with Blender, so if i was going to wish a happy valentine to something i would want to spend my life with, well, this is, tragically, it…

even in the real world.

happy valentines day everybody else.

and to you, my darling Blender, a happy valentines day and wishes for many more!

Leave a Reply

Your email address will not be published. Required fields are marked *