Posted by pentagrama 3 hours ago
For anyone not in the know, Resolve has an exceptionally capable and feature rich free version. A lot of the AI features (and >4k editing) are locked to the Studio licence which is a one-time payment, but works simultaneously on two computers (including different OS's) and allows upgrades across major versions.
I spent less than $300 on it a decade ago and my licence works fine on new v21 released this week. My least-regretted software purchase in 3 decades.
I already used Davinci to make a custom LUT, blended it with the Adobe camera standard profile to use in lightroom as a base for all my edits as Davinci's color tools are much better, and doing it this way just lets you get tones that you couldn't other wise get with Lighroom/ACR alone. This basically removes the need to have lightroom in there as a finishing step.
The only downside is by now I've got a really solid ImagenAI profile based on the 30k photos I've fed it over time, and that obviously relies on a lightroom catalog, and is also capable of applying my LUT since I've made it a slider in LR. I hate adobe, but I'm not sure I could go back to not using Imagen as now I can turn around a full wedding gallery in about 2 days.
I assumed they would add them later, I hope I'm right!
edit: nope, still only supports Canon, Fuji, Nikon, Sony and iPhone ProRAW.
I am pretty experienced linux user and i would have to buy new NVIDIA card and pray things work out.
Sadly i found out it will be much cheaper to buy refurb mac mini. So now i have dedicated machine for editing video.
They are both PatchMatch (well the healing brush certainly is), which is a heroic bit of code. Entirely deterministic statistical algorithm. Not AI by really any definition (including back then)
I don’t know enough about anything to determine which opinion is correct.
Is AI output not just matrix multiplication and a random seed?
But what I would observe is that the healing brush does not have a random seed. It will always do the same thing if applied at the same pixel.
(I am actually less sure if content-aware fill randomises; I always got the impression it did not)
This makes it both incredibly powerful and occasionally frustrating.
Because on the one hand, you can learn to apply your judgement to precisely control what it will do, and change the radius or position if you learn it is likely to fail, which becomes instinctive. I absolutely love using it to fix scratches in film scans; it's a quick, precise, controllable tool that can be used in a way that is amazingly convincing, and it ends up quite a "zen gardening" thing as a result. It'll sell you on the cheapest wacom pen once you know how efficient it can be.
On the other hand there are situations where it simply cannot work the way you want because it will always find a pattern you don't want it to.
(You can sometimes use the clone brush tool first, to manually break up the pattern that patchmatch will find)
Given a model architecture that supports it, greedy decoding + the same inputs + prompts, that's true for most LLMs today too, I don't think people consider them less/more AI because of that.
I have a bunch of photos from a Mini 3 Pro and Mini 4 Pro -- 4 years' worth -- that I haven't published because I don't have a way to process them.
No tools on Linux (RawTherapee, Darktable, RapidRAW) render their colors correctly no matter how you mess with the sliders, and all of the Github issue pages are dismissive of the problem. There is something fundamentally wrong about how Linux RAW libraries are reading DJI photos.
Lightroom on a Mac tested on another computer renders them all correctly, but I don't own a Mac nor Lightroom.
I moved my team to Resolve about 2.5 years ago and I can say with absolute confidence Fusion > AE. Resolve Studio is all you need, period. Lightroom is clearly better right now than their Photo editor but like all their other offerings (ehhh except fairlight lol still mediocre), it's only a matter of time
When it works, it's really nice and I'm sold on the node based workflows, but it's not all entirely roses.
Resolve itself is excellent, Fairlight works just Ok and is still heaps better than Audition ever was. But Fusion I feel like hasn't seen enough love in the past few years.
The whole Reactor setup for installing Fuses feels really broken and for the free users they disabled those entirely in version 19.1.
Most video is going to be AI in the near future. They see the writing on the wall. Their camera business line is going to sharply decline.
That's like saying all fine art would be photography, all film would be CGI, or all music would be synthesized electronica.
That's not how aesthetics seem to work. Artists will make more or less good use of generative AI in their work, and it will probably seep into most media in some way or another, effecting them, but arts mostly don't get replaced and AI doesn't really offer an exception to that history.
Or is it like saying most portraits will be photographs rather than paintings? There are still a lot of portraits painted (maybe even as many as the pre-camera days), but by raw numbers most portraits are created by photographers.
I don’t think they’ll see a decline in cinema camera sales due to AI soon.
Even the frontier models running on insanely powerful hardware could only generate 15 second clips in low resolutions.
And yeah, I saw some demos from Seedance 2.0, and they were awful. It's ridiculous how much people on Xitter were like "You can't even tell it's AI!" and I was like "It's trivial to tell it's AI" and could easily pick out all the markers. An individual screenshot could look good, but every time the camera angle changed, there would be a glaring inconsistency.
You people are either blind, delusional, or outright insane. AI might be used for a quick clip, or used to enhance something recorded by a camera, but "most video" is definitely wrong.
Having said that, for all the AI features, the big one would be setting key frames etc. with an agent, driving the general editing workflow with text,etc. I realize this is non trivial but it's certainly viable for a team of this calibre.
I think if BM added a paid for agent which helped execute their traditional video editing tools (even if it "only" supported a subset) then that's a subscription a lot of people would be willing to pay for, especially as their core tool is so generous.
the more SFX-end ones like facial aging and face reshaper, or the talent-replacing ones like speech cloning/ADR, feel both too prescriptive for a director to dial in what they aesthetically want, and also not good enough for the final cut. so I struggle to find where they'd be actually useful in a workflow as opposed to being a trap, looking just fine enough at a glance to sneak into a final cut while looking poor when viewed by the audience.
likewise the focal adjustment and upscaling just feel gross. the kinds of things a good cinematographer can already do, and it'll be so so tempting to use tools instead of taking time to do it right in camera because it'll look good enough in the editing bay, but I feel like it'd really stand out as fake in the final cut unless you're targeting like, heavily compressed social ads. the less you use them the better they'll work, which isn't ideal for a marquee feature.
if the blemish remover really does respect continuity it'd be a nice-to-have, but it also feels like another trap to be careless/cheap on things like makeup or lighting at the shoot, at the expense of looking fake in post
I think that's the broader angle that bugs me the most. all of these tools are convenience tools for editors, but in the end they'll really be justifications for directors/producers/studios/agencies to cheap out and do shittier work faster on the shoot. a cheap, shitty shoot covered in AI bandaids is still going to hit an audience like a cheap, shitty shoot.
Artists appreciate and use time saving tools. The AI features in Lightroom, for example, are very well received by photographers (myself included). Automatic subject, background, sky, body part masking, content-aware/generative fill and generative remove are genuinely helpful and time saving.
Where as marketing at all these corporation is trying to genericize "AI features" into anything using an algorithm. "Content aware fill", something we've had for over a decade is now "AI object removal"
"Noise suppression" is "AI voice extraction"
Motion unblur is now "AI motion unblur".
My bigger concern is that these neural network based solutions have taken the place of the former rather than supplemented them. Many tools no longer provide the algorithmic/kernel-based approach at all, and have marketed the “AI” (née ML) alternative as a strict superset/upgrade, despite its potential drawbacks.
(Interestingly while the inference-based implementations generally have higher latency (or infinitely worse, cloud and pay-as-you-go requirements), for some computationally difficult kernels the inference-based approach is actually faster!
To be clear, my use case is making weekly online videos suck a little less - not grading feature films :)
https://github.com/mhadifilms/dvr/
(or just point Claude to the API surface)
It might be chaotic, but you should be able to automate editing... maybe. :-) Ha. (You'd need a good loop to make sure Claude can see what it's doing..? Beyond screenshots I mean)
It's a baffling process flow if you're coming from Lightroom or a manual ACR workflow. But I'm excited to see where this goes. Quite simply, the output results are great. And free!
Not that I spent any extra money on the license compared to what the camera itself costed, but I also feel like the 0 money I spent was well spent :) Harder about the time commitment to move from something you know really well to something new, but the time I spent on that was very well worth it too.
Different stories shown with different treatment. With CGI, scenes zoomed out to wider shots and effects swelled even louder over lighting, intimacy, acting, etc.
Old styles didn't disappear or stop evolving entirely, of course, but the center of attention profoundly shifted and the "big" production money went with jt.
Generative AI will likely drive some kind of analogous shift in dominant film aesthetics. I don't know where, but I'm not particularly excited by it myself yet.
Likewise, there's been a ton of movies since that could in theory have been done purely with SFX instead of VFX, but which is probably must better from having used VFX/CGI, titles like The Matrix, Lord of the Rings, Inception, Avatar, etc.
Obviously there was also charm in titles like E.T. and Gremlins, and I think there might still be a market for movies like that, such as the 2019 Netflix series The Dark Crystal: Age of Resistance (I loved the 1982 version as a child, though this sequel wasn't for me).
I guess the question is: will generative AI allow new movies to be made, the same way that CGI has? Or will it just be an economic shift: the same quality CGI but at a fraction of the price?
There may still be people out there who believe that AI will never be able to create CGI as good as humans can, these might be the same people who used to say that CGI can never look realistic. And if you work in VFX, I bet that you can spot a fake mountain with the slightly incorrect shadows in the distance easily, but as a simple movie watcher, I really don't see it, especially when it's only on screen for like 3 seconds.
If you look at the ten top-grossing films in the past 15 years, it's almost exclusively computer animations or SFX showcases with some token, quip-based acting thrown in every now and then (all the Marvel universe stuff).
Basically, I'm not even sure if replacing actors with AI would be stylistically perceptible for the average moviegoer. There are interesting films still being made, but almost no one is watching them, because you have the option to watch some SFX explosions and flying superheroes in Avengers VII.
I don't think their use of it is bad at all, I'm just tired.
Putting "AI" into the feature title means "this time it actually works"
My opinion is that, for end users, if you name your feature "AI" to market it, you kind of already failed to read the room. You're writing to VCs while hoping it convinces customers.
Name what the feature does, what it gains them. Call it "smart" if you must imply some black box treatment.
Naming AI as the selling point for everything feels a lot like that Android tablet ad circa 2010:
"Your wife will love the new dual core Tegra™ chipset!"
100% this.
Maybe I live in a bubble, but consumer sentiment regarding AI seems extremely negative. Boasting "AI" features is more likely to lose sales than to create them.
I'd also get tired if it was "AI ala Microsoft/Google" where the goal is to get you to write forever with a chat bot somewhere else, but these features are very different from that.