The middle finger dunk is a TikTok trend from October 2026. Under the meme, it's actually a tight visual spec: 8 seconds, 9:16 vertical, two people, and six timed beats. TikTok's page for the trend showed 503.7K posts on 27 October 2026, which is a general topic page that mixes the AI trend with real basketball clips. Here's the shot breakdown, how the trend spread, and what an open model pipeline needs to reproduce it.
The spec: six beats in 8 seconds
Here's the entire sequence written out as an engineering spec:
Format:
- Aspect ratio: 9:16 vertical (phone screen)
- Duration: 8.0 seconds
- Audio: hip-hop beat + indoor crowd noise (made together with the video)
Cast:
- Dunker: mapped from photo 1; wears a white #7 jersey with blue and gold trim
- Defender: mapped from photo 2 (or a made-up player); wears a red #11 jersey
- Appearance: original faces, skin tones, and hair stay intact
- Badges: no team names, no league marks, no brand logos
Setting:
- Packed indoor pro basketball arena, court to ceiling
- Camera: one continuous handheld-style take
Shot list:
- 0.0–1.0 s: Dunker attacks defender, crosses him over, and slips past hip
- 1.0–2.0 s: Camera pans behind defender; dunker takes off into a wide leg split
- 2.0–3.5 s: Dunker hangs on rim with one hand, flashing middle finger at defender
- 3.5–4.0 s: Ball flushes through rim and net
- 4.5–5.5 s: Dunker drops to floor; defender flips ball back
- 6.0–8.0 s: Dunker dribbles toward lens into close-up, middle finger up with a smirk
Because the timing and action stay fixed across runs, the pipeline's only real inputs are two face images. That's why you don't need any complex scene setups once your inputs are ready.
How the trend got here
The meme didn't start inside an automated generator. It evolved over several days across multiple creators:
- 19 October 2026: TikTok user @dihanator4000 posted a real 1v1 streetball clip from an outdoor night court with the caption "Bro did an emote mid dunk", passing 1 million views by 24 October.
- 20 October 2026: French creator @teuteuf6 posted an AI remake of the same play featuring an animated character, captioned "Mon prime 1v1 basket" (my prime 1v1 basketball), clearing 1.1 million views in four days.
- 22 October 2026: Creator @scrambledai shared a 14-second AI version adding a slow-motion leap over the defender and a walk-off.
- 23 October 2026: A CapCut template titled "Your Prime 1v1 Basketball Moment" appeared, establishing the #middlefingerdunk hashtag.
- 24 October 2026: The phrase first shows up in Google Trends worldwide, while TikTok's topic feed grew past 503.7K posts by 27 October.
The base clip shows real people on an outdoor court, so every AI clip in the trend is a remake built from new photos. That's why each new video rebuilds the entire scene from scratch around two portraits.
What the middle finger dunk is not
The name gets mixed up with real NBA moments, mainly Kawhi Leonard's 2020 playoff block on a Jamal Murray dunk. In that play, Leonard got his middle finger directly on the ball to stop a score. That's a defensive block in an official league game, not this social media trend.
People also call the meme format "basketball middle finger", "my prime 1v1 basketball", and "your prime 1v1 basketball moment". If you're searching for clips or tracking the format's spread across platforms, you'll want to check all three names alongside the main tag.
Reproducing it with an open model
If you want to build this sequence locally, you can look at open-weights tooling. Released in September 2025 by Wan-AI, Wan2.2-Animate-14B has a replacement mode built for character animation and character replacement. It takes a source video and a character image, swapping that character directly into the footage.
Building the full 8-second clip this way takes four steps:
- Source a clean base clip. You need your own master video running the exact handheld motion, crossover, rim hang, and walk-off.
- Prepare character inputs. You'd prep clear portrait shots of your dunker and defender, alongside matching game uniforms if you want their clothes to align with the scene.
- Run replacement passes. Because the model takes one character image per run, you run replacement twice: once for the dunker and once for the defender. Running this locally needs a strong GPU to run the 14B weights.
- Mix the audio track. You need an external audio pass to lay down the hip-hop beat and crowd reactions, syncing the sound hits to the rim hang and footwork.
That's quite a bit of plumbing to run by hand, but it's entirely doable if you've got the hardware and time to tune it.
Or call it done in one request
If you don't want to wire up a video pipeline, we built an endpoint for it. You can do all of it on the Middle Finger Dunk page.
Treated as an API, the contract looks like this:
- Inputs: One or two images (JPEG or PNG) plus a consent checkbox confirmation.
- Dunker (Photo 1): You in the white #7 jersey with blue and gold trim.
- Defender (Photo 2): Your friend in the red #11 jersey. If omitted, a made-up defender joins the play automatically.
- Settings: a quality choice of 480p, 720p, or 1080p. There aren't text prompts to write and you won't tune camera paths.
- Output: An 8-second 9:16 MP4 delivered in a few minutes, complete with embedded crowd noise, hip-hop backing track, and no watermarks.
We build this page to deliver finished clips directly. One 720p video is $9.90 with no sign-up required, and you can buy one-time credit packs that stay valid for 12 months. If a job fails, the credits return to your account balance automatically, so you don't lose anything on an error.
Here's how the finished video looks from that one request:
Checklist before you post one
Before you generate or publish a clip to your channels, run through these requirements:
- [ ] One face per input: Make sure each uploaded file contains exactly one person.
- [ ] Clean framing: Pick images where subjects look directly into the camera without heavy occlusions or hard angles.
- [ ] Age verification: Confirm that every person featured in the source photos is 18 years of age or older.
- [ ] Image rights: Verify that you hold the legal right to use both uploaded portraits.
- [ ] Aspect ratio: The exported 9:16 vertical container fits TikTok, Reels, and Shorts feeds directly without cropping.
Disclosure: we build Seadanse, the site whose Middle Finger Dunk page made the sample in this post.
Top comments (0)