The Offset Podcast - The Offset Podcast EP059: Quality Control (QC)

Episode Date: July 15, 2026

In this installment of The Offset Podcast we're discussing quality control or QC for short.  If you've worked in production & post for long enough, you've probably encountered QC challenges and QC reports.  While passing QC can sometimes be challenging - understanding what it is, how it works, and how (at least most of the time) that its not personal attack can go a long way to treating QC as just like any part of a project pipeline.  Some of the specifics we'll explore in this episode include:When & where does QC happen?Understanding QC specs & why they're sometimes challengingThe importance of communication in the QC processWhat are the goals of QCHow is QC performed?Internal QC & client trustQC severity flags & understanding 'inherent to source'Check out offsetpodcast.com for our entire library of episodes. You can also follow us on Instagram & Facebook - just search for The Offset Podcast.Be sure to like and subscribe to the podcast wherever you found it and be sure to check out our growing library of episodes.  If you like the podcast it'd mean the world to us if you'd consider supporting the show by buying us a cup of virtual coffee -https://buymeacoffee.com/theoffsetpodcast

Hey everybody, welcome back to another episode of the Offset Podcast, and today we're taking a look at QC for quality control. Stay tuned. Support for this episode comes from Flanders Scientific and Gaia Color Direct Connect volumetric autocal. Calibration of reference displays is critical in our industry, and there's no easier way to calibrate than by using Gaia Color, which is standard on DM, XMP, and XMPC series monitors. Guy color calibration allows supported probes to be plugged in directly to the monitor for fast, accurate, and automated calibration with no computer or operator expertise required. You can learn more about this powerful system at flanderscientific.com. Hey, everybody, welcome back to another episode of the Offset Podcast. I'm one of your host, Robbie Carmen, and with me as always is my partner in crime, Joey Deanna.

Hey, Joey, how are you? Hey, everyone. Well, Joey, I can't believe in almost 60 episodes, we haven't touched this subject because QC or quality control is something that if you have worked in post-production long enough, it's going to come from somewhere. You're going to be surprised about it. You're going to have moments of anger and frustration, and then you have to fix a whole bunch of stuff. So I figure this would be a good episode to touch on this because it is something. that comes up a lot. And it's one of those things that, you know, it can be frustrating if you don't know what

there's what's being asked for. It can be frustrating if you don't know how you're supposed to respond, what you're supposed to do, what's worth fixing, what's not worth fixing. And in general, I think I just want to talk a little bit about QC as like a process for finishing, you know, projects destined for film and TV and stuff. Because QC can be a pain in the butt. and if you haven't navigated it before, it can be a little bit of a surprise

and a little bit of a frustration. So yeah, I just want to dive into that and see what's what. I'm sure you've had QC problems over your career, right, and how to address them. Like, like, in general, not what you're talking about. Right, you're perfect.

It's perfect. So I guess before we dive in, let's just define this just a little bit more because QC, I think, can kind of mean a few different things. In your opinion, what does QC mean? So QC is kind of the very, And as the abbreviation, uh, says quality control process, but the important thing is that happens at various stages, right?

Yeah, you know, mostly what we talk about is the kind of final QC of the project's finished deliverable because we're at the end of the pipeline, but QC happens at least it should happen ideally at every stage of the process from acquisition to editorial to color to final finishing and deliverable. Right. Every one of those stages should have a set of predetermined requirements. And that's the really most important thing about QC is that you need to know what standard to which you as a vendor, you as an artist need to meet. If that's not clearly defined, the whole process breaks down. So even if starting on set, if nobody defined what format you're shooting on, how we're doing slates, how we're timing things, how we're rolling, you know, do we run multiple takes per reel or do we do a new reel per take, especially if you have multiple cameras. Do we have time code on this one? Time code on that one. Do we have a multi-cam set up where we have a cam cutting between takes and B-cam continuously rolling between takes? If those kind of details are not established for every

stage, you can't do QC because what are you comparing against? And then by the time it gets to the next stage, let's take that on set example, the editor will pull it in and be like, um, why is your multi-cam setup a complete disaster? Was nobody paying attention. So then from the editor to them doing their edit, getting their timeline in order to the next stage, going to audio mix, going to graphics, going to finishing, if they don't have a set workflow where, okay, the colorist is getting this, the audio mixer is getting this and that's all planned for in advance. If they're just kind of winging it, well, guess what? The audio mix really, well, I can't take this AAF because of the way you organized it. Or the colorist might be like, well, you're coming from avid. You exported a sequence without four. Latin multi-cam clips, my whole timeline's blank. Right. And then of course with us, it's things like gamut, color space, metadata, flash frames, video quality, stuff like that. You know, but every stage of the process should have ideally, of course, with any workflow, there's some wiggle room and flexibility. But ideally, every stage from pre-production to production to editorial to finishing should have clearly defined specs and scope.

and then validated at the stage where you go from one stage to the next. Yes. I'm just going to say yes. There's a lot to unpack there. So before we dive into all that, just as a reminder, you can follow the Offset Podcasts on Instagram and Facebook. Just search for the Offset Podcasts.

You can also watch the video version of this show on YouTube. Just search for the Offset Podcast server on YouTube. And of course, we're on all major podcasting platforms like Spotify, Apple Podcasts. And you can always head over to the offsetpodcast.com to see our complete library, get additional show notes. But most importantly, over there on Offsetpodcast.com, you can submit questions, ideas, or any feedback for us to pay attention to. Yeah, man, so I agree with all that. You went through it all real quick of a high-level overview, which I love.

But let's just break it down a little bit more. I want to start on set. Because in general, I think that QC-type problems have a way of snowballing, right? and I think that, you know, when somebody doesn't care on set, that tends to ripple down through the rest of the pipeline to where ultimately it's a problem that somebody at maybe much later in the pipeline goes, why did this happen? And nobody knows, right? Nobody checked again. And I think you're your overarching point of a set of standards, a set of goals, a set of things to look out for in that process. Like, I don't think it's necessary, necessarily to be like a hundred percent dogmatic, because, you're right. Things change. Situations are fluid on set, especially, right? But like having goals in mind, and to me, the most important QC part, and you mentioned a couple of things like when are cameras stopping and rolling, when are card swapouts doing,

but like ultimately all of that kind of stuff on set comes down to data integrity, right? And I think that there has to be defined methods on set that a DIT, a camera person, an on-set producer is paying attention to, and that is, okay, what happens when we hit record, we film a scene, we take that memory card out, what happens from there, right? And that documenting that step of how things get backed up, how things, okay, here are dailies that we made versus the camera originals, here is a drive that we made versus an LTO tape on set, whatever it may be, that stuff needs to be defined, right? And things like, you know, check some validation to make sure your copy from the card is clean and backed up before you wipe the card, but also things like making sure directory structures stay intact.

You know, totally. I find it's not a big problem anymore, but early in the days of things like Sony and XD cam where you have this folder structure, people would be like, oh, cool, here's all your MXFs. Well, nothing links to anything because we don't have all the associated files. with those things or everything is kind of, you know, anarchy because we don't have the full folder structure. Those are the kind of problems that are completely unnecessary and can be easily avoided. Yeah, and I think also another thing that is important

is I think that it's just as important as documenting the process is documenting the potential mistakes that were made, right? So, you know, and that can be super highly camera level, feedback, right? Like, oh, we got these lens players that you're seeing are in fact intentional because our DP was going after them or, no,

this is intentionally exposed a stop or two under for X reason, whatever. Like, you'd be surprised how much of that data gets siloed really quickly. And like, by the time it gets to a colorist or a finishing editor or whatever, they're like,

wait, I've never heard that before, right? And so I think the more that you can document that kind of stuff, the better, the good and the bad. When it gets to editorial, I think there's some in particular QC steps in editorial that should be always considered because here's the deal. You're on set. You're moving setups. You're running fast.

There's moving gear, all that kind of stuff. You know, not a lot of time is spent reviewing the full amount of content, right? When you get to that initial editorial stage of logging footage and recording stuff, that's a great place to go, oh, well, this was a little foobar, this was a little wacky, right? And making notes of those things, recording the potential problems.

Oh, you know what? Like, you didn't run, you know, whatever, sync time code on this particular camera. Like, we're going to have to deal with, like, whatever it may be, right? Documenting those problems, figuring out a fix for those problems, if possible, because that's the earliest stage

before people are going to start getting into the creative part about it. And if they're constantly making mistakes there, they're going to ripple through the rest of the pipeline too. Yeah, absolutely. And you know, like you said, on set, it's all about data integrity. Yeah. In editorial, it's still about data integrity, but it's more about metadata integrity, right? On set, you want to make sure the full image capture is actually captured and saved and backed up and everything else.

And that's your QC focus. Editorially, you might be doing an offline edit or you might be editing online high-res footage doesn't really matter. proxies, not proxies, all of that varies based on your workflow, but what matters is you're not losing metadata, right? Because that's where the next phases are going to need. Well, you'd be surprised how many times, I mean, I'm sure our audience has heard this before in their own work too, right? Like somebody's like, you know, a colorist or a finisher is expecting to go back to camera originals or something. They're like, what are you talking about camera originals? Like, we transcoded everything to pro res proxy. That's fine, right?

And like, what did you do with it? What did you do with the originals? Oh, we needed those? Right. Like, and then you open up the timeline and everything is real number 0.01. Right. With time codes starting at and so they didn't transcode.

So this is like, if you're, you know, if you're editing with proxies or trans codes, you've got to make sure those real numbers and those file names and those time codes all come across. Or you get into some situations where, you know, one workflow example that happens all the time. We overcranked. We shot it 120 frames a second. We're going to transcode those down to 2398, edit from that. I go back to the camera originals there, 120.

None of my time code lines up. These are all things that can be worked around, but they should be thought about in advance. Yeah. And I would throw a couple more things in there, and that set, the editorial step two is sort of coming up with, because at this point in time, there's a lot of metadata that can be added, as we discussed. But there's also things that just from an editorial flow point of view are useful. and they fall into the camp of QC, right? Standardized metadata tagging with like keywords and stuff.

This is something that I'm continuously amazed by how editorial even works sometimes when like somebody calls it a car, somebody else tags it vehicle, somebody else tags it truck, right? And they like, well, I can't find this shot, right? So things like simple things like coming up with a shared...

Meanwhile, I'm like 1971 Ford. Right. With a dual carburetor or whatever, exactly. You know, coming up with standard naming, standard metadata tagging, standard vocabulary, all of that kind of stuff, doesn't seem like a really important QC step, but it is, right? Like, if you have editors constantly looking for things and not being able to find it, in my opinion, that's a little bit of a QC setup planning fail, because they should be able to quickly go, yeah, I'm searching for something, let me find it by this keyword or whatever, right?

Those kind of things, that metadata is more pervasive than ever with things like, transcriptions and AI plug-ins and all that kind of stuff. Like we're getting more of that. And if everybody's doing it slightly differently and not having discussed it as a team, it can be a real recipe for disaster. Yeah. And you know, you mentioned AI and transcription. That's where QC in the editorial process is monstrously important.

Because, yeah, AI-based image classification or transcripts, transcribing, or sorting, all these things are great tools that. make editorial faster, you got to QC all of that internally to make sure it's not putting you down a road where you don't want to be or it's not missing something and, you know, the one take you need isn't classified, right? So you never get to it in the editorial stage. You know, QCing automatic processes is just as important as QCing human processes, if not more. Totally. And so when things get to sort of us, the finishing side, mixer, sound designers, colorists, et cetera.

You know, at that point in time, we're kind of, our feet are in sort of two different worlds, right? So part of us is worried at QC, worrying about some of the things that we already mentioned, like original camera metadata, getting access to that stuff,

et cetera. But then we're now starting to worry about signal, right? And quality of signal in that kind of stuff. I'm talking about things like bit depth, video range, sample rate, channel configure,

You know, all of, you know, SDR, HDR, whatever, right? All of that kind of stuff. And if there are mistakes made in the editorial, like if a project is supposed to be, I don't know, destined for HDR and transcodes were done where it's like 8-bit, you know, 8-bit 420 prores proxy and that's all we got. Well, guess what?

We got less options now to finish this in H-Svr than we did going back to the camera. So that kind of stuff can flow. But also, I think, I tend to think that, a lot of editorial teams kind of punt stuff like that a little bit to the finishing team, to the engineer types, because they want to tell a story. I get it, right? But like, you have to realize that the steps before can cause problems for people downstream. So if you are on an editorial team and you're just punting these technical problems,

it's eventually going to become Joey's problem or my problem or if you're listening to this, right? The other thing I think that's interesting about QC from the finishing side of point is that we actually actually, I said we have our feet in both worlds, it's actually in three steps, right? Because now it's like, okay, we're dealing with editorial handoffs. We're dealing with signal stuff that affects our job as finishers. But then

we have this third part that's like, okay, well now we have to play by QC rules of the distributor and where it's going, right? So every network, every, you know, theater, festival, et cetera, is going to have different standards and QC and all that kind of stuff,

right? And here's, I think, the big thing about that is that some of those QC standards are really detailed and like OCD level. Some of them are really ambiguous, hard to follow. Somebody didn't do their homework on that. So talk about that a little bit because we've seen it. We've seen a little bit of it all, right? People are like, I mean, dogmatic about their process and their standards. And then people are just like, I don't know, just give us a pro res file, which can be just as bad and just as challenging.

Yeah. Honestly, I prefer specificity by default, but not to the point of pedanticness. Right. And what I mean by that is there are, you know, certain sets of valid configurations that should work for anything that should be fine. Right. So being super specific about those things, it depends on the workflow. You know, sometimes these things have to get pipelined through different transcoding things, stuff like that. So I would rather know in advance.

If the spec is very specific and it doesn't have, let me back up a little bit. I like the spec to be incredibly specific and valid because this is something that happens more often than you'd think is you have a very specific spec that was initially written. Okay. And then as lessons are learned as things are delivered, the producers start to say like, oh, actually, we don't really mean that part of the spec. so you could ignore that, that never gets reintroduced into the spec. That's the kind of stuff that can be kind of a nightmare for us, because the spec might actually be asking for something that's not technically possible, like an audio configuration that doesn't exist, or a set of metadata tags that doesn't exist,

or needs some special software to make or something like that. You know, that kind of stuff usually is worked around quietly, and then the specs aren't fixed. So that's when specificity bites you and bites you. Support for this episode comes from Flanders Scientific and the XMP 270 and XMP 310, the accessible, lightweight, and versatile monitors helping to bring HDR monitoring on set while also being very well suited to post-production work. Learn more at flanderscientific.com.

Well, that brings up the question, though, that I think is an important one. So we talked about onset editorial, a little finish, we're getting into talking about delivery and network stuff, is who is doing this documentation and upkeep, right? And I think that there is a tendency by pretty much everybody in it, no matter where you are in the pipeline, to think that somebody else is just going to do that work, right? Like, somebody will document this, right?

And I think that sometimes, like, there's also the idea from creatives to think about QC as a purely technical signal, like, you know, codex containers sample rate kind of thick and that's not necessarily true because i think a lot of this stuff is starting to you know a lot of qc standards obviously have impact on creative work stuff as simple as safe zones for bugs and text and you know so that's going to affect editorial and finishing where where does that stuff go right um it's also going to affect things like you know aspect ratios that are used frame rates that are used like so like there are technical

standards that affect both the creative and the technical side, but also who's doing that work. And I think the excuse by creatives that like, I don't know, some engineer will figure that out is not a good attitude to have. I think that if you're going to be, I think if you're going to be proficient in whatever part of the pipeline you're in, you have to understand specs and not just pass it off as like, that's an engineer's problem. Yeah. And on the communication side, what we see way more often than we should is somebody dug up. the last time they sent an email related to QC, sends it to a vendor, says this is our spec, and the spec is like pro res, stereo, no details.

So then you have to email back and be like, no, no, no, I'm sure you guys have something more specific. Can you get us that? So then they search their email again. They find something else that might be completely out of date, and they send it to you, and they don't know if it's current. They just know that from their perspective, I just sent the spec, right? And it comes down to the same question I ask for any workflow or any question is how much is a boat?

It depends. So what is the spec is not always the spec. And this is where our critical thinking needs to come into play. This is where we take a look at that document and say, hey, to the client, the producer, the engineer, whoever we're talking to, let me just get a quick gut check. Is this the current spec? Is there any addendums for things like HDR? Is there anything specific for how we uploaded? Is there, hey, I looked at the spec and this part isn't really well worded.

How would you like me to do it? It's much better to kind of communicate that in advance, even if you're, you know, I have a good guess. But, you know, I'd rather deliver once than deliver four times and get each little thing figured out with each delivery. And that's actually a really important thing. thing, right? Because, and we haven't gotten to this part yet, but we can jump ahead a little bit. And that is that, you know, QC is time and money, right? You go through the steps of delivery, especially on the, well, throughout the whole pipeline, but especially on the back end of the pipeline where you're delivering to a network or whatever, you know, oftentimes if you have to deliver something that's on a deadline, okay, we're going to have this by Friday, they QC it. And if there are mistakes, guess what? You got to go budget the time to go back, fix those mistakes, you know, fix those problems, resubmit. Even if the mistakes aren't real, you got to investigate them. Right.

So I think the more that that can kind of be done, as you said, clear and concise head at it's, I got one the other day that I just giggled at because I was like, okay, who the hell wrote this? It was, we require H264 pro res files. And I was like, huh, that's interesting. I've never heard of an H264 pro res file. But, you know, and they were like, oh, no, no, no, sorry. We meant pro res like in an M.

MOV container. I'm like, okay, that makes sense. But like, so like, I think that if there's something that's not clear, I think we tend to, especially on what we do as, as finishers, we're always people pleasing, want to hold clients hand or whatever. It's okay to push back on some of these things that don't make sense or a little complicated. And some of it, too, getting to the next part of this, after sort of like your, you know, finishing QC signal stuff, whatever, you're ready to deliver something. Some of these network and distributor QC process is you'd think that they would more or less kind of be the same. That's not ever been my experience.

Like, especially on the bigger side of things, you get into really pedantic, really complicated stuff. And a lot of it just seems stupid. File naming is one that drives me crazy all the time. And like, oh, you're using our old naming standard. Okay, well, where do I get the new naming standard? The best experiences that I've ever had when it comes to the process of QC and is when, especially on the delivery side,

is when networks have standardized documentation that regularly gets updated. Go to whatever, you know. Yeah, have it hosted. Right, NBC.com slash, you know, QC or whatever. There's a portal for it or whatever, right? Even, like, even to the fact of, like, some of the bigger networks that we deliver to,

like Warner Brothers Discovery, for example, they have, like, a whole online portal. When you submit a file, it's doing some basic checks. So, like, that stuff that could get you trouble down the line like, oh, we see that you submitted a file with the wrong name. You know what? We're not going to even accept it. Fix it first, fix the name, then resubmit it, right?

And there's automation that's gaining speed on this. I think the real ambiguous ones are when it's just a Word document written by somebody that's just sent it, like as you said, it's just sending it on and there's no clear and concise update, somebody who really understands it. We've had an experience, I thought, was really good. maybe in the past, I don't know where that was, maybe a year or two ago, we were delivering some stuff for Peacock, which is like NBC Universal kind of thing. And it was great because they had a, we had a meeting with, we're delivering a big film,

like a two-hour doc, had a meeting with a dedicated QC and delivery team. They gave us, we walked through the spec. What does this mean? What does that mean? How important is that? Is that important? Like answering questions live and there, making notes. And so when it came time to deliver, we weren't necessarily having to bother our clients who were just middle people, you know, middlemen in general.

We could go right to a QC contact at the network and go, hey, I noticed this weird thing about metadata. Can you explain that a little bit more? And they got right back to us. So like, there's, there's degrees of this, I guess is what I'm trying to say. Yeah. And that communication aspect, I think, is the most important part of all of this because, there is a both from the client side, the vendor side,

and the deliverable side, there is a bit of ego surrounding QC for everyone involved that needs to be consciously eliminated. And what I mean by that is this. If I send a deliverable and I get back QC notes, the instinct is, I couldn't have screwed that up.

It's my fault. Who am I going to blame? What is this? What is that? No, no, no, no. One Mistakes happen. Two technical things happen, especially if you're going through multi-stage where you need to assign metadata or things like that. Or you have multiple deliverables. Sometimes things can get flipped around. Human error happens. Human error happens on the QC side as well. We had a situation where we were delivering a whole bunch of promos for a huge campaign, and the QC House had changed their version of some software. So stuff that was completely valid by the network. Work's workflow, both the network was surprised and we were surprised was coming back as failing. And, you know, it could have devolved into a finger pointing, you screwed up, you screwed up, who screwed up thing. I just got on the phone with a QC company and said, hey, let's troubleshoot. This is what I'm seeing.

What are you seeing? And we ended up in about a five minute conversation, check in some version numbers, found they were different. And then we're like, oh, I think we've got our problem. Let me recheck it in the other version. Everything's great. These are happy. Wonderful. Now, my job is the colorist delivering these isn't to say, see, I told you so, my files were right.

Y'all are dumb. It was to elegantly, you know, kind of shuttle the client through that process. It would be like, let me just talk to them. If it's something on our end, we'll figure it out, we'll fix it. If it's something on their end, we'll figure it out, we'll help them fix it. But it's never an ego or a you screwed up or be screwed up or they screwed up. It can get adversarial very quickly as a knee-jerk reaction.

And I just want to say for everybody involved in this process, from the client to the QC House to the editor to the audio mixer to us in color and finishing, just take a step back and think, is it worth getting upset about, you know, So, of course, if somebody is trying intentionally to do something bad, that rarely happens. But just make your default assumption of no malice. We're going to try to work through the problems together. And I think you'll find the entire process gets a lot smoother and a lot happier for everyone involved. Yeah, I agree with that. To place a slight contrarian about that, I do think that it's harder for clients to understand that

than the people in the trenches, right? You know, you as a colorist talking to a QC guy, yeah, stuff happens, like, we'll make things. Clients sometimes take these QC issues as like, this person didn't know what they were doing, right? And it erodes, it erodes, trust. And that's why if you can limit the reply all chains and be more specific with your communication,

granted, with some workflows, that's not possible. But if it is possible, do it. Yeah, but I think to a greater point, like, it's that as a creative in the pipeline, one of your jobs is, as I've said before, is to understand these specs, understand the goals of QC and try to eliminate these potential. Like, if you're, you know, I do things all the time that I'm not asked to do that are not part of my job. Like, I'll give you a perfect example. I'm coloring a documentary, right? My job is to color the documentary. My job is not, oh, is this text, you know, 10 pixels outside of title safe, right?

But you know what I do? I throw up some title safe lines on the screen and I just go through all those lower thirds. And if I can make a nudge 10 pixels to the right or whatever so it's not a QC issue later, then I'm just going to do that. It saves me a hassle, saves the editor or the motion graphic artist who did that. It's a hassle. And I, depending on what it is, I might mention it to the client. be like, hey, just at FYI, all of these titles were a little out of title safe, but I fixed it,

no big deal. Like that kind of thing, because I'm aware of the spec and I know what it is, can really like pay dividends and buy you a lot of trust in respect because it's like, oh, this person is on a lookout for this stuff. They saved our bacon. It wasn't an issue with the network or whatever. And so that's what my point would be. It's like, it's not that it has to become adversarial, like, you've screwed it up or whatever. But like, it's like one of those things like, you know, Public safety things, like, see something, say something, right? Yeah, absolutely, especially because we're the last set of eyes on it.

Right, and the more that everybody can see something, say something in the pipeline, that minimizes those mistakes at the end. All right, so you mentioned this adversarial thing, which I think is really important. It's a good segue to talk about sort of what are the goals of QC, right? Because I think that this kind of gets conflated in a couple different areas sometimes, right? And I think ultimately the goal of the QC process is to ensure that delivered materials adhere to the technical timing and otherwise distribution goals of that particular outlet, right? So that from a technical point of view, all of those things should make sense.

Gamut, you know, peak levels, audio configurations, but even down to things like, ad breaks and the timing and the clock and like okay did you put bars and tone on not supposed to have bars and tone on this one or hey this one's supposed to be interlaced this one's supposed to be progressive all that technical stuff that seems to pretty pretty straightforward and i think that's oftentimes the bulk of the lot of the notes that i see is sort of like hey your ad break was one frame short or um hey you know a lot of like things like timing and clock and uh configuration I guess is how I would say that management, right? Like your audio track order was wrong.

Like this is supposed to be a mono track and you made it a stereo track, that kind of configuration thing. Where I think it gets a little complicated between technical and configuration problems. And you mentioned this earlier, but I think it's worth saying again, there is a little bit of a power trip that goes on with some people in the QC process. and I feel like they tend to sometimes step out of their boundary. Sometimes in a good way, sometimes in a bad way, and that's when they start making creative comments, right?

This doesn't look very good or this doesn't sound very good, whatever. And it's like, really? Did you just spend six months with a client in a suite, you know, working on this show? No, please shut up that, right? But like I have a love to you related to that? I've told you this before, but the worst, most ridiculous, quote, spec I've ever seen came from a document at a cable network. This cable network figured out through empirical testing what we all knew from day one, which is that any reasonable human viewer at a reasonable screen size and viewing distance combination cannot tell the difference between 4K and HD. This is a fact we know biologically because of the size of the cone and rod center.

in our eyes. They had invested a lot of money into marketing 4K. And then they realized nobody could tell the difference. They had a spec sheet that said, and again, this goes back to what we were talking about, all the way back to production, that said, we want you to make sure everything is 60 frames a second and stop down as far on the lens as you can

so you never get shallow depth of field. You get as much sharpness as possible. to make it look and I quote more 4k. That is when QC and tech specs can go complete. And literally every single production vendor that saw that just called their contact and said, we're not doing that, right? And then of course our contact was like, yeah, of course we're not doing that. We're doing it at 24 frames a second and do all the shallow depth of field you want.

You have a dispensation. So again, the specs are out of sync. So that is exactly where what you're talking about when the specs and the QC dives into the creative. That shouldn't happen. I mean, I've had it to the point where I've gotten QC notes before. It'd be like, like storyline points. And I'm like, this is not your job. Like you're not, you're not in the, like, you know, devil's advocate here.

Like you said, see something, say something. Let's say someone says the wrong line and no one knows. noticed it. If somebody in the QC process just flag something, not fail something, just flag, say, hey, they flip that line. Is that the biggest, the greatest thing for creative things in QC, I think, is verify that this is intentional. Right. Yeah. But I also the client and say, oh, whoa, I didn't notice that before. Great. Let's fix it. Or the client can say, nope, it's great. We're good. That, that you're absolutely right about that. We'll get back to that flag fail thing in a second. But I also think it's important that. the creative team, the content producers, if they are on the lookout for certain things, to make the QC department aware of that, right? So, for example, the one that pops up in our world all the time doing a lot of television work is for international distribution, we'll maybe have something like different units, right? So here in the United States, it will be imperial, right?

Like pounds or whatever, right? And then it's like, okay, it's got to be kilos over here, right? And like, it's the job in my mind of the showrunner, of the producer, etc., to go to QC, Hey, when you QC back these, all the audio and measurement and units for this version of the show are supposed to be metric and for this version of the show. Like, okay, that doesn't really, like I'm talking about more about stuff like, you know, I feel I feel like you could have brightened up to the left side of her face, right? Like, no, like that's not your job, right? Now, along those same lines between like configuration projects, problems, one of the technical, some of the technical things that I think about that I get flagged in QC all the time or things like metadata issues. packaging problems, right?

Like those kind of things. I think those are fully apropos for the QC people to talk about. But let me say this, that I think that there's a question that comes up all the time that I hear people say of like, what does it really matter, right? And I think you need to take a setback about QC from a distributor network's point of view. It's less about saying, like, and the one that always comes to mind is RGB gamut errors, right? Remember when we used to get flagged for RGB gamut errors all the time, right? And there was like this idea that like you were going to somehow blow up somebody's TV, right,

with like the three pixels that are out of gamut in the left hand corner of the screen for three, three milliseconds, right? It's, it's never been about like damage to somebody's TV or, I mean, in the early days, maybe it was. In the early days, yeah. Yeah, overmodulation or whatever. But I'm just saying, like modern QC has never been about that. It's more been about standardization that everybody plays by these. rules, so therefore the product that you get is more consistent show to show series to series

movie to movie. And so I think that's like part of like the personal attack thing that people feel is that like, well, this is my project. It's different than that project. And that's true. But the goal from a QC perspective is just to make sure that everybody is playing within the same, you know, on the same field, if you will, in terms of the standard. And in streaming world and an HDR world, that becomes a thousand times more important. Because let's talk about a couple things we related to HDR and Dolby Vision. What you're looking at on your reference monitor

is not always going to be what the end user sees. Now we have downstream tone mapping involved. Every downstream tone mapping algorithm is operating under the optimistic assumption that the master was done to a specific spec. So if you have gamut errors, for example, and all your signaling doesn't indicate the gamut correctly, it might look really good on your reference monitor.

It gets to a streaming platform on an iPad. It would look absolutely wrong. So the other side to that coin is the pipelining that these networks and these streamers have to go through after the master is done. Because when something goes on a streaming platform, for one property, they make probably 1,500 transatlantic. Transcodes, right? I have a 720 HDR 10 megabit. I have a 1080 SDR 25 megabit stereo and then 5-1. And so basically every possible combination of their app can serve to a device based on the devices,

capabilities and the connection. They have a done transcode for it. All of that back-in process is automated. So you could have, and we ran into this, you know, on a on a pretty, big project where we were delivering IMF packages. In theory, the IMF package should be, if it's a standards compliant IMF, great, you can take it, pipeline it from there. Wonderful. That's the ideal world. The world we were in was the client hadn't fully implemented IMF, so they were taking the internal files, pipelining them in a proprietary way they had developed, and all of their specs were built around that. So we ended up, You know, it was an aspect ratio problem that I had where there were some letterbox and they needed the aspect ratio tagged in a very specific way that our IMFs were not.

And I can all day long say, well, it's a standards compliant IMF. You said you need an IMF. But the reality was they needed the metadata in a very specific way so they could pipeline the J2K file down the road. Because what we had given them, if they had just pipelined it, no questions asked, would have stretched incorrectly on certain trans codes. So in streaming world where the final product is not necessarily what we're looking at, every single detail being exactly what they expect, both from the tone mapping, color space, all that metadata side,

but also the pipelining side is absolutely essential. So you might get a note back from QC where you're like, well, no, I know it's okay. That file is all right. It might be an all right file, but it might not work in their pipeline. And you don't know the thousands of lines of code that their pipeline is running at the back end. You got to get it to what they want. Yeah, it's funny.

I've thought to myself often in the past 10 or 15 years, we've really transitioned from like traditional bass band video problems, right? Like levels, gamut, stuff like that, really transitioned to just a me. metadata problem for most QC, right? Like nobody's really caring so much about those two pixels that are out of gamut for, you know, a nanosecond anymore. But they are caring about, oh, you know what, your HDR 10 static metadata? Hey, nobody uses it, but it's not create right on this file. We need it correct.

Or your Dolby Vision metadata or your Atmos metadata or whatever it may be. And that stuff, you're right to point out the pipelines because I think that's where a lot of content creators, because it's, that's not their world, they don't really understand why the ask is the ask because they're like, what the hell is the deal? And they don't understand how fragile some of those pipelines are. Right. And 100%, like one of the, you know, one of the things that I, so I've been, I've been working on, not to self-aggrandize a little bit, but I've been working on a metadata editor piece of software, right? And as I've dug into building that tool, I've kind of thought to myself a lot of times,

Like, why does this even matter? Right? Like, why do I have to do it this way? And then it's sort of like, if you think about it, it's like, okay, in that pipeline that gigantic network is using, yeah, it might not matter for one tool, but step 13, it matters for that tool, right? And so what you think of as a content creator as just like pain in the ass, like, why are they making me do this really doesn't have anything to do with your initial delivery.

Somewhere on AWS, an FFMPEG run just failed. Right, exactly. And so it's stuff like that that it doesn't seem to affect anything, but it might affect those pipelines. Support for this episode comes from conform tools. Conform tools allows you to translate timelines between Premier and Resolve and other NLEs while automatically solving common issues that normally need to be fixed by hand. Avoid time-consuming trim and transfer issues and securely send large media files. to collaborators at a fraction of the size,

and in minutes instead of hours. With a growing toolbox of features, let conform tools handle the tedious stuff so you can focus on the creative. Built by Post professionals, Conform Tools helps editors, colorists, and conform artists move faster and finish stronger. Learn more at conform. Tools.

Now, let's turn our attention to how QC is actually done by these people, right? people, right? So I think I think that's an important part of understanding it. So the first part of this, and if you have, like us, you know, grew up in the 90s, early 2000s when, you know, the age of decks and linear suites and all that kind of stuff, there was often, at the place I worked out for a long time, we called it the penalty box. And that was like an entry level job where you had to go work in the penalty box. And what you were doing was just play back on tapes with a set of scopes in front of you, measuring at that time, you know, things like, you know, you know, things like, you're You know, what do they call that one, the screen door and like all those kind of, you know,

technical signal things, right? That still happens, believe it or not. There are people who sit around with a set of scopes in front of them looking at the file, QC in that, right? But first and foremost, what often happens more and more these days is automated QC. And I know you have a lot of experience with that. In broad strokes, where does human versus automated kind of means? as one before the other, do they flip-flop?

And what is automated QC trying to do? Yeah, in most cases, it's not the same for every delivery and every client, but in most cases, the automated QC is trying to get through all of the easily identifiable stuff before we have to pay someone an hour to watch an hour show, right? So it's got to be rec 2020. Okay. A robot can check if it's wreck 2020. A robot can play the file down and see if there's potential audio dropouts. A robot can play the file down and analyze for...

Duplicated frames or whatever. Duplicated frames or frame cadences or interlacing problems or any of these little technical things. And there's really advanced... I'm kind of minimizing it a little bit. A lot of the automatic QC software these days is really, really advanced in the problems it can identify and look for. The only trick is, nine times out of ten,

that automated QC report will have like a hundred things listed and it'll still be a pass. And those hundred things listed are usually things like, you know, oh, this could be an audio dropout. No, it was a pause between people talking. You know, they are,

they're very prone to false positives and most kind of presets that these networks do, avoid that. right so they'll give you the whole list they won't fail you for any thing that isn't egregious most robotic QC only fails if it's something that's absolutely smoking gun and it's supposed it's supposed to supposed to I use that in air quotes catch as you said catch the big things like okay your audio config is completely wrong uh oh you your track right your time code starts at the wrong starting time

code or, you know, that kind of stuff. And it seems to me that I get it for that reason, right? Like, that makes sense as a first line defense to where instead of your, instead before you're paying somebody to look at it, watch out it, make an opinion on it, you're doing that. And that software, by the way, I agree with you, man. It's gotten

really sophisticated. I mean, just to mention a couple of the tools. Q-Scan comes to mine. Telestream has one called Vidchecker. There's a Veneerer Pulsaar. There's a couple other ones out there. And they all more or less do the same thing. You as the network or distributor, you set up the rules and the things that you want the QC software to pay attention to and look for. And by the way, these can be

other things like blanking, title safe, action safe, stuff like there's a lot of things you can do, the metadata. I mean, one that burned me the other day, and we're helping a buddy of ours with this, is the idea of container versus essence level metadata, which was a, you know, which was a slightly new one to me, that you could have your container report one thing and the actual frames in the file report a different thing. And if, hey, in this particular case, they weren't matching. Hey, that's a QC fail, right?

So you get that human factor in after this automatic QC. What are the humans doing? Are they verifying that the automatic stuff was correct and it matters? Or are they doing something a little bit more beyond that? It's both. Like I said, the automatic stuff. stuff will usually pass with a whole bunch of false positives. The human side, ideally, again, not the same for all deliveries and all networks, but the

human side should, in an ideal world, watch the entire content from beginning to end, make sure there's no egregious video issues, spot check all of those areas that might be a false positive, might not, that's still past the automatic QC, but you still want to look at, And then they make a judgment call as to if it's going to go in the final QC report. And they kind of watch through the whole thing. Check all those spots. Check all the specs again. If like you talked about there's timing issues,

they will go in and double check the in and out segments of everything. Because the automatic system might not see a fade correctly. Or you might have an act break that comes out of black later than the audio for creative reasons. Right? That's something a good QC person could easily immediately say, okay, the automatic said your break time is wrong. No, the audio comes in. I see creatively why they did that. It's a creative choice.

We don't need to put that on the report. Right. So where the automation ends is where the human should begin. And hopefully the automation could be a help to the human. Yeah. Yeah, I agree. So in either, in both processes, generally speaking, a. a report of some sort is generated and delivered back to the client, right?

Like, if things just pass, you might, you might just get, hey, an email and say, hey, you pass QC, everything's, everything's hunky-dory and move on with your life. If you don't pass for whatever reason, most QC, and they come in different flavors, configurations, or whatever. Some of them are pages and pages and pages and pages. Some of them are summary reports. But generally speaking, in my experience, what they do is they generally rank the severity of the problem in terms of pass, fail, you know, kind of, it's usually like a stoplight, right? Green, yellow, red kind of thing, right? Where the red is stuff like, nope, this is a clear failure, this doesn't meet our spec for whatever reason. The stuff that's green is like, yeah, we flag this, but okay, cool, we're

cool with letting it fly through. I'll give you several examples that in a second. And then yellow is something like, hmm, we need to talk to somebody about this, right? And oftentimes I find the yellow stuff to be the most frustrating because that is where you get into the land and we're going to bring up a new phrase here native to source original inherent to source I think that's what I was looking for inherent to source tell us Joey what the hell inherent to source means and why is it become sort of the in the adversarial content creator QC relationship why has inherent to source become a bad word? No, it's a great word. It's the word I use for all of them.

We do a lot of documentary work, which means, okay, this clip is from a film transfer that was done in 1945 on a video camera telecine and then, you know, copied with a wire recorder and, you know, whatever, somebody recorded on VHS for a rebroadcast, and that's the only version that survives. And the QC note is slightly out of focus. Right? Or dropped frames or whatever. Yeah, exactly. It is inherent to the source. The best source, we've made the best possible effort to get the best source for this, this image. We either don't have the money to get a new film scan done or we don't, the other, you know,

a better source just doesn't exist. We've done our dual. The client has done their due diligence to try to figure out what the best source is, we have to live with this. Right. Now, can I say something about that? That's really kind of a tangent, but it really makes me giggle these days, is that so many content creators are aware of what QC departments are trying to do. And as I said earlier, they proactively try to fix problems so it's not a problem, right? So in the day, the day and age that we live in with AI, everything, right?

People are using like AI restoration tools or whatever to like generate new frames from bad frames or to sharpen this up or whatever. I've seen so many QC notes now about that kind of fix, right? Where the- Bro has seven fingers. Right. If you had just left in the original content and said inherent to source, it wouldn't have been a QC problem to begin with. Right. And said a lot of times that you can do yourself.

more harm than good by trying to sometimes fix things and I want to be careful about this. Try to fix things that are not well defined or approved as a fix, right? So it might be that the network comes back and flag you know, flags your soft archival material, right? And you might say, hey, look, we have a way of fixing this, but we need you, the network, to be a judge of whether this is acceptable or not before we go fix all 87 instances of. of this, right? And it becomes a creative question. How much fake sharpening do you want to add? That's up to the client, you know? Yep, yep, yep. And so I think that's one of the things that when you get a report back like that, you should stop in your tracks and go, okay, what are they really

saying here? Is this something that we did intentionally? Is this something that we did thinking we were fixing the problem, but it actually caused the problem? And like, just be prepared for those situations. Okay, cool. I can show you this pre, you know, up-res fix or whatever, which one do you? you prefer and get some parameters about that because I have seen that kind of thing last weeks, if not months. The most important thing in my opinion, when you get a QC report is you got to triage it, right? You got to, like you said, red green, red yellow, I have to say red green blue. That wouldn't make sense.

Red yellow, green, right? You got to look at those reds immediately, immediately have a plan to address them and present that to the client. Like, okay, yep, we got to meddata. or wrong. We'll fix that immediately. Great. Okay. You want to, that is the first and foremost thing that you want to communicate to the client. We've got it. It will be solved. We don't want any ambiguity there. Green stuff, we can kind of just, you know, spot check it. We like it. We already watched this down. We're already good with it. Great. The yellow stuff is when we got to kind of look at and say, I need to talk to the client about this. I need to talk to the QC person about this.

I need to double check my timeline in these five spots to see if I forgot to turn on my noise reduction or something like that, right? But triage it. So it's like solve these red things quickly, communicate about the red things quickly, then figure out the yellow things. And then lastly figure out the green things. And there's a bit of nuance there too, I think, that there are discussions that need to happen. Like I kind of think of this as like maybe three or four levels. there's conversations that happen with everybody involved, the network, distributor, your client, your team.

Then there are discussions that happen just between you and the client. And then third, there's discussions that happen just with you internally and your team, right? And sometimes you internally and your team is just you and yourself. Right, right. But it's still an important step. I bring it up because I think that sometimes, and I'm pointing the finger at myself on this one, that the goal of transparency and communication sometimes is a mistake. If, you know, if you made a mistake and you realize it was your mistake, making the network

and the client all aware that like, oh, man, I really screwed this one up bad, doesn't do a lot for confidence, right? So you're better off in situations like that of going, well, that's a mistake. Let me fix it. and just not just fix it there's no such thing as a perfect delivery right like there's going to be something that somebody's always going to be happy with right I'm not happy with rather um now let me ask you let me ask you this because there's a situation that comes up where you can't necessarily fix something right whose decision is it and how do you proceed when there is something like that

film scan that can't be rescanned or whatever like I don't think that that onus of making that call is on the operator, the colorist, the editor, the, you know, whoever, right? Like, that's got to be somebody a little higher up the food chain. Do you agree? Absolutely, because that becomes where the technical and the creative kind of butt heads, essentially, is, is this low-quality piece of video so essential to the story that the client needs to make the case to the network or the streamer that we need to have it in there? Or is the alternative, okay, it's going to cost $100,000?

to get this footage re-scanned. Is it worth that? Do we have the budget for that? That becomes kind of a, you know, a three-headed monster of, is it so essential to the story? Is it worth the, if the, if the fix is just cost, is it worth the cost? None of those things are our kind of world.

The best thing we can do in that situation is advise based on what, what we know and where our zone is, which is the image quality and the technical side. Say, okay, if your options are get rid of it or re-scan it, the rescan will get you in theory this much better quality, right? Or if it was easy to cut around and there was an easy fix to get rid of it, you could propose, hey, here's a optional easy fix to get rid of it, you could propose, hey, here's an optional easy fix to

get rid of this shot that you could make an editorial and everybody would be happy, but it's up to you, right? We can present the details of possible solutions and advantages or disadvantages, but I feel like that's where our role in that kind of thing ends. And I agree. And I also think that like it shouldn't really ever come to, it should never that like a lot of those big issues should never really come as a surprise and wait until final QC, right? Like, I'm thinking of, like, recently I did a project where there was a lot of 70s and 80s, four by three material, and the network had this thing about, like,

you cannot have any pillar box content. It all needs to be full raster. So how do you, okay, cool, you take a 720 by 486, you know, file, and you push into it 400%. Then the QC comments would be, this is soft, this is pixelated, whatever. And so what happened was our client, that showrunner, that EP, interface with their counterpart at the network.

They discussed the problem. And ultimately, there was a solution made where it's like, okay, this kind of stuff can remain four by three for this duration, right? And if it's a longer clip, we need to do this kind of treatment on it. And that will satisfy our problem. So by the time, like that was all decided during the editorial process, during the story. like, you know, the rough cut process, it was never a surprise because it was handled as an issue up front. And I think there was a lot of things that kind of get confused, that blurry line of content technical,

that could be solved much earlier in the process, but unfortunately become QC problems. And to me, that's a mistake. That's a mistake of the creative process when you get to that point. We didn't know our 4x3 footage couldn't be Pillarbox. Come on. Like, come on. Every stage in the process, like we've talked about, should have ideally an established set of expectations. As we said from the beginning, exactly, 100%.

Now, let me ask you one more question. We'll wrap this guy up. So the content kind of questions, like I said, I have a little bit of a love, hate, with them, because I don't think it's necessarily the job of QC to do, like a network or distributor QC. However, I want to walk back one thing about this because I do think that there is a certain attitude that people in the creative pipeline, editors, colorists, etc., like, that's just something that happens externally from it. I'm going to make the point of, especially where we are as colorists and as finishers and sort of the small end of the funnel, your job should be, and you do this all the time. you talk about to me all the time there's a certain level of internal

QC that needs to happen to save face to save a project to save money before you ever get to the distribution side QC and that's a big problem where I see a lot of finishers passing the buck and it's like no you owe it to your clients

to do that and I think it's just as simple as a watchdown, right? Always. And like, and a couple like just big gut checks is the clock right, is the time code right, are the ad breaks right, are the, is the audio configuration right, is the file for, like, I'm not talking about like you have to spend, you know, 50 grand on dedicated QC software. I'm talking about the gut check simple things that you can do. You'd be surprised how many people.

What does I call it the idiot check? The idiot check. How many? It always surprises me. I caught a friend of mine the other day who rendered something from Resolve and then just uploaded it to the network. And I was like, oh my God, what are you doing? I'll never do that. They're like, what, what, what, what?

And I'm like, you just sent that file without looking at it. Be like, well, I looked at it and Resolve. Yeah, but you didn't look at the file. Right? And sure enough, like the third time that they did that, they had set like an endpoint on their timeline. So now the file was like 20 minutes short and I'm like you would have just delivered that file, right? And it would have been a mistake and it would have been embarrassing. So you're right.

The idiot gut check kind of stuff like you as especially where we sit in the pipeline. It has to happen as it is. It has to happen every single time. There's no excuse for not checking the big gotchas, right? Yeah, it's your due diligence. Right. Nobody's going to fault you. for the two pixels that are out of gamut

for a millisecond in the corner of a screen, right? But if you just rendered out a stereo file when it's a 12-track deliverable, yeah, that's on you, right? And that could have been prevented. Yeah, absolutely. And it just comes to taking, you know, taking pride into your work.

You know, every so often, it's tough. You know, we get a lot of jobs that are just kind of in, out, it's commerce, it's not art, but at the end of the day, no matter what, no matter what the job is, if you took the job on, you should finish it with competence and with the full process of you being confident in what you've delivered. And that means watching it down and doing your due diligence every time. Awesome.

Very fun discussion. If you have any horror stories or things to add to the conversation about QC, feel free to drop us a comment if you're watching this on YouTube or on Spotify or Apple Podcasts. Of course, you can follow the show on all those platforms. You can also follow us on Instagram and Facebook to search for The Offset Podcast. And you can always head over to The Offset Podcast.com to watch our full library and get additional show notes. Also, if you wouldn't mind, we'd really appreciate if you could consider some supporting the show by visiting this link right here on screen to go buy us a couple virtual coffee. Every dollar contributed there helps to make new episodes and helps further the Offset Podcast. So we really appreciate that.

And then just as a last little goal of ours, we're getting close to a thousand followers. on Instagram. If you have a friend or a colleague that could benefit from watching the Offset Podcast, please send a link their way and if they're on social media, have them follow us that way

because the more followers, the more the algorithm works better for us and all that jazz. So we really appreciate that. So Joey, fun conversation as always. I love, hate QC, but it's an important topic to talk about. So for next time, I am Robbie Carmen.

And I am Joey Deanna. Thanks for listening.