AI editing, a track that stifles money.
In traditional production, the production process of a video roughly includes pre-planning, mid-shooting, post-production and editing and distribution, etc. AI disrupts this production process, and the head players in the industry tend to focus their attention on the promising Sora's, but ignore the more practical and able to solve the user's pain points of the application.
Short video slices.
Just as a number of AI video generation products rolled to life and death, trying to prove to the outside world their potential for commercialization on the ground.stapleSlicing and editing's AI products are already sneaking in profits. One of the best of these is the Opus Clip, whose core function is toUsing AI technology to automatically identify the highlights in long videos and generate viral videos suitable for distribution on short video platforms.
The product, created by Chinese, gained 5 million users in just 7 months and did almost $10 million in ARR (Annual Recurring Revenue). In addition to Opus Clip, according to Product Hunt, there are tons of similar products on the market that benchmark Opus Clip.
And there's no bigger reason behind why this track has been so hot than the fact that it's really liberated some of the creators' productivity.
All kinds of short video slices that are popular on the extranet are cut by AI
Editing has always been a time-consuming and boring part of the video production process. Especially after the rise of short video platforms represented by Jitterbit, theNew slicing gameplay has been derived, based on live broadcasting, film and television variety shows and other long video content for secondary editing and multi-platform distribution to gain more traffic and exposure.
The country's leading cargo anchor, "Crazy Little Brother Yang" will also beLive SlicingMade into a business with an annual revenue of hundreds of millions of dollars. According to the official disclosure data of Little Brother Yang's company, Three Sheep, a total of more than 11,000 people were authorized to receive slices from Three Sheep Network in 2022, with a per capita income of 17,000 yuan. Even according to the minimum commission share ratio of 5:5, Xiao Yang's earnings through live slicing reached hundreds of millions of dollars.
In this model, head anchors earn commissions for licensing IP, and downstream accounts earn by editing and distributing live sliced content.
But for downstream slicing accounts, revenue is heavily dependent on the head blogger. And, after the market is saturated, the revenue that is strongly tied to traffic will diminish.
It also means thatEven if you put a lot of time and effort into editing your videos, there's a good chance that you'll end up getting less traffic and revenue than expected.Batch cutting can increase the probability of hitting the flow, but if you do it too much, you'll feel like you've become a tool on the assembly line.
The advent of AI, on the other hand, makes the ROI ratio of the slicing business look high.Many products are also very attractive slogan: just one key, AI can automatically edit the long video into ten or even dozens of short videos that can form a viral spread on social platforms.
Real-world testing of 3 popular AI slicing tools.
Let's see which one.mostWith?
Whether it's long-form video bloggers trying to distribute their content to short-form video platforms, or traffic plays such as live-slicing bandwagons and short-form drama slicing and drivin...The core requirements are all about distilling the highlights of long videos for redistribution.
There are many AI editing tools on the market today.We've picked 3 popular products that specialize in "long cuts and short cuts" and have a certain amount of free usage., tried it out.
Opus Clip
Opus Clip has recently launched an internal beta of its next generation model, "ClipAnything", which is officially calledThe firstMultimodal AI editing tool, can extract and edit any clip from any video by visual, audio and emotional dimensions.
It's easy to use. Directly upload your own computer's local video, or paste the video URL, click the "Generate" button, it will help you cut the video into multiple short video slices.
You can also enter prompt words thatCommanding AI clips with your mouth. For example, have it analyze which part of this video is the funniest.
- Cue word: Compile all the hilarious moments
Because the selected clip is the classic American sitcom "Office", we chose "Entertainment" and "Drama shows" for the genre, and finally the AI outputs 9 video slices and scores them according to several dimensions such as The final AI outputs 9 video slices, which are sorted according to the scores of Hook, Flow, Value, Trend and other dimensions.
In this dialog clip, two salespeople are conducting a customer pitch simulation under the supervision of a leader. One of the salesmen, Jim, who plays the role of the customer, intentionally says his name is "Bill Buttlicker" in order to play a prank on the other salesman, Dwight. In American culture, "Buttlicker" means "ass-kisser".
In the real test, the 9 slices given by the AI were basically localized to this laugh track. We picked one of the shortest and most concise of these to see the effect: the
In addition to having the AI find the laughs in the video, you can also have it analyze the clips in the video that are most likely to go viral on social media or just have it locate famous scenes from classic movies and TV shows.
For example, the famous scene in "Legend of Concubine Zhen Huan", "Yizhang Hong".
- Cue word:Find the part where someone is kneeling and begging for mercy.(Find the part of the video where someone is kneeling and begging for mercy)
More than 40 minutes of the original video, this time the AI gives a slice, but also basically positioned to the "red" famous scene, the slice is also used in the top and bottom of the split screen, multiple people in the same frame, and other editing techniques to pack the video.
Descript
This tool is also an AI editing software that is highly recommended by many bloggers on Tubes.Descript originally started as an audio podcast, so the interactive interfaceIt's easier for creators to view the transcript and edit the video by marking up the transcript.
If you want to make a jump cut, just delete the text in the middle.
Descript also supports AI to analyze and extract clips, click the robot button on the top right, find "Create clips" and click it to set the number of clips you want to generate, the length of each clip and the appearance of the clips.
It's worth noting that Descript also allows you to enter cue words for the AI to find the clip you want.
Here we hope to find clips with the keyword "Elon Musk" in them, and see how Trump blew Musk off in his post-election speech.
However, only one of the five slices given by the AI mentions "Elon". Look at the results.
Vidyo.ai
This tool is also mainly positioned to convert long podcasts and videos into suitable slices for distribution on short video platforms. The specific use is similar to Opus Clip, just import the video, AI will automatically analyze and edit the highlights.
Here we took a link to an earlier video of Musk and Altman talking and threw it at the AI to cut out clips with the keyword "AI" in them.
https://www.1ai.net/20450.html
In the end, I was given 8 slices, most of which mentioned the keyword "AI".
At first glance the results look pretty good, but one of the highest AI rated cutscenes sees people out of the shot at the end, and if you want to adjust the characters to a reasonable position, you need to pay for a membership to use it.
The basic package of 33 euros per month (about 250 yuan), really can not afford to use. Converted into yuan, the price is also quite connotative people, who spend who is a fool's sense of vision.
In contrast, the conversation slices generated with the Opus Clip are quite "TikTok"-like in their pacing and camera combinations.
AI cutouts, how far have they evolved now?
As far as the real-world experience is concerned, all 3 tools are very similar in terms of functionality.From long videos to short videos, AI can help you locate the highlights, edit, and provide subtitle templates, AI auto-generated emoji, and other features that can package the video twice.
Also, they both support secondary fine editing.For the AI-generated slices that are not precise enough, you can manually make secondary edits, you can fine-tune unwanted sentences, add voiceovers and brand logos, and so on.
In the past, the entire cutting and editing process relied entirely on manual processing, which was not only time-consuming but also boring. Especially if the amount of material to be sliced is very large, it is easy to turn this into an assembly line job.
the reason whyMany bloggers overseas are using AI to edit their long video content.The whole experience can be quite silky smooth, especially for tubing bloggers.Longer videos posted on Tube are also available for secondary distribution on short-form channels such as Tube Shorts and TikTok.
We also tried to use Opus Clip to generate a live clip of "Mr. K".The cue word requirement is for the AI to cut out full product bandwagon clips that attract people to click on them from a 1 hour live recording.
As you can see, the final cut out slice is still very stellar and pinpoints the bandwagon clip of an item.
However, Opus Clip's built-in default fonts don't support Chinese, so they will be displayed as garbled in the video. You can upload fonts that support Chinese in the font settings, but this feature requires a paid membership, which costs $15/month.
Opus Clip free version currently supports 60 minutes of video analysis and editing, if there is a greater demand for batch generation of short video slices, you can also consider opening a membership.
A more frugal use would be to turn off the subtitle display in Opus Clip, and just batch export the AI clipped slices and subtitle them again.
On the domestic front, Ali Tongyi Labs had open sourced a program called"FunClip"The automated video editing tool that can perform speech recognition on videos and convert speech to text.
Users can select a text clip from the recognition results or specify a speaker to have the AI cut out a corresponding video clip.
While it's not like tools like Opus Clip, which allows AI to straighten out slices with one click, and FunClip, which requires manual screening of content, it still makes editing a lot easier than timeline-based editing tools like Adobe Premiere.
FunClip Java Open Source Address.
https://www.1ai.net/10553.html
There are also many domestic tools that support AI one-click mixing and editingAs in the case of the little-known B-siteAutocut. Many Knowledge Zone and Tech Zone UP owners use Autocut to optimize spoken word and batch mix video clips.
Tool address:
https://www.1ai.net/23392.html
Tools such as "Instant Creation", an AI creation platform under the Jitterbug Mega Engine, and New List's Little Bean SproutIt is one step closer, not only to realize one-click intelligent mixing and editing, but also to support the AI with the script, merchandise selling points and other needs, to generate the finished film material.
The advantage of this approach is that mixed cutscenes can better achieve the effect of "de-emphasis", reducing the risk of platforms to limit the flow.
Tool address:
Tool address:
https://www.1ai.net/23395.html
Of course, even if today's AI can already cut your mouth broadcasts, mix and cut material, and even let AI generate dozens of slices with just one cue word, the shortcomings of AI editing are also obvious; its understanding of long videos and narrative content is still limited, and it is not yet at the stage of large-scale commercialization, let alone replacing the human editors.
But as OpusClip founder Zhao Yang said, "AI can only do 80%, the rest of the 20% still need people to take over." The relationship between people and AI is not simply to replace and be replaced, but to collaborate and complement each other.