Unpopular opinions about AI and video editing:
1) AI won't replace editors. It'll replace editors who only knew the buttons.
2) Auto captions made every video look the same. The fix is still human.
3) An AI first draft is only useful if someone knows what's wrong with it.
4) The more footage AI can sort, the more taste matters, not less.
5) AI edits feel fine and forgettable. Fine is the most expensive word in video.
6) The editors who win next year won't be the fastest. They'll be the ones who can explain why a moment works.
7) Clients don't want AI. They want it done well and on time. Most don't care how.
Great take! Often times when chatting with prospective clients I convince them to go with one edit at my rate, no commitment, then if they like it we could talk about a longer term relationship. Some have gone with it, the decided to pivot to a cheaper editor. They then compare the experience and the results, and almost always come back
Going on about a month of using ChatGPT for a lot of automation in my video edits.
They’re still feeling about 60% of the way to what I’d like them to be. The last 40% is me battling the AI to dial it in, as opposed to me manually editing them.
Also noticing that for certain tasks the time it takes to execute is much longer than what I could do myself. Part of dialing in the workflow is figuring out how to efficiently prompt so each turn doesn’t perform tasks not needed in that turn…
@thelegendbul Truly. The problem with everyone’s edits looking the same are that audiences will become desensitized to it and its effectiveness will plummet. Results will come from people using their imagination to step outside that box
True motion designers design work that evokes a feeling, has a strong overarching theme, and tastefully breaks the rules
Agentic “motion designers” are just creating the same things over and over, with rounded edges and smoothed speed ramped motion.
You’re not an editor or a motion designer just because Claude is giving you decent outputs
@linus5x I gotta be real. I don’t believe what you’re claiming. The numbers don’t add up. First red flag, no link to the video in question. In the replies you mention the name of the channel, which analysis shows the inflated view counts are not tied to organic growth or engagement…
I’ve found that if you give it a ton of parameters to work within, it gets close. The results are still far from perfect, but depending on the type of edit you’re working on it can get somewhat decent results.
For example I tell it what to exclude like wobble, out of focus shots, bad exposure, awkward facial expressions or gestures, etc. I also feed it a reference video of something I’ve manually cut before.
The other thing is if you’re working on an explainer and trying to source b-roll from the internet it’s all about pointing it to where to pull from
@Noukin_Camera Been so curious about running local models. I’m planning on upgrading to a Mac Studio with a large amount of ram and processing power. Gonna be expensive but in the long run it should be cheaper than burning through tokens
@stalman Straight up. If you’re willing to betray your spouse *multiple times* there’s no telling how low you’re willing to go. Can’t be around those people in business. High integrity business partners are the only way to go
You’re going to be blown away by this new workflow. I’ve been a professional editor for about 15 years and the things this can do are crazy.
The more detailed you are when prompting, the better. I typically will turn on voice dictation and describe in detail the type of cut I’m looking for. Describing everything from how to organize my clips, to story structure, and export settings. Give it a shot!
@ArslanA50689007@Faris_edits Best way in my opinion to gain new clients are through strong referrals. Any clients you’ve been working with and have built a relationship with would be happy to refer you to their friends
@thefinnmckenty Worst thing we’re up against now are usage limits. I’m running the $500/mo ChatGPT plan and the $100/mo Claude and these workflows are burning through my usage
The reality of this is that the $20/mo subscription cannot support an end to end agentic editing workflow. I’m currently running the $500/mo workflow and just about hit my usage cap each week.
Also to add, there is still a good amount of work that needs to be done on the backend to get the edits over the line. YouTubers will still hire editors, they just probably won’t need as many. Goodbye to the era of 5+ editor teams