-
Do you conform inside Resolve?
-
Is this for a tapeless only workflow?
Depending on what it is you are doing, it seems to me that you want to limit the rounds of compression (whether they are deteriorating your material or not)… So After Effects would not be the right place to go after Resolve (at least not for the entire project)… because if you needed to go to tape you would need to export everything then bring into your mastering tool… which, depending on the software could require a further compression after you render your project out of AE…
It seems to me, that you are hoping for a “one size fits all” solution, and at least in the “affordable” (for people, not facilities or seasoned established freelancers) there is no such solution, and even then there isn’t. If there were you wouldn’t have the options of FilmMaster, Baselight, Pablo, SGO, etc… There would just be “The Tool”…
It’s also worth noting that as a whole, Adobe Production Premium likely has all the pieces you are looking for…
-
Yes, it’s for a tapeless only workflow.
Somebody suggested Blender actually. It has 3D that supposedly rivals Maya, Compositing, matchmoving, tracking, roto, color grading and basic video editing with EDL import. I never knew that. I was always a bit skeptical with the whole open source tools support thing.
-
[MichaelMaier] “I was not sure if I should post this in the basics forum or here. Something tells me this is not really a basic question.
Once you are done with your grade, what do you do? You export the files out to be re-imported in the NLE and finished/final exported to DCP, Blu-ray etc? Or do you finish it from Resolve? If so, how do you do with the soundtrack/audio mix?
I guess this is more of a workflow question. I wonder for example how big feature films do. If they export the final output straight from Resolve or not. It seems to me that it would be preferable to do it from Resolve to avoid re-compression. I know big budget features use uncompressed but at least to avoid another step?”
[MichaelMaier] “This is what I’m talking about. I think rendering out of Resolve as DPX 16bit (at least) would be best. Maybe EXR is even better. But then there’s the question of where to finish it. Not all NLE work well with DPX let alone EXR.”
The budget DIY solution is to export ProRes 4444 and take it into FCP 7, FCP X (which actually has some pretty cool abilities with respect to routing multichannel audio and batch-generating multiple deliverables formats) or perhaps Premiere Pro.
Technically, yes, you are losing something with ProRes 4444 vs. DPX. As a practical matter, you will likely never be able to tell the difference. ProRes 4444 is widely used as a mastering codec in the indie film world. It’s similar in quality to HDCAM SR, which has generally been considered an acceptable master format. In fact, the last time we delivered a product to have a DCP made, we had prepped a DPX sequence, and our contact at the other facility basically told us that yeah, they can handle DPX, but the vast majority of projects are delivering ProRes these days.
As for generational quality loss from compressing twice, that’s not really a big deal. First, ProRes is specifically designed to minimize generational quality loss, for just these sorts of scenarios. Secondly, while I’m not sure how Premiere handles things, in FCP 7 and X, if you export a sequence containing ProRes footage to a ProRes file, re-compression only occurs where necessary — for instance, if you’ve layered titles on top of the video. If a segment of video is untouched, the ProRes data is simply copied from the input file to the output file, rather than being decoded and then re-encoded.
If you do want to master to DPX, though, Premiere Pro does support it, although I haven’t tested this workflow. The next step up in terms of price would probably be Smoke.
Depending on the types of projects you work on, however, mastering in Resolve may indeed be viable. Note that Resolve supports alpha channels in ProRes 4444 files, which gives you a pretty easy way to get title over image, etc. You can bring in audio, sync it up, and embed it in your outputs. The main limitations with this approach are a) you’re going to be rendering out a bunch of elements in other apps and bringing them into Resolve, where in a fully-fledged finishing tool you could do everything in one app, b) if you need fancy audio channel configurations, etc. in QuickTime outputs there’s no way to set that up, and c) if your program is long enough to have been broken up into reels, you might need to splice them together in an external app. Although the last limitation goes away if you’re rendering to an image sequence format; then if you setup your timecode correctly you can just render all your reels to one folder and get a continuous sequence. (Though note that for e.g. mastering a DCP the facility you deliver to may prefer reels.)
—
Digital Workflow/Colorist, Nice Dissolve.You should follow me on Twitter here. Or read our blog.
-
Hey Chris. This is a great reply! It’s exactly the type of reply I was expecting to get when I started this thread. Extremely useful and insightful. Thanks for that!
Please see my replies bellow.
[Chris Kenny] “The budget DIY solution is to export ProRes 4444 and take it into FCP 7, FCP X (which actually has some pretty cool abilities with respect to routing multichannel audio and batch-generating multiple deliverables formats) or perhaps Premiere Pro.”
I’m totally PC based now. So FCP is a no go. I don’t use Premiere either.
[Chris Kenny] “Technically, yes, you are losing something with ProRes 4444 vs. DPX. As a practical matter, you will likely never be able to tell the difference. ProRes 4444 is widely used as a mastering codec in the indie film world. It’s similar in quality to HDCAM SR, which has generally been considered an acceptable master format.”
But in my experience when using Prores files in Resolve, at list in Windows, it seems to clip the highlights. The same for example doesn’t happen to DNxHD. After importing it to the media pool, if you right click on the file> clip attributes> and choose data levels instead of auto, the clipped highlights in a DNxHD clip shows the full range and are brought back, showing highlight detail that seemed lost. The same doesn’t happen to Prores clips. I’m talking about clips captured by a camera in either DNxHD or Prores. Not clips converted to Prores from another codec.
From talking to several other people who also experience the same problem, it seems to be a Prores problem. On top of that, Prores on windows can have gamma shift problems etc. I think it’s a great codec, but only if you are on a Mac. Just my personal opinion. For that reason my whole workflow is DNxHD based, at least till it hits Resolve.[Chris Kenny] “In fact, the last time we delivered a product to have a DCP made, we had prepped a DPX sequence, and our contact at the other facility basically told us that yeah, they can handle DPX, but the vast majority of projects are delivering ProRes these days.”
Maybe their motivation would be that Prores is probably much easier for them to handle than a DPX sequence.
[Chris Kenny] “As for generational quality loss from compressing twice, that’s not really a big deal. First, ProRes is specifically designed to minimize generational quality loss, for just these sorts of scenarios. Secondly, while I’m not sure how Premiere handles things, in FCP 7 and X, if you export a sequence containing ProRes footage to a ProRes file, re-compression only occurs where necessary — for instance, if you’ve layered titles on top of the video. If a segment of video is untouched, the ProRes data is simply copied from the input file to the output file, rather than being decoded and then re-encoded.”
I don’t know. I’m really paranoid with the compression thing. I guess it’s trauma from the old DV/DVCAM/DVCPRO days. 🙂
To be safe I would rather go DPX or at least TIFF.[Chris Kenny] “If you do want to master to DPX, though, Premiere Pro does support it, although I haven’t tested this workflow. The next step up in terms of price would probably be Smoke.”
But what about After Effects for example? I have heard of people finishing on it and if we can even start thinking of doing it on Resolve, After Effects could also do it, maybe even better since it has a better timeline and you can do any graphics and titling there. I would still grade in Resolve though. But instead of going back to the NLE, After Effects would maybe make more sense?
There’s also Blender 3D, which is the closest thing we have to Smoke and even closer than AE since it has a proper timeline and video editing function, along with 3D CGI, powerful node based compositing, matchmoving, tracking, rotoscoping, color grading etc.
[Chris Kenny] “Depending on the types of projects you work on, however, mastering in Resolve may indeed be viable. Note that Resolve supports alpha channels in ProRes 4444 files, which gives you a pretty easy way to get title over image, etc. You can bring in audio, sync it up, and embed it in your outputs.”
That’s interesting.
About audio, would it really be enough, enough channels etc?
[Chris Kenny] “The main limitations with this approach are a) you’re going to be rendering out a bunch of elements in other apps and bringing them into Resolve, where in a fully-fledged finishing tool you could do everything in one app,”
Yes, but if finishing on a NLE you would still need to be rendering out a bunch of elements in other apps. At least in Resolve you eliminate one more step and also one less compression.
I think that there’s no discussion that using a fully-fledged finishing tool is much better than doing it on Resolve. But the point here is if one doesn’t have such a tool as Smoke. What do after Resolve or in Resolve? I think this workflow you are talking may be very viable. The question is if it would be viable for long form, such as a feature film. But sounds interesting nonetheless. Good idea.[Chris Kenny] “b) if you need fancy audio channel configurations, etc. in QuickTime outputs there’s no way to set that up,”
Could you elaborate a bit on that? I’m not sure I get it.
[Chris Kenny] “c) if your program is long enough to have been broken up into reels, you might need to splice them together in an external app.”
Does anybody break shows into reels anymore these days? I thought that was only important if you were going to transfer it to 35mm.
[Chris Kenny] “Although the last limitation goes away if you’re rendering to an image sequence format; then if you setup your timecode correctly you can just render all your reels to one folder and get a continuous sequence. (Though note that for e.g. mastering a DCP the facility you deliver to may prefer reels.)”
Really? I wonder why since at the time of projection there’s no need to change reels and it will be a straight through shot.
Chris, thanks again for taking the time to type that great reply. It already helped loads and is changing the way I was thinking of approaching my workflow. The thing is this is for a feature film for the festival circuit that I may get in, in a month or so. Since I have never done this before I need to re-access all my workflow, specially finishing.
-
[MichaelMaier] “From talking to several other people who also experience the same problem, it seems to be a Prores problem. On top of that, Prores on windows can have gamma shift problems etc. I think it’s a great codec, but only if you are on a Mac. Just my personal opinion. For that reason my whole workflow is DNxHD based, at least till it hits Resolve.”
Hmm. We grade Alexa-originated ProRes 4444 (on Windows) on a fairly regular basis, without apparent issue. In terms of outputs, we’ll just move finished projects to a Mac if we need to render to ProRes — and we’ve found smaller gamma shifts (no visible difference, very slight change on scopes) with ProRes than with essentially any other QuickTime codec, including QuickTime-wrapped DNxHD and Blackmagic’s 444 Uncompressed QT codec.
Although QT is infamous for gamma shifts in general; I certainly understand the temptation to avoid it. It just opens up the ability to use so many additional tools. For instance, if you want to encode a DVD, you severely restrict your options for doing that if you require an app that can do it directly from a DPX sequence.
[MichaelMaier] “I would still grade in Resolve though. But instead of going back to the NLE, After Effects would maybe make more sense?”
If you need its toolset After Effects may be a sensible option, but I think to some extent it leaves you with some of the same sort of limitations Resolve has in terms of configuring audio, etc. In fact, last time I checked I couldn’t figure out how to even have After Effects play audio except with a RAM preview, which might make it a little hard to watch through your finished piece prior to export.
[MichaelMaier] “There’s also Blender 3D, which is the closest thing we have to Smoke and even closer than AE since it has a proper timeline and video editing function, along with 3D CGI, powerful node based compositing, matchmoving, tracking, rotoscoping, color grading etc.”
As far as I know Blender doesn’t support video I/O hardware, so the question again is, how are you going to be able to watch the piece with ‘real’ monitoring as you’re working on it?
[MichaelMaier] “I think this workflow you are talking may be very viable. The question is if it would be viable for long form, such as a feature film. But sounds interesting nonetheless. Good idea. “
We generally fully conform features in Resolve, even when we’re outputting to ProRes and/or generating final deliverables in another app. Why? Because we never know when clients are done making changes. Sometimes they’ll want to come back months after a ‘final’ output, with a slightly tweaked edit, or a desire to take the grade in a particular scene in a different direction. If your workflow involves rendering out of Resolve and then doing a bunch of additional conform work in another app, and then you later need to make changes in Resolve, you may have to painstakingly transfer over or redo that additional conform work with the new set of Resolve outputs. So we’ve found that while it might seem like more work up front to get every title, every VFX shot, etc. dropped into place in Resolve, it saves time in the end.
It also means the client can just watch the whole movie through like a movie (well, except for reel breaks; see below) while it’s still easy to change everything.
[MichaelMaier] “Could you elaborate a bit on that? I’m not sure I get it.”
A QuickTime file doesn’t just contain a particular number of audio tracks; those tracks can be configured in specific ways. So, for instance, having six mono tracks in a QuickTime movie is not the same as having a 5.1 mix in it, or three stereo pairs, or whatever, despite the fact that all of those are six tracks. As far as I’m aware, while Resolve can export pretty much any number of audio tracks, there’s no way to specify what they are like this.
Honestly I’m not sure e.g. Smoke actually addresses this either, though. By far the easiest way we’ve found to handle it is via roles in FCP X. Of course you might not actually need to care about this when delivering many projects.
[MichaelMaier] “Does anybody break shows into reels anymore these days? I thought that was only important if you were going to transfer it to 35mm.”
Long-form projects tend to bog down a bit in Resolve if they’re not split into reels, particularly with respect to the time it takes to save. We generally break features up into reels of 300-400 shots each.
[MichaelMaier] “Really? I wonder why since at the time of projection there’s no need to change reels and it will be a straight through shot.”
DCP itself actually has a concept of reels. That is to say, within the DCP package there are usually multiple reels. The media server strings them together seamlessly during playback. I don’t believe this is actually a technical requirement of DCP (that is, I think you could deliver a single reel feature if you wanted to), but it’s a widespread practice.
Breaking things down into more manageable chunks tends to make it easier to troubleshoot, and may mean having to redo less rendering/encoding if things go wrong somewhere.
—
Digital Workflow/Colorist, Nice Dissolve.You should follow me on Twitter here. Or read our blog.
-
>DCP itself actually has a concept of reels. That is to say, within the DCP package there are usually >multiple reels. The media server strings them together seamlessly during playback. I don’t believe this is >actually a technical requirement of DCP (that is, I think you could deliver a single reel feature if you >wanted to), but it’s a widespread practice.
It’s the standard way of doing things for the reasons you mention and others. In theatrical features, there are changes going on right up until release time. There are also multiple versions, some of which have different material. The DCP operates with the concept of a composition play list, which specifies which frames to play from which files in which order. This allows for packages to be created that have the same “essence” files, but different files for things like audio, subtitles, etc.. Not only that, but you can create “supplemental” packages that allow you to insert sequences or shots from different essence files in the proper place during play out. Having the picture finished by reels allows this kind of flexibility to be much simpler to implement and program. You can create a DCP with one full length essence track, but no studio level theatrical features are done that way.
-
Another outstanding post Chris! Thanks.
[Chris Kenny] “Hmm. We grade Alexa-originated ProRes 4444 (on Windows) on a fairly regular basis, without apparent issue.”
Maybe the Alexas are set to REC709? This highlight clipping I mention with Prores only happens with clips recorded in full swing. REC709 will already probably shrink that range to fit within video levels.[Chris Kenny] “and we’ve found smaller gamma shifts (no visible difference, very slight change on scopes) with ProRes than with essentially any other QuickTime codec, including QuickTime-wrapped DNxHD”
I only use DNxHD wrapped in quicktime since many programs have a problem with MXF. Maybe I should do some more tests but I haven’t found any significant gamma shifts with DNxHD QT.
[Chris Kenny] “As far as I know Blender doesn’t support video I/O hardware, so the question again is, how are you going to be able to watch the piece with ‘real’ monitoring as you’re working on it?”
I really have no idea. If it doesn’t then yes, could be a problem. Although I would never grade in Blender. I’m in love with Resolve. 🙂
But they made a whole movie in Blender called Tears of Steel. It’s the latest open movie from the Blender Foundation and it’s live action+VFX. So somehow they managed it.
Besides the potential I/O problem, it seems to me like the best option apart from Smoke. You got all in there in one app and you can do real video editing.
[Chris Kenny] “We generally fully conform features in Resolve, even when we’re outputting to ProRes and/or generating final deliverables in another app. Why? Because we never know when clients are done making changes…So we’ve found that while it might seem like more work up front to get every title, every VFX shot, etc. dropped into place in Resolve, it saves time in the end.
“Sounds very interesting. If you have the time I would love to know the workflow you use for this. Finishing in Resolve seems to be the most appealing to me for some of the same reasons you mention.
[Chris Kenny] “A QuickTime file doesn’t just contain a particular number of audio tracks; those tracks can be configured in specific ways. So, for instance, having six mono tracks in a QuickTime movie is not the same as having a 5.1 mix in it, or three stereo pairs, or whatever, despite the fact that all of those are six tracks. As far as I’m aware, while Resolve can export pretty much any number of audio tracks, there’s no way to specify what they are like this.
Honestly I’m not sure e.g. Smoke actually addresses this either, though. By far the easiest way we’ve found to handle it is via roles in FCP X. Of course you might not actually need to care about this when delivering many projects.”
So you are saying for example that you couldn’t conform and output a show with a 5.1 track from Resolve?
Thanks again Chris. This has been a great chat and I’m learning a lot.
-
[Mike Most] “It’s the standard way of doing things for the reasons you mention and others. In theatrical features, there are changes going on right up until release time. There are also multiple versions, some of which have different material. The DCP operates with the concept of a composition play list, which specifies which frames to play from which files in which order. This allows for packages to be created that have the same “essence” files, but different files for things like audio, subtitles, etc.. Not only that, but you can create “supplemental” packages that allow you to insert sequences or shots from different essence files in the proper place during play out. Having the picture finished by reels allows this kind of flexibility to be much simpler to implement and program. You can create a DCP with one full length essence track, but no studio level theatrical features are done that way.”
Ok. But so it also means that you no longer need to mind the reels during editing right? It can be arbitrarily split anywhere and using any criteria. Because back in the film projection days, unless you wanted the audience to know exactly when the projectionist would change the reel, you had to mind it during the editing and have the splits in places that would make sense.
Reply to this Discussion! Login or Sign Up