> DaVinci Resolve 21.1 now supports integration with AI assistants such as Claude, Claude Code and ChatGPT Codex, allowing customers to analyze projects, organize media, adjust settings and batch render using everyday conversational language. Customers can ask their AI assistant to create highlight edits from long form video, remove unwanted clips and render deliverables, so they can spend more time being creative rather than getting bogged down in repetitive tasks.
These 'integrations' everywhere just mean we get more APIs that can be accessed programmatically, which I think is good for users, even if you don't plan to use to use AI. It's like the optimistic web 2.0 APIs everywhere days
No no, if you've ever spent serious time editing, it's a big deal -- if they pull this off correctly, it will be a game changer.
Editing has so much grunt work, where you just have to spending time organizing, collecting files, renaming things, finding missing files... Just tedious project management stuff that is excellent for an AI to do.
It’s because those codecs require SDK’s that are expensive to license. That’s not a Linux specific issue - it’s always been that way for the non-studio version of Resolve (with some exceptions on MacOS since the ProRes codecs are already part of the OS).
Primarily because Resolve is meant as a tool for professionals - and as a result it can’t get away with using things like Ffmpeg under the hood.
Ffmpeg (for example) doesn’t use officially licensed codec SDK’s to handle certain codecs - it uses reverse engineered tooling.
If you’re working on a personal project or something that is just getting pushed to YouTube that won’t matter 99% of the time.
However if you are working on a professional color grading pipeline for a project that might go to broadcast or into theaters, there are very specific QA tests that the final rendered product must pass - and these unofficial reverse engineered rendering tools often result in rendering errors that can fail these QA tests (and in a way that is basically impossible to troubleshoot).
It’s already enough of a challenge to manage a color pipeline using the officially licensed tools - they can’t really risk that kind of issue with their pro and semi-pro customers.
That said, it’s important to note that MP4 is a just a container/wrapper, not a codec.
So an mp4 file might contain video (or just audio) that is rendered using one of many possible codecs depending on what was chosen by the person doing the render. That will determine whether Resolve whether the free edition of Resolve will support it or not.
Haha yeah was about to post this very comment. I'm glad to see they have their priorities in order instead of checks notes supporting MP4 on Linux, the most common type of goddamn encoding. GoPro footage literally does not work with Resolve outside Windows.
This is one of the the cases where “the language model won't necessarily respect IP law” is really relevant. Individual users may be able to get away with this because they're too scattershot to sue, and users in some jurisdictions may actually be fine regardless, but broadly, the patent licensing around things like H.264 is not something BMD can afford to ignore. AFAIK the biggest reason Google was pushing WebM in the first place was because they, a huge corporation, were balking at being under the thumb of the MPEG LA!
I appreciate that it can be confusing for users that patents can still apply to techniques where implementations exist that are both libre and gratis from a copyright perspective, but this has been an issue for Linux-adjacent software distribution for ages—like, see how Debian used to put certain stuff that was encumbered by patents in the US but not in other jurisdictions in special “non-US” repository sections, that sort of thing. It's not Black Magic's doing that the codec space is like this. Don't assume it's just lack of engineering capacity or denying you for fun.
Linux Vivaldi builds use libffmpeg.so from chromium-ffmpeg, which includes decoders for most codecs used on the web (no h265/mpeg2 tho), and VA-API which will handle all hardware accelerated codecs on particular platform.
This looks similar to using Microsoft provided Media Foundation codecs. No idea why Black Magic doesnt do it like Linux Chromium browsers.
Do you have any information on how the actual patent licensing is set up for Vivaldi? I wasn't able to find much on it, though I see the note in https://wiki.archlinux.org/title/Vivaldi that it “automatically downloads” the relevant files, which is at least suggestive that they can't just redistribute the binaries normally. The other closest thing I was able to find was https://github.com/chromiumembedded/cef/issues/3559 which describes a specific proprietary_codecs option in CEF, which is not quite the same thing but ought in some sense to be related… the way they describe having to set ffmpeg_brand=Chrome normally for that to be available doesn't sound promising?
I would be moderately surprised if whatever arrangement is going on there allows full use of the codecs for video editing software and not just for consumer decode of Web resources.
That doesn't sound right to me at all. I'm pretty hazy on the specifics of VA-API, but I wasn't under the impression that the graphics hardware carries 100% of the decode implementation nor had codec licenses attached to it, and this Phoronix article from 2022 describing Fedora having to back off from having accidentally shipped H.264 via VA-API without appropriate licensing seems to be evidence in the direction I expected. https://www.phoronix.com/news/Fedora-Disable-Bad-VA-API
If there were definitely no legal encumbrances, surely they'd want to keep shipping the version with the proprietary codecs in, because that would make them a superior option for users who need those? AFAICT someone upstream thought that having the toggle to avoid patent issues was important, which if true nixes the idea that VA-API just gives you an automatic unencumbered route and you don't have to worry about it after that.
I'm not really sure what actual claim you're making regarding the relevance of “pro-patent views”, but I might interpret it as something like “Red Hat is more likely to play it safe, whereas less IBM-enculturated distributions are more likely to bet on interpretations that let the users take on the risk”. Which I think is probably true; for instance, https://www.reddit.com/r/linux/comments/xqn20n/comment/iqbmv... describes how other distributions chose differently and moved to re-enable the proprietary codecs by default. (Some of them seem to be based in friendlier jurisdictions, mind. I'm not completely sure where BMD is, and I'm not sure how much it matters given they want to sell worldwide, but their full corporate name resolves most strongly to Australia: https://opengovau.com/business/30098098287 and Australia doesn't seem to exclude software from patentability: https://www.ipaustralia.gov.au/patents/what-are-patents/what...)
Anyway, if that's what you meant, I don't think that distinction is nearly as applicable to the situation with Resolve. BMD isn't a producer of a Linux distribution containing a broad variety of stuff, where “users happen to do stuff with the open source software we provide that violates some patents” can maybe be waved off as not their responsibility. BMD is very obviously directly operating in the commercially-distributed professional video tooling market, so codec licensing will be something they have to deal with as a keystone part of their business. Adding something to their software to actively enable people to do professional-style encoding for those codecs without the patent licenses would be playing with fire.
Seems like an unusually boring Resolve release. There's nothing there I particularly want to try out. Nothing wrong with that, I've just got used to every release having a ton of shiny new stuff in it.
I guess the agentic stuff is probably exciting to somebody.
I've had Claude Code drive DaVinci Resolve before in three ways:
- Generate .otio files that are imported. Unfortunately, .otio doesn't fully support everything Resolve can do.
- Resolve has a scripting engine. Claude can write Lua scripts, then execute the script (through UI Automation)
- Computer Use. It's quite token intense, but it works.
I didn't try it, but for the paid version, someone also built a MCP server based on the scripting engine.
I'm programmatically adding analysis to sim racing lap guides, therefore my flow is probably different from the usual editing flow, but I have to say I found Claude Code + Opus to be quite capable of making edits with computer use.
I used IntelliScript a lot. I give Claude a subtitle export, ask it to cut it and import the cut subtitle track with IntelliScript. Resolve then creates a video version that’s cut based on the subtitle changes
I helped my girlfriend use the MCP server as she had a project where all the media files had been moved to a new source location, and using Codex, it managed to hunt down all the files and stitch it all back together. Ended up saving quite a lot of manual boring labour, and was quite cool to see it work in real time:)
I spent a few tokens last year trying to create a Python lib for native (undocumented) Resolve files. I was making progress, but it was slow (I ended up scripting Resolve in different ways). I imagine that current models could reverse-engineer the native file format relatively quickly.
I've been using it for many years, and it's grown in power and functionality with every release. BMD has offered free upgrades to Pro users (with no subscription) for many years, which is refreshing given the current state of software licensing.
For those upset with the new agent integration, keep in mind that one of the big knocks on DaVinci Resolve is that it's hard for beginners. Powerful tools like color grading and nodes take years to truly master (and that won't change), but making it easier for users to collaborate with the agents they already use to help learn and perform basic tasks should help with the learning curve.
Ultimately, professional video editing is an art. An agent might save time with asset management, project/timeline setup, etc., but it won't replace the real experience and taste required for larger projects. I would rather spend the extra time on the nuanced/deeper creative work.
No offense whatsoever. But this whole line "but it won't replace the real experience and taste required for larger projects" is not how I see it. I wish people are more humble at this juncture of AI progress. This is not to say humans won't play a part but saying xyz cannot be replaced is a big stretch. Saying its not there currently is okay but saying it won't replace I don't see where that's coming from.
> But this whole line "but it won't replace the real experience and taste required for larger projects" is not how I see it.
Why not? Personally (although I'm not saying others have to see it this way too), I'd wait until I can see what AI can replace, before believing so, rather than the opposite.
So far, AI isn't able to "set direction" at all in any sensible way, especially for a creative project like TV or movies, the results are just too bad (today). Although it's excellent at "following directions", which is why these new APIs are great and exciting, it lets people who are good at setting direction to leverage these tools to have the direction concretely followed by something, without having to create a group of collaborators.
Until we've seen proven that AI can indeed "set directions" ("have taste"), I think it's fair to assume they're unable to do so in a way that it produces something that humans generally enjoy. And very far from the kind of masterpieces a human can create.
> it lets people who are good at setting direction to leverage these tools to have the direction concretely followed by something, without having to create a group of collaborators
It lets the unskilled access the tools of the skilled without having had to learn all the pitfalls, traps, and safety concerns. With video editing there may not be much at stake but we've already seen how sloppy code ~= vulnerable code when the prompter does not know what they are doing. What's going to happen when project managers start vibegeneering cars and buildings.
The way I see it, it’s all about how you use it. Let people have tools; their output stands on its own legs.
As pro software, the UI has a learning curve. Many YouTube videos are out of date, or unnecessarily drawn out.
Additionally, as a hobbyist, project settings, color spaces, clipping, etc can be tricky to miss.
I’ve already been using Claude to (1) help me with UI, (2) QA my delivery, and (3) occasional fixes for audio repair; waveform sync (can’t afford time codes).
I make all creative decisions - including most menial ones - myself, I wouldn’t have it any other way; I enjoy the act of creating. I welcome this integration.
I’m sure some users are gonna agentically grade or whatnot. I don’t like it, but I’m also don’t believe in telling other people what they should do with their own work and creative process.
I would gladly enjoy to collaborate with an agent if I can learn more about video editing.
I edit my video only for passion and to learn something new though, so I guess that professionals may see this in a different way.
But I don’t think that it’s much different from how Claude code revolutionised the software scene.
Same. I edit lots of stuff in DaVinci, but am quite bad at it. As in, I spend forever figuring out how to do the stuff I want to do. I don't need the AI to invent stuff for me or take any decisions, just help me achieve my vision. Often it's quite simple stuff, where I'm forced to watch a too long video tutorial to figure out how to do something. But if I instead can ask it "help me to configure it so that the subtitles get baked into the video" or whatever it would be nice.
I understand what you're saying and I agree, ostensibly, but on the caveat that it's hinged on the "we don't know what we don't know" aspect of reality. If we restrict our domain to what's forseeable, it's still apparently the case. Astra's demos have me scratching my head, diminishing returns are kicking in harder and faster than anticipated. It seems we've entered a phase to where frontier companies have completely abandoned improving the substrate because they can't justify it. It seems like it's all just harnesses and reasoning orchestration at this point. Hell, everybody but Google seems to have given up even on projection layer improvements (and Gemini is unusably bad in other ways).
Watching this next generation closely, because there are just a shitload of red flags starting to crop up that don't seem to be going away.
The person you are replying to seems to have a depth of experience with editing over many years.
It seems pretty clear that AI capabilities are jagged. There are good reasons to believe based on research and the subjective evaluation of experts who are quite friendly to AI that there are some tasks creative tasks that AI has been really quite good at (totally disrupting stock illustrations) some where it does so much better than typical person, but not nearly as well as well as an expert (song creation) and some where, so far, it pretty much fails, and sees almost no progress even as models get better and better. (long form creative writting). Indeed, writting style sometimes seems to fall backwards with recent releases.
Also, it's subjectively, but commonly held to be true that most of the content created when AI can execute well lack taste, unless the human creating with AI has it. The term "slop" exists and in legible for a reason.
I mean, maybe this is just a question of different people speaking about different time horizons.
But it seems perfectly sensible for someone to say that there is a lot of automation that can happen in the near term, and that they see no reasonable evidence that AI can do interesting or even coherent work this space.
I don't know that shallow accelerationist boosterism (or panicked doomer despair ) that handwaves away the distinctive variations AI performnce is that enlightening.
Taking away some of the more monotonous and "uncreative" tasks in any job can be taking away some of the variety in the work. Variety is nice. Variety affords us downtime. As the job gets stripped down to its core function, while a machine keeps the pipeline full so that the work can continue uninterrupted, the downtime slips away.
And great for beginners, now they get to pay some chatbot company to cover the gaps left by detereorating software design. Lucky they're already using agents... are they?
This feels inverted to me: downtime affords us variety. If I have little downtime then I'll strongly prefer known, efficient methods of doing something, rather than experimenting with new ones.
> As the job gets stripped down to its core function, while a machine keeps the pipeline full so that the work can continue uninterrupted, the downtime slips away.
I have to disagree. You are already many levels away from the actual pixels. This is what software is for. Drawing a line for you personally is dismissive of the rest of the user base. Just because there is a so called agent added to the mix doesn't remove the human. It includes far more humans and enables them to use the tool. This is design if done right. Also you seem to be critiqueing the concept not the implementation.
I don't sense that you have thought this out and you are cycling talking points even if you don't mean too.
I have a hard time seeing your way. I build software. Tools for humans to do stuff. Whatever they want. It gets complex. There are methods to make it more accessible to more people. I want to go to there.
I feel you bro that the human is getting squeezed out. I do not see it that way in the sense that the human no longer is the nexus.
CEOs and tech bros and hype are all a distraction. You are a valuable person in the conversation and we all need you to participate with eyes wide open. I assume you are young. That's not discounting it is welcoming.
I've been thinking a lot about this across many different mediums, not just filmmaking, and I think it's one worth digging deeper on.*
What is lost when losing the (understandably annoying) process of the grind? By letting LLMs do the busy work, do you lose the opportunity to stumble into randomness? Do you lose the understanding of what is possible/interesting with the medium?
And do you end up in this derivative future where decisions are made by a system that was trained on all the previous uses of it?
Obviously that's fairly hyperbolic, but I have absolutely felt the pull of it all. When it decides to take a couple extra steps I didn't ask for creatively and I think "eh, that's fine."
*EDIT: I wrote this and looked back and this SOUNDS like an LLM wrote it. See what I mean??????
Not only are the upgrades free, the full license is still only $295 and can be freely migrated between macOS, Linux, and Windows (unless something has changed, serial numbers can be activated on two systems at a time and transferred freely so long as you have Internet access).
I'm no expert, but I never found Resolve materially more difficult to learn or use than Final Cut Pro or Premiere. And if you're doing anything moderately complex, it's potentially easier to learn because things that require external tools in FCPX and Premiere are built in to Resolve Studio.
$295 one time permanent license with free upgrades and can be migrated between different computers?
I don't have a single reason to do any form of video editing, let alone professional video editing, but that kind of pricing model makes me want to buy a license just to make sure that companies like this stick around.
Black Magic supports all three platforms and they’re not exactly a large company. They show every day that there is no excuse/reason not to support multiple platforms. Anything else is just an excuse by a lazy developer. The developer for Blender is also the same. A user of both programs.
Black magic also make a program that support/ties into the camera hardware on smartphones and they support both Android and iOS.
> $295 one time permanent license with free upgrades and can be migrated between different computers?
I've been using Resolve since version 14, still getting upgrades to it! My license does only work on one "computer" at a time though, so when I switch between Linux and Windows, I need to re-enter and activate the license, as when I do that it deactivates the previous (other OS) one. You have seemingly unlimited amount of times to switch though, as I've probably done it on a bi-weekly basis for quite some time.
I don't know if something has changed, but if you buy resolve now you can have it activated on two computers at the same time. I've got my resolve studio license activated on my mac laptop and linux workstation at the same time.
Having a powerful video editor you can grab any time you need to is useful in a way few other bits of software are. Sure, for trimming and even some basic editing or watermarks, ffmpeg works great, but there are times when having full editor ui is beneficial
Plus you get the license for free when you buy a camera from them!
I’ve been using DaVinci Resolve Studio for 12 years now with frequent updates, with the license I got with the original Pocket Cinema Camera for £800 (still a fantastic pocket size cinema camera, shoots raw cinemadng and ProRes with a beautiful sensor).
It really is amazing. DaVinci used to be something that required a dedicated workstation with proprietary hardware cards, and most used it via renting editing bay time in studios equipped with it. Like just buying one of the workstations and then renting it out was a viable business model if you had teh startup capital (and I randomly knew someone who operated such a business years ago).
For me it was the other way around - bought the license and got the speed editor for free. Though it may be a different model, I think the stand-alone price for it was way less than $400.
yes to agents! how i had wished i could just point an agent at my footage and instruct "do something sensible with color grading and audio cleanup". i just want something presentable for a hobby video podcast. it doens't have to look a million bucks, and i have no time or desire to learn the intricacies involved. just something that would get me 70% (or 50%!) the way there would be more than fine.
Resolve is not any harder to use than any other NLE.
Although... BMD has actually made it harder, by adding bloat like the Cut page. I have attempted to find value in it a couple of times, and didn't. Now I disable it.
BMD needs to stop throwing crap into Resolve and buckle down and do two things:
1. Fix bugs, including UI defects. The Deliver page is a mess. The clip selection between the Edit timeline and the Fusion page makes no sense. And lots more.
2. Unify the node views. Resolve is a bunch of stuff BMD bought and threw together on different pages. The "integration" is janky and can, especially for the beginner, seriously degrade your work. Once example is how Fusion compositions inexplicably downgrade everything to timeline resolution. So you can bring 4K footage into a Fusion comp, but if your output timeline of the moment happens to be merely HD, the entire composition (including keys) will be degraded to work at that resolution. I mean... really? And beyond that, the relationship between Fusion comps and the timeline events they wind up in is extremely fragile and breaks arbitrarily and unrecoverably all the time.
A single nodeview for all operations would be a long-overdue game-changer for editing software.
Have you ever considered that you aren't the target audience for the cut page? It's supposed to be paired with a speed editor and is mostly intended for assembling rough cuts. You probably aren't cutting features or television or multicam so it's usefulness is probably out of your range of use.
Helping more beginners might upset some who have put in the labor.
Over time many things become easier to learn and do, and those who might feel upset should make sure their own feet are moving as well in terms of their learning and growing.
I view the agent integration (bearing in mind I haven't tried it yet) as an easier way to replace writing python scripts to do repetitive tasks, for people doing particularly repetitive work. A macro recorder might be more optimal to the anti-AI crowd, but this is probably easier to implement for the developers, while being more flexible for the end users.
I assume that this can be connected to local models, not just cloud models.
I've edited well over a thousand hours of video on my (Debian) DaVinci Resolve workstation. It's a rock-solid video editor that I can spend a day in without having to worry about a crash.
But I desperately wish Blackmagic would bring support for VST3 plugins and JACK to Linux. Or, at the very least, allow me to use my MIDI control surface in Fairlight on Linux.
As it stands, I have to do all of my audio production in Reaper, and that's a bit cumbersome.
> It's a rock-solid video editor that I can spend a day in without having to worry about a crash.
Is that praise? I would hope the bar is higher.
I bought a used computer four years ago to edit 4K video, tens of thousands of raw files off my big Sony cameras, write, light code and research. Many thousands of hours of hard use.
In five years I’ve never had a single program crash and certainly not the whole system go down. Uptime is many months.
I’ve been using it in Linux for >2 years, including their special cutting keyboard (blanking on the name).
Works without a hitch, including multi monitor, gpu acceleration etc
I use it on Manjaro, installed via AUR.
That said, IIRC it only works with nvidia GPUs and requires X11, and is supposedly finicky to get the environment right for (but the AUR gets it right for me)
I don't know about recent versions, but I managed to install 18 on Linux Mint.
Making it run wasn't easy, and it can't decode MP3 and some video codecs because of licensing issues. It also can't decode WebM. This is kind of a problem because of a lot of free music you can download to add to videos and a lot of video clips you can download from Pexels for example are encoded with codecs it can't decode, so you have to re-encode it with FFMPEG first to a supported codec.
The biggest issues are the supports of video cards (nvidia is the king) and different popular codecs. The problem is partly solved by using paid version, but the gpu/codec support is still not even with Windows, so you have to use ffmpeg to re-encode from/into supported formats.
Tried to work with ut for 5 months on Ubuntu. Installation was painful. Have to use some tutorial. No AAC audio codec support as well, which is the code of my smartphone camera. Randomly decided to not work after Ubuntu updates. I'm on Windows now...
Runs flawlessly on my strix halo Ubuntu 24 machine after fixing the install process [1]. I have an ffmpeg script that batch converts all my files into the supported codec.
The good thing about DaVinci Resolve's Linux system requirements is that these are so outdated that basically every developer links their plugins with static libraries they build themselves; and Resolve itself bundles lots of .so files with the app itself. Also Rocky Linux 8 is a good baseline Linux distribution: simple system libraries interdependence, so trivial to export and bundle those.
In other words, binaries created for RockyLinux8 with portability in mind are easily portable to sane modern Linux distributions.
(That doesn't make developing on Rocky Linux 8 fun at all)
As an anecdote, I recently needed to colour-grade some short footage for personal purposes. My workflow at the time was to essentially ask an LLM for guidance on how to use DaVinci Resolve to perform the task, step by step. I would make an adjustment, take a screenshot of the controls and the result, and repeat that process a handful of times, eventually ending up with a pretty nice result.
I can definitely see the utility of something like this for cases where a media budget just wouldn't have existed in the first place.
I don't see how companies like this can survive going forwards. DaVinci Resolve is effectively ~200 Astra/Fable prompts, maybe even fewer if you're willing to forgo polish.
I'd give you $100K no questions asked if you manage to get something even close to Resolve with 2K Astra/Fable prompts, polish included. You vastly underestimate how much work goes into "polish" when software with huge surface area is decades old.
I think there's a really interesting tension between what you're saying and what the parent post is saying, and it's something that's been discussed for a long time but LLMs are throwing a new wrench into it: a given user probably only uses 5-10% of the features of a piece of software, but across the whole body of users you get close to full coverage.
I use DaVinci Resolve for editing GoPro, Canon DSLR, iPhone, and Drone FPV footage. I import clips, use the razor to cut them and rearrange them, occasionally crop and zoom, and then encode them down to h264 or whatever. There are a bazillion features that I sort of know exist but haven't ever had any need to use. Could I make a tool using an LLM that covers my DaVinci workflow? Almost certainly, and I don't think it would take anywhere near 200 prompts. I've made significantly more complex bespoke data analysis pipelines for drone flight telemetry in like... a dozen prompts. But I have no illusions that what I would make as my "DaVinci Clone" would have anywhere near the functionality of DaVinci, just the tiny piece of that functionality that I use.
> a given user probably only uses 5-10% of the features of a piece of software
Then vibe your own 5% editor, but saying that a software so mature and complete as Resolve can be done with some AI and time is something I expect coming from the mouth of a teenager, not an adult.
I'm pretty sure your editing workflow can be achieved in the Preview app on macOS. If your editing workflow is just basic chopping clips out of a longer video then yes, you could vibe code it.
Heh, you're probably right. Something that did make me laugh pretty hard... I had a GoPro timelapse of an IMAX theatre screen being replaced and I thought to myself "surely in the vast sea of DaVinci Resolve features I've never used, un-fisheye has got to be there". Sure enough, I found it, it looked great, and it's the only time I've hit the "you need to buy the pro version to use this feature if you don't want a big watermark" gate.
regardless of whether it’s possible, there is enormous value in the community, ecosystem, and learning materials centred around Reaolve as it is right now. the workflow has been refined over years. no one shot slop effort will have that network effect behind it, and if the workflow isn’t a drop-in replacement it’s dead in the water anyway.
you do know this company also sells myriad of hardware , can't Fable those just yet. and I am sure majority of their income comes from that part of their business, selling equipment to studios etc.
You can say anything, why stop at claiming 200 prompts, why not say you think Astra/Fable can one shot Davinci Resolve! And you can EASILY make the BM cinema cameras and other hardware by prompting Fable to run a 3D printer. (≧▽≦)
How would you make an application designed for creating Hollywood grade cinema easier? The fact Resolve isn't two orders of magnitude more complex is astounding to me!
Resolve is like a top of the line professional camera: there are so many more knobs, buttons, and dials on those because top-end pros run everything manually changing settings as they take photos based on changing light conditions and what the subject is doing. Resolve isn't going to assume what you want and do it: it's a professional tool for professional users. The fact they are still giving it away is amazing to me.
The free tier is very generous and it’s only $300, no subscription. Those Fable credits would cost a lot more than that. You’d spend thousands of dollars to save $300.
Moreover, not everyone wants to have the hassle of building and maintaining their own version of a thing that isn’t compatible with anyone else’s version of the same thing. And if you hire anyone to help you, now you have to train them on how to use your app. If they leave, those skills are useless for their next job.
(I’m not even covering the idea that you could do it in 200 prompts. I got Claude to create a video player and that alone took over 300 prompts.)
Would be SO sick! I feel like there is such a big appetite for it and people would do the free labor to support it. I wonder what keeps it in this horrible purgatory of install hell.
Resolve is extremely RAM hungry. For 4K video I wouldn't want to be editing with less than 12GB of VRAM. I'm guessing that excludes most integrated graphics.
Source - I work as a professional editor (amongst other things). Please commenters, do not come at me with your five minute low bit rate projects that required only a LUT and some contrast and 'worked fine' in DaVinci on a potato.
I should have also mentioned DaVinci also heavily uses CUDA on Windows, and metal on Mac. I'm assuming the functions implemented in these frameworks do have fallbacks on Vulkan or similar, but I doubt they're as optimised. I know there are some issues with Resolve on Linux for example. Any halfway decent Nvidia card with enough VRAM should work fine - i.e.: a 30 series card is grand. But I'd be skeptical integrated graphics would have the horsepower for denoising or a host of other build in effects. Rendering would likely be slow also. Although this is the least important element of the process on modern hardware.
For people like me who have no idea what this is, I looked it up:
DaVinci Resolve is a proprietary application for non-linear video editing, color correction, color grading, visual effects, and audio post-production. It is developed by the Australian company Blackmagic Design for macOS, Windows, iPadOS and Linux.
...
DaVinci Resolve has been used for the color grading and/or editing of feature films such as Alien: Covenant, Avatar, Best of Enemies, Deadpool 2, Jason Bourne, Kingsman: The Golden Circle, La La Land, Love & Mercy, the Pirates of the Caribbean series, Prometheus, Anora, Robin Hood, Spectre, Star Wars: The Last Jedi, and X-Men: Apocalypse.
I'm all for AI features here. The software is so feature packed and conplex, that after years of usage I still need to search and look up tutorial videos every single time I edit something.
How to ease key frames, how to fix my HDR captured footage for SDR, how to export the best quality video, how to burn in subtitles with word highlights, how to fix my audio aaand so on.
Either the AI will show me the particular feature and setting or a YouTube tutorial will. I think the AI might be faster and easier in this scenario.
I used Google's Agent Development Kit to make a super simple agent outside of Resolve that I can use to say "take a 10s clip from file1.mp4 starting at 00:01:25 and a 20s clip from file2.mp4 starting at 00:02:00" then tell it to preview and render. Speeds things up.
Have anyone tried using this as a substitute for CaptureOne/Lightroom? Any experiences? I'm aware it's a quite new feature added to the suite, but would love to one day stop paying the pricey subscriptions the other two alternatives demand
I tried and as a individual (single/one at a time) image editor is it fantastic, beats Lightroom by a mile and the node editor gives me the ability to "emulate" panels from CaptureOne (and OFX plugin support means I can just LLM my own if I need).
Library management is where Lightroom (and honestly CaptureOne) still wins, I couldn't find a good way to manage a 100k photo library.
I created one project and tried importing the whole library, but it took more than 8 hours and didn't make meaningful progress (~10%?) so I stopped (compared to 15hr for the whole thing in Lightroom).
I also did a project for about 1K photos and while the import worked, the library management itself was lacking. I could not find ways to do image stacks for bursts, group by metadata, etc. The obvious filters are there, but I think it needs more work.
That’s what I was afraid of. Thank you for the answer! Seems like I’ll be stuck with C1 for a little longer. Their sessions and culling tools I’ve found quite helpful for the occasional event shoots I do, and I don’t think I’d be ready to give it up-even if it would save me a little on the subscription.
I spent a little while trying. My knowledge of lightroom didn't really carry over. I couldn't work out how to do things like masks. I'm sure if i spent longer and watched some tutorials I'd get further but I decided to just let the photo mode cook a little longer before I try to switch over.
I make motorcycle travel videos for YouTube, and the bulk of where my time goes in the preparation of a timeline: chronologically cutting the highlights, synchronising dual camera footage. If Claude + Davinci Resolve will enable me to cut this time slightly, even if it's 10%, it's a welcome change. Ever since I made the jump from Premiere Pro, I've been positively surprised with how good Resolve is. Excited to try this out!
The Claude connector is a nice idea, but there are lots of Python APIs missing in Resolve.
Even simple interactions like moving a clip 3 seconds in the timeline will not work with the MCP.
I just tried it, and got only the message:
"Resolve's scripting API also has no way to reposition an existing timeline item".
They must be planning to be bought by one of these companies. Otherwise I dont see why they would not just continue with their strategy of integrating AI tools within their product. They just reduced DR to just another tool. This is like opening a five star restaurant in a food court at the mall.
156 comments
[ 0.20 ms ] story [ 14.8 ms ] thread> DaVinci Resolve 21.1 now supports integration with AI assistants such as Claude, Claude Code and ChatGPT Codex, allowing customers to analyze projects, organize media, adjust settings and batch render using everyday conversational language. Customers can ask their AI assistant to create highlight edits from long form video, remove unwanted clips and render deliverables, so they can spend more time being creative rather than getting bogged down in repetitive tasks.
smh
Why can Openshot, Kdenlive, any other Linux-native video editor edit MP4 files but DaVinci cannot, unless you pay for it?
Genuine question.
Primarily because Resolve is meant as a tool for professionals - and as a result it can’t get away with using things like Ffmpeg under the hood.
Ffmpeg (for example) doesn’t use officially licensed codec SDK’s to handle certain codecs - it uses reverse engineered tooling.
If you’re working on a personal project or something that is just getting pushed to YouTube that won’t matter 99% of the time.
However if you are working on a professional color grading pipeline for a project that might go to broadcast or into theaters, there are very specific QA tests that the final rendered product must pass - and these unofficial reverse engineered rendering tools often result in rendering errors that can fail these QA tests (and in a way that is basically impossible to troubleshoot).
It’s already enough of a challenge to manage a color pipeline using the officially licensed tools - they can’t really risk that kind of issue with their pro and semi-pro customers.
That said, it’s important to note that MP4 is a just a container/wrapper, not a codec.
So an mp4 file might contain video (or just audio) that is rendered using one of many possible codecs depending on what was chosen by the person doing the render. That will determine whether Resolve whether the free edition of Resolve will support it or not.
I appreciate that it can be confusing for users that patents can still apply to techniques where implementations exist that are both libre and gratis from a copyright perspective, but this has been an issue for Linux-adjacent software distribution for ages—like, see how Debian used to put certain stuff that was encumbered by patents in the US but not in other jurisdictions in special “non-US” repository sections, that sort of thing. It's not Black Magic's doing that the codec space is like this. Don't assume it's just lack of engineering capacity or denying you for fun.
This looks similar to using Microsoft provided Media Foundation codecs. No idea why Black Magic doesnt do it like Linux Chromium browsers.
I would be moderately surprised if whatever arrangement is going on there allows full use of the codecs for video editing software and not just for consumer decode of Web resources.
I'm not really sure what actual claim you're making regarding the relevance of “pro-patent views”, but I might interpret it as something like “Red Hat is more likely to play it safe, whereas less IBM-enculturated distributions are more likely to bet on interpretations that let the users take on the risk”. Which I think is probably true; for instance, https://www.reddit.com/r/linux/comments/xqn20n/comment/iqbmv... describes how other distributions chose differently and moved to re-enable the proprietary codecs by default. (Some of them seem to be based in friendlier jurisdictions, mind. I'm not completely sure where BMD is, and I'm not sure how much it matters given they want to sell worldwide, but their full corporate name resolves most strongly to Australia: https://opengovau.com/business/30098098287 and Australia doesn't seem to exclude software from patentability: https://www.ipaustralia.gov.au/patents/what-are-patents/what...)
Anyway, if that's what you meant, I don't think that distinction is nearly as applicable to the situation with Resolve. BMD isn't a producer of a Linux distribution containing a broad variety of stuff, where “users happen to do stuff with the open source software we provide that violates some patents” can maybe be waved off as not their responsibility. BMD is very obviously directly operating in the commercially-distributed professional video tooling market, so codec licensing will be something they have to deal with as a keystone part of their business. Adding something to their software to actively enable people to do professional-style encoding for those codecs without the patent licenses would be playing with fire.
I guess the agentic stuff is probably exciting to somebody.
I didn't try it, but for the paid version, someone also built a MCP server based on the scripting engine.
I'm programmatically adding analysis to sim racing lap guides, therefore my flow is probably different from the usual editing flow, but I have to say I found Claude Code + Opus to be quite capable of making edits with computer use.
Works pretty well!
Yosh is that you?
For those upset with the new agent integration, keep in mind that one of the big knocks on DaVinci Resolve is that it's hard for beginners. Powerful tools like color grading and nodes take years to truly master (and that won't change), but making it easier for users to collaborate with the agents they already use to help learn and perform basic tasks should help with the learning curve.
Ultimately, professional video editing is an art. An agent might save time with asset management, project/timeline setup, etc., but it won't replace the real experience and taste required for larger projects. I would rather spend the extra time on the nuanced/deeper creative work.
Why not? Personally (although I'm not saying others have to see it this way too), I'd wait until I can see what AI can replace, before believing so, rather than the opposite.
So far, AI isn't able to "set direction" at all in any sensible way, especially for a creative project like TV or movies, the results are just too bad (today). Although it's excellent at "following directions", which is why these new APIs are great and exciting, it lets people who are good at setting direction to leverage these tools to have the direction concretely followed by something, without having to create a group of collaborators.
Until we've seen proven that AI can indeed "set directions" ("have taste"), I think it's fair to assume they're unable to do so in a way that it produces something that humans generally enjoy. And very far from the kind of masterpieces a human can create.
It lets the unskilled access the tools of the skilled without having had to learn all the pitfalls, traps, and safety concerns. With video editing there may not be much at stake but we've already seen how sloppy code ~= vulnerable code when the prompter does not know what they are doing. What's going to happen when project managers start vibegeneering cars and buildings.
As pro software, the UI has a learning curve. Many YouTube videos are out of date, or unnecessarily drawn out.
Additionally, as a hobbyist, project settings, color spaces, clipping, etc can be tricky to miss.
I’ve already been using Claude to (1) help me with UI, (2) QA my delivery, and (3) occasional fixes for audio repair; waveform sync (can’t afford time codes).
I make all creative decisions - including most menial ones - myself, I wouldn’t have it any other way; I enjoy the act of creating. I welcome this integration.
I’m sure some users are gonna agentically grade or whatnot. I don’t like it, but I’m also don’t believe in telling other people what they should do with their own work and creative process.
Watching this next generation closely, because there are just a shitload of red flags starting to crop up that don't seem to be going away.
It seems pretty clear that AI capabilities are jagged. There are good reasons to believe based on research and the subjective evaluation of experts who are quite friendly to AI that there are some tasks creative tasks that AI has been really quite good at (totally disrupting stock illustrations) some where it does so much better than typical person, but not nearly as well as well as an expert (song creation) and some where, so far, it pretty much fails, and sees almost no progress even as models get better and better. (long form creative writting). Indeed, writting style sometimes seems to fall backwards with recent releases.
Also, it's subjectively, but commonly held to be true that most of the content created when AI can execute well lack taste, unless the human creating with AI has it. The term "slop" exists and in legible for a reason.
I mean, maybe this is just a question of different people speaking about different time horizons.
But it seems perfectly sensible for someone to say that there is a lot of automation that can happen in the near term, and that they see no reasonable evidence that AI can do interesting or even coherent work this space.
I don't know that shallow accelerationist boosterism (or panicked doomer despair ) that handwaves away the distinctive variations AI performnce is that enlightening.
And great for beginners, now they get to pay some chatbot company to cover the gaps left by detereorating software design. Lucky they're already using agents... are they?
This feels inverted to me: downtime affords us variety. If I have little downtime then I'll strongly prefer known, efficient methods of doing something, rather than experimenting with new ones.
> As the job gets stripped down to its core function, while a machine keeps the pipeline full so that the work can continue uninterrupted, the downtime slips away.
This I agree with, though.
I don't mean to be harsh, only passionate.
I have a hard time seeing your way. I build software. Tools for humans to do stuff. Whatever they want. It gets complex. There are methods to make it more accessible to more people. I want to go to there.
I feel you bro that the human is getting squeezed out. I do not see it that way in the sense that the human no longer is the nexus.
CEOs and tech bros and hype are all a distraction. You are a valuable person in the conversation and we all need you to participate with eyes wide open. I assume you are young. That's not discounting it is welcoming.
Thank. God.
I never want to rotoscope again.
And frankly I'm not fond of editing either.
What is lost when losing the (understandably annoying) process of the grind? By letting LLMs do the busy work, do you lose the opportunity to stumble into randomness? Do you lose the understanding of what is possible/interesting with the medium?
And do you end up in this derivative future where decisions are made by a system that was trained on all the previous uses of it?
Obviously that's fairly hyperbolic, but I have absolutely felt the pull of it all. When it decides to take a couple extra steps I didn't ask for creatively and I think "eh, that's fine."
*EDIT: I wrote this and looked back and this SOUNDS like an LLM wrote it. See what I mean??????
Ultimately, and this applies to so much of life, "to each, their own," right?
I'm no expert, but I never found Resolve materially more difficult to learn or use than Final Cut Pro or Premiere. And if you're doing anything moderately complex, it's potentially easier to learn because things that require external tools in FCPX and Premiere are built in to Resolve Studio.
I don't have a single reason to do any form of video editing, let alone professional video editing, but that kind of pricing model makes me want to buy a license just to make sure that companies like this stick around.
This is commoditizing Adobe for them.
Black magic also make a program that support/ties into the camera hardware on smartphones and they support both Android and iOS.
I've been using Resolve since version 14, still getting upgrades to it! My license does only work on one "computer" at a time though, so when I switch between Linux and Windows, I need to re-enter and activate the license, as when I do that it deactivates the previous (other OS) one. You have seemingly unlimited amount of times to switch though, as I've probably done it on a bi-weekly basis for quite some time.
I’ve been using DaVinci Resolve Studio for 12 years now with frequent updates, with the license I got with the original Pocket Cinema Camera for £800 (still a fantastic pocket size cinema camera, shoots raw cinemadng and ProRes with a beautiful sensor).
One of the best founders and companies around.
[Citation needed]
Although... BMD has actually made it harder, by adding bloat like the Cut page. I have attempted to find value in it a couple of times, and didn't. Now I disable it.
BMD needs to stop throwing crap into Resolve and buckle down and do two things:
1. Fix bugs, including UI defects. The Deliver page is a mess. The clip selection between the Edit timeline and the Fusion page makes no sense. And lots more.
2. Unify the node views. Resolve is a bunch of stuff BMD bought and threw together on different pages. The "integration" is janky and can, especially for the beginner, seriously degrade your work. Once example is how Fusion compositions inexplicably downgrade everything to timeline resolution. So you can bring 4K footage into a Fusion comp, but if your output timeline of the moment happens to be merely HD, the entire composition (including keys) will be degraded to work at that resolution. I mean... really? And beyond that, the relationship between Fusion comps and the timeline events they wind up in is extremely fragile and breaks arbitrarily and unrecoverably all the time.
A single nodeview for all operations would be a long-overdue game-changer for editing software.
Over time many things become easier to learn and do, and those who might feel upset should make sure their own feet are moving as well in terms of their learning and growing.
I assume that this can be connected to local models, not just cloud models.
But I desperately wish Blackmagic would bring support for VST3 plugins and JACK to Linux. Or, at the very least, allow me to use my MIDI control surface in Fairlight on Linux.
As it stands, I have to do all of my audio production in Reaper, and that's a bit cumbersome.
WRT audio I'm not sure as I only did this with audio less drone footage.
Is that praise? I would hope the bar is higher.
I bought a used computer four years ago to edit 4K video, tens of thousands of raw files off my big Sony cameras, write, light code and research. Many thousands of hours of hard use.
In five years I’ve never had a single program crash and certainly not the whole system go down. Uptime is many months.
I thought that was the expectation
Works without a hitch, including multi monitor, gpu acceleration etc I use it on Manjaro, installed via AUR.
That said, IIRC it only works with nvidia GPUs and requires X11, and is supposedly finicky to get the environment right for (but the AUR gets it right for me)
Only recently has AMD even become viable for LLM inference on consumer chips and that is the same compute codepaths Resolve would depend on.
So effectively AMD is the issue, AMD on linux is great, AMD on linux for a workstation was kind of hit or miss.
Making it run wasn't easy, and it can't decode MP3 and some video codecs because of licensing issues. It also can't decode WebM. This is kind of a problem because of a lot of free music you can download to add to videos and a lot of video clips you can download from Pexels for example are encoded with codecs it can't decode, so you have to re-encode it with FFMPEG first to a supported codec.
The biggest issues are the supports of video cards (nvidia is the king) and different popular codecs. The problem is partly solved by using paid version, but the gpu/codec support is still not even with Windows, so you have to use ffmpeg to re-encode from/into supported formats.
[1] https://github.com/nhma20/strix-halo-resolve-24-04/
The good thing about DaVinci Resolve's Linux system requirements is that these are so outdated that basically every developer links their plugins with static libraries they build themselves; and Resolve itself bundles lots of .so files with the app itself. Also Rocky Linux 8 is a good baseline Linux distribution: simple system libraries interdependence, so trivial to export and bundle those.
In other words, binaries created for RockyLinux8 with portability in mind are easily portable to sane modern Linux distributions.
(That doesn't make developing on Rocky Linux 8 fun at all)
I can definitely see the utility of something like this for cases where a media budget just wouldn't have existed in the first place.
I use DaVinci Resolve for editing GoPro, Canon DSLR, iPhone, and Drone FPV footage. I import clips, use the razor to cut them and rearrange them, occasionally crop and zoom, and then encode them down to h264 or whatever. There are a bazillion features that I sort of know exist but haven't ever had any need to use. Could I make a tool using an LLM that covers my DaVinci workflow? Almost certainly, and I don't think it would take anywhere near 200 prompts. I've made significantly more complex bespoke data analysis pipelines for drone flight telemetry in like... a dozen prompts. But I have no illusions that what I would make as my "DaVinci Clone" would have anywhere near the functionality of DaVinci, just the tiny piece of that functionality that I use.
Then vibe your own 5% editor, but saying that a software so mature and complete as Resolve can be done with some AI and time is something I expect coming from the mouth of a teenager, not an adult.
Resolve is like a top of the line professional camera: there are so many more knobs, buttons, and dials on those because top-end pros run everything manually changing settings as they take photos based on changing light conditions and what the subject is doing. Resolve isn't going to assume what you want and do it: it's a professional tool for professional users. The fact they are still giving it away is amazing to me.
Moreover, not everyone wants to have the hassle of building and maintaining their own version of a thing that isn’t compatible with anyone else’s version of the same thing. And if you hire anyone to help you, now you have to train them on how to use your app. If they leave, those skills are useless for their next job.
(I’m not even covering the idea that you could do it in 200 prompts. I got Claude to create a video player and that alone took over 300 prompts.)
Way more people should know about this solution.
https://www.danieltufvesson.com/makeresolvedeb
I tried it a few times but I think it didn't work at all with those cards.
Source - I work as a professional editor (amongst other things). Please commenters, do not come at me with your five minute low bit rate projects that required only a LUT and some contrast and 'worked fine' in DaVinci on a potato.
DaVinci Resolve is a proprietary application for non-linear video editing, color correction, color grading, visual effects, and audio post-production. It is developed by the Australian company Blackmagic Design for macOS, Windows, iPadOS and Linux.
...
DaVinci Resolve has been used for the color grading and/or editing of feature films such as Alien: Covenant, Avatar, Best of Enemies, Deadpool 2, Jason Bourne, Kingsman: The Golden Circle, La La Land, Love & Mercy, the Pirates of the Caribbean series, Prometheus, Anora, Robin Hood, Spectre, Star Wars: The Last Jedi, and X-Men: Apocalypse.
(Wikipedia)
How to ease key frames, how to fix my HDR captured footage for SDR, how to export the best quality video, how to burn in subtitles with word highlights, how to fix my audio aaand so on.
Either the AI will show me the particular feature and setting or a YouTube tutorial will. I think the AI might be faster and easier in this scenario.
Library management is where Lightroom (and honestly CaptureOne) still wins, I couldn't find a good way to manage a 100k photo library.
I created one project and tried importing the whole library, but it took more than 8 hours and didn't make meaningful progress (~10%?) so I stopped (compared to 15hr for the whole thing in Lightroom).
I also did a project for about 1K photos and while the import worked, the library management itself was lacking. I could not find ways to do image stacks for bursts, group by metadata, etc. The obvious filters are there, but I think it needs more work.