What You're Trying to Make
You've seen the baseball one — the live broadcast quietly holding on a fan in the stands, except the fan turns out to be AI. The basketball version is the same illusion moved to the most expensive seat in the building: the camera cuts to courtside, you're sitting front row at what looks like the NBA Finals, and the commentary booth can't help but react. The comments fill with "wait, who is that courtside??" — which is the entire point.
This guide walks through how to make one yourself: what photo to use, exactly what prompt to write, which AI model to run it on, and how to post it so it actually travels. By the end you'll have a repeatable, generation-ready workflow.
The courtside cut is the celebrity-row variant of the fan cam. Where the stadium fan cam is about being oblivious, courtside is about aura — you're allowed to clock the camera, because of course they cut to you.
Where This Came From
The fan-cam format was born in Korea in May 2026, when a single AI-generated clip of a focused KBO baseball fan racked up tens of millions of views before anyone realized she wasn't real. Within days creators jumped it to every sport — NFL jumbotrons, F1 paddocks, Wimbledon's Centre Court, and the one that's peaking right now: NBA courtside. Know Your Meme catalogued the broader format as the Korean AI courtside trend on May 12, 2026, and tools like Media.io now ship dedicated "NBA courtside" generators.
The timing helps. The 2026 NBA Finals are on, the series hits Madison Square Garden for the middle games, and MSG's celebrity row is already the most-screenshotted real estate in sports. Dropping yourself into that seat is the most on-trend version of the format available this month.
The Fastest Way — Use the Courtside Template on Starrd
The Courtside template is live in the Starrd library. It packages every step in this guide — the telephoto broadcast framing, the static NBA score bug, the celeb-cam cutaway structure, the commentary, the identity lock — into a single upload.
- Pick a clear face photo. One person, front or three-quarter view, eyes open, decent lighting. A regular phone selfie beats a retouched headshot.
- Open the Courtside template in the Starrd app or web library.
- Upload the photo and tap generate. The template personalizes the broadcast prompt to your face, runs the character sheet for identity lock, then generates a 12-second courtside catch on Seedance 2.0 — three broadcast angles cutting to the celeb-cam, then the aura beat.
One credit, a few minutes. No prompt writing, no model picking, no lens math.
Courtside
Upload one photo, get the courtside catch — front row at the Finals, the booth loses it. 1 credit, a few minutes, no prompt engineering.
The rest of this guide is for people who want to roll their own — pick a different arena, swap the matchup to your team, or run it on a model other than Seedance.
Or, Build It Yourself — What You Need
Three things:
- A clear face photo of the subject. Front or three-quarter view, eyes open, neutral or faintly confident expression, good lighting. One person only.
- Access to an AI video model that accepts a reference image. Seedance 2.0, Kling 3.0, Runway Gen-4, or any wrapper built on them.
- A platform to post on. TikTok, Reels, and X are where this format goes viral.
You don't need editing software, courtside tickets, or a jersey.
Step 1 — Pick Your Reference Photo
The photo you feed the model is the face that ends up courtside. Choosing well here saves you wasted generations later.
Use:
- A clear, well-lit photo of one person
- Front-facing or three-quarter angle
- Eyes open, neutral or faintly confident expression
- Natural lighting, minimal filtering
Avoid:
- Group photos (the model gets confused about who's who)
- Sunglasses or anything covering the face
- Low resolution or motion-blurred shots
- AI-generated reference images (compounding artifacts)
Unlike the upper-stands fan cam, the courtside version rewards a bit of styling — this is the celebrity-row seat, so an elevated, "got-ready" look reads as authentic rather than out of place. Just keep the face clearly yours.
Step 2 — Pick Your Arena and Matchup
This decision drives the whole scene, so lock it before you write the prompt.
NBA Finals at Madison Square Garden (most on-trend)
The Garden's courtside is the celebrity-row gold standard, and with the 2026 Finals live, an MSG courtside cut is maximally current. Use a real, plausible score bug — the matchup, the quarter and clock, a "NBA FINALS · GAME 3" tag — and keep it perfectly static through the clip. Wrong or animated graphics are the #1 way these get clocked.
Your Home Team's Arena
Less universally recognizable, but it lands harder with your own audience. Swap the arena, the team colors on the court, and the score bug abbreviations to your team. The broadcast grammar is identical.
Generic "national broadcast" arena
If you don't want to tie it to a specific real game, keep the arena unbranded — packed lower bowl, bright hardwood, a generic score bar. Slightly less punchy, but it never goes stale as a real series plays out.
Never put a real, identifiable person courtside at a game they didn't attend in a way that implies something about them. Korean media already covered one fabricated stadium clip that wrongly suggested bad behavior by a real public figure. Keep it yourself, your friends with consent, or clearly fictional.
Step 3 — Write the Prompt
The instinct is to make it cinematic — sweeping moves, dramatic light, slow motion. Resist all of it. A broadcast cutaway to courtside is near-locked, with at most a slow push-in. Boring camera, expensive subject. That's the formula.
Copy this and swap in your specifics:
Single continuous live national basketball broadcast cutaway, 12s, 16:9, ultra-realistic broadcast screenshot feel — NOT cinematic.
A confidently composed fan sits COURTSIDE, front row, at a packed NBA arena (Madison Square Garden energy). [Elevated courtside outfit — e.g. fitted black leather jacket over a sleek top, one delicate gold chain]. The bright hardwood and blurred players in [team] colors are directly behind them; the packed lower bowl rises in the back.
Telephoto broadcast lens, ~135mm equivalent, eye-level celebrity-row cutaway camera, slow subtle push-in only. Shallow depth of field — subject's eyes in sharp focus, court and crowd softly blurred.
Persistent broadcast graphics, PERFECTLY STATIC the entire clip: lower score bar reading "NYK 78 SAS 74", "3RD 5:18", small "NBA FINALS · GAME 3" tag. Small network logo glyph upper corner. No "LIVE" tag.
[0-5s] Subject lounges back, eyes on the game, unbothered, oblivious to the camera. [5-8s] Eyes flick up — no surprise, just small knowing recognition that the cam found them. They settle into the lens with strong, calm eye contact and a slight head tilt. [8-10s] One slow self-directed grooming gesture — a light hand through the hair, or a watch adjust. A small confident smile holds. [10-12s] Eye contact does not break. Smile holds to the cut.
Audio: live arena ambience only — crowd roar swelling and settling, sneaker squeaks, a ref whistle, muffled PA. Two male commentators: one trails off mid-call — "...oh, would you look at this courtside—"; the other, chuckling — "whew, that's pure aura right there." No music.
Unstaged candid broadcast moment, natural skin texture, slight live-TV compression, telephoto compression. No cinematic drama. No eye contact until the 5s mark. Pure live TV capture aesthetic.
The non-negotiable elements:
- "Telephoto broadcast lens, ~135mm equivalent" — the single most important line; the compressed-background TV look flows entirely from this.
- A static score bug — the courtside cut lives or dies on a believable, non-moving NBA graphic. Animated or shifting values give it away instantly.
- The oblivious-then-aura beat — 5 seconds unbothered, then the knowing eye contact and one self-directed grooming gesture. This is what separates courtside from the plain fan cam.
- "No music" — the moment a music bed plays, the broadcast illusion dies. Ambient arena audio only.
- Commentary, not narration — two announcers half-losing the play is the punchline. Keep it short and overlapping.
Step 4 — Pick a Model
- Seedance 2.0 — what Starrd runs on. Best-in-class prompt adherence, native 12s, strong audio generation (it'll voice the commentary). The safest pick, because the whole format depends on the model not going cinematic.
- Kling 3.0 — excellent for realistic on-camera characters and identity lock; great for the eye-contact aura beat.
- Runway Gen-4 — solid all-rounder, good identity preservation, slightly cinematic default you have to suppress.
- Veo 3.1 — capable and generates synced audio, but trends polished; needs extra "broadcast not film" prompt work.
If you have no preference, start with Seedance 2.0. (See our Seedance vs Kling vs Veo breakdown.)
Step 5 — Generate and Iterate
First generations rarely nail it. Common failures and fixes:
The score bug animates or changes mid-clip. Add: "Broadcast graphics are a static image overlay — identical position and values in every frame, no animation, no resetting."
The subject reacts too big (surprise, wave, cheer). Dial it down: "No surprise, no eyebrow raise, no wave. Only a small knowing smile and one calm self-directed grooming gesture."
It looks too cinematic / slick. Add: "1080i 60fps broadcast feed, broadcast color science not film, zero cinematic grading, slight compression artifacts."
The face drifts from the reference. Use a clearer front-facing reference, or weight the reference image more heavily if the model supports it.
The players behind look like melting screensaver motion. Crop tighter on the subject so less of the court is visible, or push the background further out of focus.
Budget 3–5 generations before a keeper. If you're 8+ deep, something structural is off — reread the prompt and hunt for anywhere you drifted toward "cinematic."
Step 6 — Post It
The generation is half the work. How you post determines whether it gets 200 views or 2 million.
Caption framing. The viral originals never said "made with AI." They captioned like a real broadcast moment — "who is that courtside 😭" or "front row at the Garden last night." That's how the algorithm files it as sports content, not an AI demo. (Platforms increasingly require AI-disclosure labels — comply, but your caption copy still matters within those rules.)
Keep it 16:9, then center-crop to 9:16. The telephoto compression only reads right in 16:9 natively; generate wide, then crop vertical for TikTok/Reels.
Timing. Post during US prime time, ideally around a real game window — sports content peaks when games are live.
Don't over-edit. No text overlays, no music, no zoom effects. The clip has to look unaltered. TikTok edits give away that you're trying to make it go viral.
Common Mistakes That Tank Your Video
- A moving or wrong score bug. The fastest courtside-specific tell. Static and plausible, always.
- Going cinematic. "Dramatic," "epic," "stunning" anywhere in the prompt pushes the model to AI-video defaults. Cut all of it.
- Camera movement beyond a slow push-in. Pans, orbits, and whip-zooms read as "produced."
- A big performative reaction. Courtside aura is unbothered. Save the gasp.
- Wrong lens. Wide angle looks like a phone. Telephoto compression is the entire look.
- Music. A music bed is the single fastest way to kill it.
- A recognizable celebrity reference face. The model will fight identity preservation and viewers will instantly clock it — plus it's the version of this trend that gets people in trouble.
Window of Opportunity
The fan-cam format has been running since May and the courtside variant is peaking now, riding the live Finals. Trends like this have weeks, not months, before saturation. If a viral courtside clip is the goal and the prompt engineering above sounds like a chore, the Courtside template at the top of this guide is this exact workflow in one tap.
Fan Cam
The original stadium fan-cam catch — same broadcast illusion, upper-stands version.
Frequently Asked Questions
How do I make an AI courtside NBA video? Pick a clear face photo, choose an arena (Madison Square Garden / the current NBA Finals is the most on-trend), write a telephoto broadcast prompt that frames you in a front-row courtside seat with a static NBA score bug and the court behind you, generate on a model that supports reference images (Seedance 2.0, Kling, Runway, or Veo), and post it like a real broadcast moment. Or use the Starrd Courtside template to skip the prompt work.
What is the courtside AI trend? It's the basketball version of the AI fan-cam trend that started in Korea in May 2026 with KBO baseball. Creators use AI to make it look like a live national broadcast cut to them sitting courtside — the "who is that courtside?" celebrity-row moment. The person was never at the game; the whole clip is generated from one photo.
What's the difference between the courtside cut and a regular fan cam? A fan-cam catch is in the upper stands and the subject is oblivious — micro-actions only, no eye contact. The courtside version is the celebrity-row cutaway: front row, expensive styling, and you're allowed to clock the camera. The best ones lean into "aura" — a small knowing smile, a self-directed hair or watch adjust — unbothered, like you expected to be on TV. The booth losing it is the punchline.
What AI model is best for courtside videos? Seedance 2.0 has the best prompt adherence, which matters because the format depends on the model NOT going cinematic. Kling 3.0 is strong for realistic on-camera characters. Runway Gen-4 and Veo 3.1 also work but need extra "broadcast-style" prompt language.
Why does my courtside AI video look fake? Usual tells: a cinematic camera move (cutaways are near-locked with only a slow push-in), wide-angle lens instead of telephoto compression, a music bed (use ambient arena audio only), an animated or wrong score bug, or a big performative reaction. Keep it candid and let the styling carry it.
Do I need to disclose that the video is AI-generated? Yes — TikTok, Instagram, and YouTube require AI-generated content to be labelled, and South Korea's AI Basic Act (in force since January 2026) pushed the trend toward disclosure. Comply with platform rules, and never make a real, identifiable person look like they did something at a game they never attended.
Can I make this without writing the prompt myself? Yes. The Starrd Courtside template handles the telephoto framing, the static NBA score bug, the celeb-cam cutaway structure, the commentary, and the personalization to your face. One upload, one tap — no prompt engineering required.
Related Reading
- How to Make a Viral AI Baseball Fan Video — the stadium fan-cam version this trend grew out of
- KBO Fan Cam AI: The Stadium Goddess Prompt Explained — line-by-line breakdown of the original viral fan-cam prompt
- How to Make a Viral AI F1 Paddock Video — the same broadcast illusion, F1 garage edition
- Seedance 2.0 Prompt Guide — the full framework for writing AI video prompts that don't look like AI video
- Seedance vs Kling vs Veo — which model to pick and why