Seven Shorts formats
I Tested Every Viral Shorts Format With AI
I made one short in each format. Below is the prompt, what it cost, and the file that came back.
This is the written version of my YouTube video. Watch the original video
What I tested
The Short I open the video on has about 80 million views, and a similar one cost me a dollar and under five minutes. Cheap is not the same as a channel I would actually repeat. I scored each format on cost, how much work it takes, and whether I would build a faceless channel around it.
Three of them start in Shorts Studio, where the camera style is already built in. The other four start in Wizard, where I paste a reference and ask for a finished short. YouTube’s inauthentic-content rules are what I scored against: a single generated clip with no story is the risky shape. Narration, captions, and stitching clips together are what I treat as the fix.
01CCTV and doorbell
One clip, one dollar, nowhere to go next
A security-camera short where something ordinary goes wrong.
These spread because it looks like a camera caught something it shouldn’t. In Shorts Studio I pick the CCTV format and describe the scene. While I type, the composer is set to Seedance 2.5, 8 seconds, 720p.
a cat ringing a doorbell as a hundred other cats are patiently waiting, as the door opens, all cats swiftly run away as they were pranking the home owner
Shorts Studio, CCTV format.
Typed on screen. I read the same line aloud, with “because they were” where the box says “as they were.”
The clip I play a moment later is an earlier take. Its saved direction asks for tens of cats, and that file is the one below. A separate generation that says “a hundred of other cats” is a crowded porch, and it is not the clip in the video.
a cat ringing a doorbell as tens of other cats are patiently waiting, as the door opens, all cats swiftly run away because they were pranking the home owner
The direction saved on the Shorts Studio generation that plays in the video. 16 credits. The on-screen card reads 16 credits = $1.
That price is the advantage. One clip, a minute or two, no special skill. It is also the weakness. There is no ongoing story and no recurring concept, so the next video has nowhere to start. The channel I found that repeats this format uses one idea over and over, a dog saving a child. That is the only way I would try a CCTV channel.
The cat clip has none of the things I use to stay clear of inauthentic content: no narration, no stitch, no edit. I might still monetize a simple one-clip CCTV video. I would not build the channel on it. I give CCTV and doorbell videos a 2 out of 5. Ranking videos, later on this page, are what I would wrap around this format.
02POV wholesome stories
A full story, and a $7.15 bill
A narrated first-person short built to make someone tear up.
The shape is a rescue: a firefighter pulling a child out of a fire, or someone saving an animal. The reference I pasted is this Wild Alarm Short, a penguin running to a car to get away from a seal. That channel is past 700,000 subscribers from this format.
make a wholesome narrated multi-clip story similar to https://www.youtube.com/shorts/0d956JF95EA. Make sure our video follows the same principles that made the example video go viral. I want our video to be so emotional that people tear up at the end.
Typed in Wizard. The YouTube link shows up as a chip on screen. I also read it aloud.
More moving pieces means a round of notes. This is the one I sent about the opening.
pls rethink the beginning. We can see the deer running away which doesn't make sense. Feel free to generate more clips to also hook the viewer better
Saved in the same Wizard chat. I didn’t read this one aloud. On camera I called it one round of revisions.
After that cut I sent these two notes. The captions are in the file.
great, love this, add captions and music
Saved in the same Wizard chat. I didn’t read it aloud.
hmmm so it's impossible for you to add music?
Saved in the same chat, after I said the captions were great. I didn’t read it aloud.
What came back is about 40 seconds. I hear the ice crack, a mother deer and her baby are in the hole, I crawl out and pull the baby out first, then her. I spent $7.15. Every clip is generated, which is why the next format costs less.
YouTube has said channels that lean on animals in exaggerated distress can be demonetized, so I would not push that. The story, the narration, and the captions are what keep this on the safer side of inauthentic content. I give POV wholesome stories a 3 out of 5.
03Interesting facts
One prompt, $4.50, and a channel I would actually run
A narrated short that explains one fact with a run of clips.
Lots of channels already do this, in different niches. Farzan Films tells history, including a short on the Opium Wars. Discovery Dose does nature. Their short “Why Bears Leave Most of Their Salmon Behind” was about a week old in the video and already near 7 million views, and their biggest sit around 100 million. The reference I pasted is this Discovery Dose Short, the one I called the crabs short.
analyze why this video went viral https://www.youtube.com/shorts/obZcEJj5jf8 and make a similar multi-clip video about an interesting fact for me. Make sure our video follows the viral principles you've identified from the reference short
Typed in Wizard and read aloud. The Short’s link shows up as a chip.
It came back in one pass, so I didn’t send a revision. The short is about Sichuan, where pesticides wiped out the bees and farmers pollinate pear blossoms by hand with brushes made from chicken feathers and cigarette filters. $4.50, and under 10 minutes.
The stronger version of this workflow is to let Everygen find the idea, write the script, and make the voiceover, then find real footage yourself and generate only the gaps. That spends fewer credits and shows the real thing. Using someone else’s clip is fine when the narration and the edit are the new part. I would build a channel on this, especially if you already edit. I give interesting facts a 5 out of 5.
043D explainers
Two styles, and the most expensive test
A 3D short that explains one idea, in the Zack D. Films shape or the Dr. Data shape.
Zack D. Films is the channel that set this style. Dr. Data looks different and does the same job: 3D that explains something, like how stitches work or how people stayed clean before soap. I ran both. Everygen already has the Zack D. Films style, so I didn’t have to brief an animator.
analyze viral shorts on this channel https://www.youtube.com/@zackdfilms/shorts and identify exactly WHY they went viral. Then, create an original short in the same style that will also go viral
Typed in Wizard and read aloud. The channel link shows up as a chip.
The plan on screen uses Kling 3.0 for the shots of the person and Seedance 2.5 for the joint cutaway. Before the export above I sent two notes. The first keeps a typo, “when then see,” as I typed it.
make it closer to zack d films. Make sure the videos perfectly visualize the voiceover. Make sure the cracking the knuckles action is hyper realistic and satisfying to watch. The same goes for the close-up inside-the-bone 3d visualization shot – I want viewers to go WOW when then see what actually happens inside their bones when they crack their knuckles
Saved in the Wizard chat, as typed. I didn’t read it aloud.
nah let's bring back that guy at the beginning as the first clip, then cut to the current first clip
Saved in the same chat, after I asked if it was good to go. I didn’t read it aloud.
The short explains why a knuckle pops, and why the same one won’t pop again for about 20 minutes. The card on screen says 148 credits, or $9.25. I would still go back and forth before I posted it. The pictures have to explain the sentence, not only look good.
I didn’t leave the Dr. Data prompt on screen. On camera I called it the same prompt with that channel as the reference. The saved message is its own text, and it is the one that produced this short.
analyze the most viral shorts on this channel https://www.youtube.com/@dr_data_dr/shorts, identify exactly WHY they went viral. Then create a short in the same style that will also go viral thanks to the viral principles you identified from the reference channel
Saved in the Wizard chat that produced this short. Not shown in full on screen. I described it as the same prompt with Dr. Data’s channel as the reference.
This one is how people told time before clocks: the sky, then Egyptian obelisks, then the stars. The story came back in one pass. I asked for captions, and they’re burned in. The card on screen says 120 credits, or $7.50.
These explainers hold a shot longer than the facts shorts. One clip stays up for more than 5 seconds while the sun rises, sets, and the character runs back to the cave. This is the most expensive format I tested, and the one that can show something you cannot film. I don’t see a monetization problem. I give 3D explainers a 4 out of 5.
05Nursery rhymes
Easy topics, slower ad money
A short animated lesson for kids, with music.
Animation plus a song, one simple lesson. On screen I show “One Two Three | Learn Number | With Slide” at 5.5 billion views, and “ABC Vehicles: Learn & Play” at 399 million. I wanted basic shapes. Shorts Studio has the nursery-rhymes format, so I pick that and type the line. The composer shows Seedance 2, 8 seconds, 720p. The short that came back is 15 seconds.
make a short that's going to teach kids basic shapes
Shorts Studio, nursery rhymes format.
Said in the video, and the direction saved on this generation. The composer shows the start of the same line while I type it.
It comes back in one pass. You don’t need a plot. Shapes, colors, counting, animals: the topics are already there.
The catch is the ads. YouTube says skipping personalized ads on kids’ content can lower revenue. I put the other formats on this list at about 20 to 30 cents per thousand views, so a thousand dollars is about 4 to 5 million views. Kids’ content I put at about 5 cents per thousand, which is closer to 20 million views for that same thousand dollars. That is ad revenue only. Once the characters are recognizable, merch is the other path. I give nursery rhymes a 3 out of 5. Fast to make, slower money at the start.
06Satisfying shorts
A hydraulic press, now that the physics hold
One shot of something getting crushed, in the candy-press style.
Until recent models, AI was bad at this kind of physics. The reference Shorts on screen are candy compilations. The biggest of them shows 566 million views. They all use the same press and the same kind of candy, so I screenshot the setup and attach it. Shorts Studio has the hydraulic-press format. The composer shows Seedance 2, 8 seconds, 720p.
Image1, the screenshot I attachedmake a video of hydraulic press squashing candy. Use @Image1 as the starting frame
Shorts Studio, hydraulic press format. @Image1 is the screenshot above the prompt.
Typed on screen, including @Image1. I also say it aloud. The on-screen card reads 16 credits / $1.
Under two minutes, and the press behaves. What I like is that these are bingeable. I watched a long run of the reference channel the day I found it. The catch is that anyone else can make the same video, and it has the same monetization problem as the CCTV clip: one shot, no story, not much of an edit.
I don’t give satisfying shorts their own score in the video. The last format is what I would add on top of them.
07Ranking videos
Generate the hook, keep the rest real
A numbered short that opens on a generated joke and then cuts to real clips.
Rankzilla is the channel I use. I point at a short of theirs with 350 million views that opens on a generated hook and then cuts to real footage. That is the workflow: generate the hook, use real clips for the rest. You spend less, and you don’t have to invent every joke.
analyze the most viral shorts on this account https://www.youtube.com/@RankZilla23/shorts and give me 5 ideas for a ranking short like they do that will also go viral
Typed in Wizard. The channel link shows up as a chip. I also say the instruction aloud.
One idea that came back was why dads can’t be left alone. I searched “dad funny moments” on TikTok, took the first clips that fit, and sent this. The two links are pasted with no space between them, the way they were saved.
https://www.tiktok.com/@the_hale_homestead/video/7529065799927991582?q=dad%20funny%20moment&t=1790914562462https://www.tiktok.com/@laughingloop0/video/7687926045286173985?q=dad%20funny%20moment&t=1790914562462 let's go for #3, can you take clips from these tiktoks, then generate a crazy funny clip to use as the hook yourself, and stitch this viral short together?
Saved in the Wizard chat and partly visible on screen. The TikTok links render as chips. The visible line matches this instruction, including “let’s go for #3.”
About 70 cents, because I generated one clip. A second generated clip would be about twice that. I would budget 30 to 60 minutes, mostly for finding the real clips by hand. The export above is the stitch. The numbered ranking overlay is not on it. I add that in CapCut, the way Rankzilla does, and Rankzilla also adds transitions and sound effects. Those make the video funnier, and they are part of making it different from the originals.
You can rank CCTV clips or satisfying clips the same way. That gives those formats a repeatable concept and a real edit, which is the monetization job: the video has to be meaningfully different from the clip it came from. Some channels add a reaction voiceover between the clips. I don’t think it’s required. I give ranking videos a 5 out of 5.
Where I would start
If I started a faceless channel today I would pick ranking videos or interesting facts. Both have more topics than I could post, and mixing real footage with a generated piece keeps the cost down and the video better.
Pick the one you would enjoy making every week, pick a niche, and make the first few. The prompts above are the ones from the video.
Make the short you would actually repeat
Ranking videos and interesting facts are the two I would start with. Paste either prompt into Wizard and swap in your own reference.