Bugfest 2025.2

Flame Between Renders
1

this is super anecdotal but as a supervisor i see my artists struggle seemingly more and more trying to get stuff out of flame that isnt riddled with bugs. this is anout 2025.2

Even with all the workarounds in place , i am loosing trust in this tool seeign whats going on.

some stupid anecdotal stuff that I cant even put a finger on really, as hunting these, proving them and writing tickets is not something I can do all day, of course I try to but its difficult.

→ Conforming XMLs from premiere still has the action corruption bug, gladly we have a workaround script now, but we didnt for over a year!! and it wasnt adsk that gave us a fix either.

->Publishing from Arrriaw crashes flame on mac.

→ Publishing with “from source” colormanagement bugs out, so you cant have mixed colorspace footage to publish to acesCG without first pre-cobverting everything yet again.

→ Dont get me started on BFX, we dont use it anymore at all.

→ Random corrupted tlactions on export

→ timeline fades would randomly not comp through to the bg, resulting in black frames in the timeline on export.

That adds to mayor deficencies that cost me time and thus money

→ No metadata in EXRs , we take care to record lens metadata on set and flame just throws it away, this is very very much not nice.

→ No multiuser collaboration function.

→ No support for most codecs, we have to pre convert most stuff , or actually now due to arriraw bug - everything , this wastes time.

→ No path translation worth talking about

→ No solid way to do project versioning, archiving is no fun , too large, too bulky, too annoying.

→ Tineline resizing from xmls is not ideal, should be relative nowadays and not absolute… (resize vs action)

→ publishing is super rudimental, it always adds random nodes, position of elements to each other is not observed, relative handle frames is not a thing, cant add a template, stuff we have in NukeStudio basically.

→ No way to properly handle quicktime NCLC tags, we have to resort to external tools to fix flames behaviour.

→ scanning folders during conform takes way too long

→ no way to nest timelines, in this current world of thousands of deliverables this is bad, Connected conform is nice but it doesnt fix having to dublicate everything for “clean” versions. waste of time.

→ no advanced export function of creating multiple deliverables into specific folders and colorspaces and codecs, (see Hiero/ NukeStudio for how a export tool should be) .

→ conform tools are super nice , best in business when it comes to matching footage to a edit / however support for industry standard premiere xmls is just a mess. have to rely on user created tools to patch it.

→ you cant properly export a timeline from flame as a file (xml/aaf), doesnt support timewarps etc… insane!

Really I had a tough last 2 weeks with missed deadlines and quality issues, and its bit grinding on my nerves a bit, i can hire 2 resolve operators for the price i pay in flame subscriptions, maybe thats a better solution for me .

Of course there are good parts that I love, source sequence is magical, the whole versioning of comps with openclips is great. but thats all shortcuts of stuff a human can easily do and something you can with enough willpower code yourself and throw it into resolve…

1 Like
2

I’m kind of thinking that Autodesk shouldn’t be approaching you for any testimonials for Flame right about now @finnjaeger?

3

Watching fusion tutorials now maybe … man yea seeign my creditcard statement compared to the issues i have to deal with there just is a large disconnect thats driving me a bit insane.

1 Like
4

Totally get it @finnjaeger.

The thing is, there isn’t a standout option B out there. Not saying that Flame is the best nor the least worst but when you’ve invested a lot of time into a piece of software it is hard to make the switch unless the advantages are clear.

5

no totally.

i go through this rollercoaster every few months thats why i end up knowing pretty much any software now -.- and none do what I want.

always end back at flame , but its frustrating still

2 Likes
6

i dont know how many flames you have but above 5 i would imagine you would want some kind of data ingest team / workflow where it’s someone’s job to ingest and properly color manage all the footage.(resolve?) relying on multiple people, freelancers, etc to properly color manage footage seems like a guaranteed headache… or maybe i misunderstand this item!

2 Likes
7

we have someone dedicated for this actually /

We always publish out every plate as exr acesCG properly like how it was written on stone slates in 1200BC

we used to use nukestudio for all of this ingest and publish of nuke scripts, now we use flame to di the same with batches and openclips.

The problem is we have to go :

  1. Convert everything flame cant read into a format it can , which is many things

  2. conform from sources in flame

  3. push sources sequence to resolve/scratch

  4. convert footage in resolve to exr, prores or so

  5. reconform to new created media in flame

  6. publish shots from flame .

While it should be

  1. conform from original sources in flame

  2. Publish shots/batches from flame

many extra steps, media duplication, rendering , reconforming all of that necessarry due to lack of raw camera media support in flame.

Its a waste of time, even for the media i/o / conform person with lots of steps that can go wrong inbetween.

its not like Nukestudio doesnt have similar issues, proresMXF doesnt work there since nuke 15.1 and you cant batch- change any raw settings .

Hell even arri reference tool is writing corrupted exrs if you use any kind of compression

8

does resolve not have any feature like the sources sequence in flame where it will consolidate all the needed media into one string out? the reason i ask is it seems you could just build a ‘ingest’ conform in resolve and do the transcoding to EXRs in one go instead of round tripping from flame to resolve to flame again.

dont get me wrong, i’m with you and wish we conform from original sources once but i dont think that will ever happen in flame… there’s too many cameras too many codecs, it’s constantly changing, updating… resolve is prob a better place to start for that reason alone…

9

I actually want to do some testing with Assimilate Scratch again for this purpose as it supports all the codecs, extracts all the metadata and I believe you can do something clever to create a source sequence (this is something I could be very wrong about). It would also be a great tool for versioning as it has a crazy good export tool.

Mistika Workflows could be worth exploring as well for similar reasons.

In the end though, you could end up using several tools in this type of workflow which used to be the strength of Flame, the only usable all in one (bar grade). That hasn’t been the case now for quite a few years which is unfortunate.

1 Like
10

I hear you brother, as do I.

1 Like
11

it does not have that feature, I have stupid workarounds but they arent as good as sources seq in flame

We do use it like this right now but it leads to doubled media and just another step that we dint need to have.

a few caveats why resolve is not great for shot publish:

→ no sources sequence, so you have to render all timelines with timecode as framenumbers so that segments will overwrite thmselves and you end up with split exr sequences that are your sources, this works well but yea kinda overkill and can take a long time, and you can only use source name and source tc as framenumbers all into one folder its fugly.

→ no sequential shot naming, so have fun renaming every segement in the sources sequence by hand. and no you cant access this with the API…

→ well you dont get batch files or nuke scripts with it… so … eh. you need to publish again with flame anyhow :joy:

→ no good way of doing “halfres” only fixed sizes like flame , nukestudio is better here (we get a lot of mixed 8K dci/8K16:9 that i would like to just halfres)

and here comes the kicker, I often actually go “backwards” as in publish with flame, then match paths for all shots and shotnames manually and overwrite the published exrs so that the exrs have metadata for matchmove etc to pick up!

its a bunch of crazy “conform ninja stuff” that is working around all the different apps deficiencies.

its crazy, i dont want to have to do this, why cant we have a single functioning conform and plate publish tool… ffs

12

let me know if you find anything, i an still super confused by scratch and all i can do in scratch is convert prores raw and dng etc to something useable :joy::joy:

13

I am going to explore it in Scratch. I believe there is a way to group multiple timelines and manage the media within that.

For my purposes, Scratch can use timeline markers to create shot/element names so as we are working on Features & Episodic series, that is very much the domain of either the VFX Editor or assistant editor from the editorial team.

A little birdie has told me that the scripting thing is currently being addressed and that Scratch is being shown some love.

Just saying that the Scratch player is awesome whenever I have used it.

I am also aware that there could be loads of shortcomings which is why I intend to test it and not just implement it :slight_smile:

1 Like
14

keep me updated, i have asked them for many things like that, i really want a versioning system as well, it has so much under the hood potential but it has been on the backburner for a long time it seems and their focus was on LiveFX

1 Like
15

It’s because Gerk sadly passed away and he was a force of nature.
were he still with us he probably would have worked out how to drive the displays on the sphere from a high school calculator.

1 Like
16

I was not aware of that. What he did with Cyclops and Optical Flow was incredible and game changing at the time.

1 Like
17

You should talk with Alex McClean. He’a a Scratch master and also well versed in Flame and workflows. He would likely be able to figure if there’s a path.

2 Likes
18

I’ve had a couple of chats with @Mazze about it. He is a top dude who happens to be their product manager and highly familiar with Flame too. Would love an intro though @allklier if that is at all possible.

1 Like
19

Of course. Let me connect you.

3 Likes
20

hey @finnjaeger - I have been thinking about this workflow Rubik’s cube and had a few ideas…

one thing that i request on every job i lead is a copy of all the shoot dailies on our server. that way i can scrub through the footage myself very quickly (dailies i get from DIT are usually ProResLT HD) to find that precious clean bkgd frame or any other elements i might need. So it occurred to me you could have flame create a sources sequence from the dailies instead of original camera files. when you push sources sequence to resolve (step 2 above) you would have to unlink and relink to the camera files at that point (i’m assuming this is an easy thing to do in Resolve but not sure) and then all the footage you need is transcoded to ACEScg EXRs and then unlink / relink in flame to those new transcodes and you are off to the races. i guess this eliminates the step 0 “Convert everything flame cant read into a format it can, which is many things”

another way you can approach this is shift a lot of this busy work back to resolve. Have the data ingest person / resolve operator conform the spots themselves with whatever handles and then transcode all those sources to ACES exrs and export to your server. that is conforming the same spots twice for sure and there could also be some redundant transcodes or frames if you’ve got a project with mulitple cutdowns and what not but that can get sorted in flame via sources sequence. It’s less efficient than the ideal workflow you are proposing but if you look at pushing a lot of menial labor from what is likely a highly paid flame artist to what is essentially an edit asset level person… there’s an argument you are freeing up that flame artist to really focus the majority of their time on more complex tasks…

curious to know your thoughts and hope you’ve had a great holiday @finnjaeger

1 Like