How do you Flame?

I would love to see some type of analytics of my flame use. Time spent in certain nodes. Features used or not. I know @fredwarren mentioned something like this and I would find it fascinating to look at my own :thinking:

Do you work mostly in batch?

Can you do your entire job on the timeline?

What is your go to keyer or colour correction node?

Are there some features you never use? 3D? Paint? Particles? Expressions?

Do the new features pass you by (SLAM tracker, Perspective Grid) or do you get on them straight away and give them ago?


Desktop Action. Keeps my memory healthy…


Pretty much exclusively batch.
Rarely use timeline.
I’m a great believer in the MK, I love the ability to create a mini-batch inside it.
Go to colour corrector is, the colour corrector, followed by the colour warper
I always try to use the new tech and can’t envisage going back to a time pre ML, motion warp tracking, SLAM tracker.

As a side note, before anybody flame me for sticking up for the modular keyer, I’m in a unique position where I never have to hand setups over to anybody.


Just by guessing:
40% Batch
45% Timeline
10% BFX
05% Effects Tab + Desktop FX (Tools)

I rarly use 3D and Particles, as it is just smoother and faster doing these in another software. Also never use the old gmask and also only once a year the MK if rly needed.

My most used nodes for me (mostly beauty work): gmask tracer, color correct, elbow, action, color warper, followed by comp, paint, ML Face and 2d transform.

Every new feature is tested, but only used if actually faster/cleaner

1 Like

This is funny to me because I’ve been keeping a little living and growing document for my purposes/ amusements that I call the Batch Manifesto on my phone for the last year or so. This is it- it’s really just for me, and there’s plenty to disagree with and I don’t actually know what I’m doing most times most likely but since you asked. I do shot work in batch:

Batch Manifesto

Action is sometimes (but not always) superior to comp nodes, and Vice versa

Don’t be afraid of linear.

Linear is your best friend when it comes to blurs and comp math.

Linear is not so friendly when it comes to sharpening, resizing, tracking, and regraining.

There is always a better way to do something that you have done one thousand times before.

Better ways that take more time sometimes aren’t worth it. Constantly search for better ways anyways.

Black points and white points are important.

Chromatic aberration, not so much.

Colour management has a steep learning curve. It’s worth the climb.

The rule of halves: add the amount of glow that you think looks like enough, then cut it in half.

Pixel spread is a last resort.

Spill is in the eye of the beholder.

Measure twice, cut once. Pull the exposure up when you measure.

It’s ok to cross the streams in a schematic, if necessary.

It’s never okay to go backwards in a schematic. Left to right, never right to left.

Swapping inputs in comp nodes isn’t helpful, it’s confusing

Comp like it’s an element you are giving to someone else.

Trust other people. Better yet, learn as much as you can from them.

Cycle your channels to check your grain.

Premultiply when you need to, unpremultiply when you feed into action

Premultiplied is just multiplied.

Unpremultiplied is just divided.

Premultiply if effecting matte and fill identically.

Unpremultiply if effecting matte and fill differently.

Color correct unpremultiplied.

Blur premultiplied.

Color correct before blur.


Learn to camera track, it can’t hurt.

A good camera solve can solve almost anything.

Anamorphic lens blurs are biased on y.

Bright backgrounds erode dark, defocused foregrounds.

Reflections are tough. Always will be.

Distortion workflows are not as complicated as they seem.

Redistort only the element you’re comping,
never the entire plate.

Too much is easier to dial back than too little is to amp up.

The Rule of Cool: Photorealism means nothing if it doesn’t look cool

Concatenation errors are not very noticeable, but comp like they are

Steal everything you can from Nuke people

Pixel math is normal math in linear. Pixel math is weird in video.

Use negative values to your advantage. But nobody wants negative values in a final comp.

Don’t waste time with flame particles unless you really want to. Just use after effects. Or Houdini. Or anything else.


Amazing post


Yeah. Loved this @BrittCiampa :heart:

1 Like

Are you serious? Good for keeping dementia at bay i suppose.

We used to do this though. Remember? I would have a fancy naming convention for file saves pre batch:


Was an action going into …


… a colour correct and so on and so forth. Yikes :scream:

Exactly! I remember those times.

(It was just a joke)

One of the best responses of all time.


You have a full compositing book my friend…

1 Like

@BrittCiampa That’s an awesome list. Thanks for sharing.


Ha! It’s an ever evolving work in progress!

I would also say don’t do CC feeding any matte to the CC node. CC, then use comp node with matte to comp it. CC has always had some odd behaviors when feeding it a matte directly.


I heard about this before. Is that still the case, or is it legacy? Can you show an example?
I always use the matte input in CC nodes, for example to create a shadow for some text. Never had any problem.

Its usually fine as long as you clamp the matte

aces cg

pattern browsed comp versions

pattern browsed anim playblasts - bulk update my timeline every time there are new shots to present.

work whole job in one daily desktop, pinned WIPs reel, flick easily from batch comp to batch comp. one button to save whole job.

reel groups for seperate storage areas.

use write node to publish flame comps to network so pattern browsing can version up easilly.

all comps in batch. no BFX unless they are titles in rec709 or minor QC fixes post grade.

preset tokening in batch names and render names.

preset tokening for shots - __v1 for example.

WIPs reel pinned to desk - so any batch setups can change, and u can always see your WIP and update it.

as many WIPs as possible connected conform, usually built manually from master for piece of mind.

right click ‘duplicate connected segment’ for mass comp additions across all edits.

after intial conform and setup is done, im spending as little time as possible updating shots manually on TL and relying fully on pattern browsing, reminding all my artists to maintain a standard of metadata from render_v1 to keep the pattern working.


Too many exceptions to the rules, I comp like this until I get this kind of job. I tend to try and learn everything and keep the rule keep it tidy like someone will pick up the setup at any time and try to balance speed and quality to maximize the project which means working sometimes in unusual ways.
To clarify in a commercial setting I tend to work batch since its easier to hand off, however in episodic the timetable sometimes can be a few days and breaking all the shots out would take time away from doing shots so I tend to bfx some of it unless they are large then I batch it, I got a weird graphics job and worked strictly timeline since it was all time slip based so largely bfx with matte containers which really sped it up but its wild how flexible you have to be in order to take all kinds of jobs its nice to have a wide range of skillsets to pick and choose workflows that best suite the job.


Thanks @Dan

I am loving the pattern browsing. I have to get all militant with people and their naming convention but secretly I like that.

I love the manual right click ‘duplicate connected segment’ . Helps to fill in the gaps of the connected conform.

Hey @BrittCiampa. Do you (or anyone else) have any recommendations for learning color management? I’d like to share some resources with my team. Thanks in advance.