At a glance
- What it is: upload one photo and AI replaces the performer in the viral Bulin 47 street freestyle, while Kevin Wai with the speaker, the red-lit restaurant entrance, the camera moves and the original dembow audio stay.
- Lowest per-render cost we found: ≈ $4.40 at Genjutsu's yearly-plan rate (148 credits, full 19s) or from ≈ $5.3 on Popliqo's Bulin 47 Freestyle template 480p tier (155 credits, 16s, best-value plan rate).
- Simple one-off options: Starrd ≈ $7.50 flat; bulin47ai.com $7.00–$9.99 per video depending on pack; Seadanse flat $9.90.
- There is no free tier. Freebies are below the cheapest video (Genjutsu's 3-credit signup bonus vs an 8-credit minimum), watermarked previews (vswp), or a silent 4-second re-creation (Renoise).
- Most tools use the same 19-second segment of Kevin Wai's clip; Popliqo's template makes 16 seconds, Seadanse 15, EasyVideoAI offers up to 23.
- Dates and names differ across sites (1 vs 2 October, "Wai" vs "Way", song out 7 vs 8 October). We flag what is verified and what is not.
- Reddit has no threads on this trend as of 14 October 2026 — verified by manual search. The conversation lives on TikTok and Instagram.
What is the Bulin 47 AI trend?
Bulin 47 AI is the AI remix wave built on a spontaneous street freestyle in Madrid, Spain, in early October 2026. Dominican dembow artist Bulin 47 — real name Hancel Teodoro Vargas Reynoso — was about to walk into a restaurant when Colombian creator Kevin Wai (TikTok @kevincitooficial11), known for stopping people on the street with a portable speaker, played him a dembow beat. Bulin 47 answered with an improvised freestyle and a dance move, and both posted their angles to TikTok the same day.
The moment that went global is one line: "No ando en película ni ando en paquete, ya todos me conocen, soy Bulin 47" — Starrd's gloss: roughly, "I'm not living in a movie, and I'm not up to anything shady. Everyone knows me, I'm Bulin 47."
The view counts kept snowballing, and each source reports a different number on a different date:
- 50 million+ plays across platforms, reported by Dominican newspaper Listín Diario around 7 October (cited by Starrd and Summrs).
- 60 million+ across platforms by 8 October (LosMocanos).
- 65 million on Instagram alone by 12 October (Summrs).
- The original TikTok posts: about 3.1 million views on Kevin Wai's upload and 3.3 million on Bulin 47's own (@bulin47rd), as tracked by bulin47ai.com on 9 October.
The AI trend started around 5–6 October 2026: creators recast the performer — most often with footballers like Messi, Ronaldo and Mbappé — while keeping the speaker, the crowd and every camera move. On 7 October tutorials flooded TikTok, and a one-photo template was used more than 170,000 times in its first two days (Seadanse). Bulin 47 leaned in, declaring "from today until December, every day is Bulin Day" (LatiNation, 8 October), and on 8 October he and Kevin Wai released the studio single "Dame Hombro", which hit #10 on YouTube's US trending chart within 24 hours (Summrs; the release date is 7 October on at least one tool site — see the conflicts table below).
A note on conflicting dates and names
Tool sites contradict each other on basic facts, so here is what we could and could not verify as of 14 October 2026:
| Fact | Version A | Version B | Our read |
|---|
| Street encounter | 1 October (Seadanse) | 2 October (bulin47ai, Dreamega, Summrs) | Majority says 2 October; Seadanse stands alone |
| First AI recasts | 5 October (Seadanse) | 6 October (bulin47ai, Dreamega, Summrs) | Majority says 6 October |
| "Dame Hombro" release | 7 October (Renoise FAQ) | 8 October (Seadanse, bulin47ai, Summrs, Genius page dated 8 Oct) | Majority says 8 October |
| Creator's name | Kevin Wai | Kevin Way | Both in wide use; Dominican press writes "Way", tools write "Wai" |
| Performer's outfit | Dark sunglasses + pink mesh shirt (Genjutsu, EasyVideoAI, bulin47ai, Summrs) | Tan checked three-piece suit (Starrd) | Likely different cuts of the same night; unverified |
| Core clip length | 19 seconds (most tools) | 23.15s / 16s / 15s / 4s (EasyVideoAI / Starrd+Popliqo / Seadanse / Renoise) | Tools built on different segments |
The safest way to cite any number in this niche: attach a date and a source, because the answer changes weekly.
Price comparison: what each tool charges
All rows were read from the listed pages on 14 October 2026. "Not stated" means the page does not publish the number — we did not borrow figures from third parties.
| Tool | You upload | What you get | Price (read 2026-10-14) | Not stated | Source |
|---|
| Popliqo (Bulin 47 Freestyle template) | 1 photo | 16s vertical 9:16 MP4, ~9 min estimated render; sample file carries an audio track (we probed it: 480×854, 16.04s) | Base 310 credits (720p default tier); 480p = 155 / 1080p = 775 credits. From ≈ $5.3 at 480p (best-value plan rate) | Whether the audio is the original recording | popliqo.com |
| Starrd | 1 photo | 16s 9:16, Seedance 2.5 at 480p, original audio, frame-by-frame lip sync | 150 credits ≈ $7.50; one-time, credits never expire, no subscription | Higher-resolution pricing | getstarrd.app |
| Summrs | 1 photo | 19s 9:16 MP4, no watermark, original dembow audio; only the performer changes | Not stated on the site (blog page has no pricing) | Price, resolution tiers | summrs.com |
| bulin47ai.com | 1 photo (optional crowd themes, e.g. World Cup kits) | 19s 9:16 MP4 with original sound (+ silent copy), no watermark, 5–20 min render | 480p = 10 credits, 720p = 22 credits; $9.99 per video; Trio 3-pack $24.99 (≈$8.33 each); Crew 100-credit pack $69.99 (≈$7.00 per 480p video); packs never expire | — | bulin47ai.com |
| Seadanse | 1 photo | 15s 9:16 up to 1080p, original sound, no watermark; runs Wan 3.0; no sign-up | Flat $9.90 for 15s at 720p; failed renders auto-refunded | Exact prices for 480p and 1080p (shown on the button at runtime) | seadanse.com |
| Genjutsu AI | 1 photo, or your own Bulin 47 clip + photo | 5–19s 9:16, original sound; full 19s only on Seedance 2.5 (other models ≤15s) | Full 19s: 148 credits (480p) / 332 (720p) / 588 (1080p); cheapest 15s-480p = 23 credits. Plans: $9/mo = 240cr, $29/mo = 780cr, $99/mo = 2,650cr; yearly $86–$950; lifetime packs 220cr/$12, 2,050cr/$99. Full freestyle ≈ $4.40–5.55 on subscription rates, ≈$8.07 on lifetime packs, up to ≈$30.83 at the $5/24cr trial-pack rate. 3 free credits < the 8-credit minimum; watermark until first purchase |
Not stated = the page did not publish the figure; we did not fill gaps with numbers from elsewhere.
Two things the table makes obvious:
Per-video pricing spreads nearly 9x — and the biggest lever is which rate tier you land on. The same style of clip, a 15–19s vertical recast, runs ≈$4.40 at Genjutsu's yearly-plan rate, from $5.3 on Popliqo's 480p tier, $7.50 flat on Starrd, and up to ≈$41 for EasyVideoAI's 23s 720p cut at pack rates. Trial packs are consistently the worst deal: Genjutsu's $5 trial pack prices the full freestyle at ≈$30.83 — about six times its yearly-plan rate.
Every "free" offer is structurally unusable. Genjutsu's 3-credit signup bonus cannot buy the cheapest 8-credit video. vswp's free preview is 5 seconds with a watermark. Renoise's free first clip is a 4-second silent re-creation, not the original freestyle. As Genjutsu itself puts it: "There is no unlimited free version, here or, as far as we found, on the dedicated Bulin 47 sites."
What each tool actually does
Popliqo's Bulin 47 Freestyle template
Popliqo's template sits at the cheap end of full recasts: 155 credits for 16 seconds at 480p (≈$5.3–7.0 depending on your plan), with a 1080p tier (775 credits) above it for maximum sharpness. Upload one photo, pick a quality tier (480p = 155 credits, 720p = 310, 1080p = 775 — the price shows before you pay, with 720p as the default), and the render estimate is about 9 minutes. We downloaded the template's sample output on 14 October 2026 and probed it: a 16.04-second, 480×854 MP4 with an AAC audio track, so the clip ships with sound. Popliqo runs a subscription model (entry: the 7-day Trial at $4.99) with automatic refunds on failed generations. The template went live on 9 October 2026 — within four days of the trend starting.
Starrd: the flat $7.50 option
Starrd's template renders 16 seconds at 480p on Seedance 2.5 with the original audio and frame-by-frame lip sync, for 150 credits (about $7.50), one-time, credits never expire. Its guide is also unusually practical about DIY pitfalls (see the how-to section). Pricing above 480p is not stated on the page.
Summrs: a guide without prices
Summrs' how-to blog (12 October) describes a 19-second, no-watermark, original-audio recast in four steps — and publishes no pricing at all. If you are comparing costs, the page cannot help you; that is a content-marketing choice, not an accident.
bulin47ai.com and Seadanse: the fixed-price one-offs
Both are purpose-built for this single trend. bulin47ai.com charges 10 credits per 19-second 480p video ($9.99, or $24.99 for a Trio of three — about $8.33 each; the 100-credit Crew pack at $69.99 brings a 480p video to about $7.00) with optional "crowd themes" that redraw bystanders as invented characters, and Stripe checkout with credit packs that never expire. Seadanse charges a flat $9.90 for 15 seconds at 720p with no sign-up and no watermark, running Wan 3.0. Both refund failed renders automatically.
Genjutsu AI: full price transparency, with a 6x spread between best and worst rates
Genjutsu publishes everything: the per-model credit matrix (the full 19-second freestyle costs 148 credits at 480p, 332 at 720p, 588 at 1080p, and only Seedance 2.5 makes the full clip — Seedance 2.0, 2.0 Fast, 2.0 Mini and MiniMax H3 top out at 15 seconds) and a real pricing page. The plans, read 14 October 2026: $9/mo = 240 credits, $29/mo = 780, $99/mo = 2,650; yearly $86–$950; lifetime packs 220 credits/$12 and 2,050/$99. At those rates the full freestyle runs ≈$4.40–5.55 on a subscription, ≈$8.07 on a lifetime pack — or ≈$30.83 if you only ever buy the $5 trial pack. Free accounts get 3 credits, less than the 8-credit cheapest video, and exports carry a moving watermark until your first purchase.
EasyVideoAI: consent-first, per-second billing
EasyVideoAI bills 80 credits per input second at 480p (180/s at 720p) against a $10 = 1,000-credit starter pack, so a 12-second clip runs about $9.60. You must check a likeness-permission box before generating, reviews are free, and credits are reserved only at the final generate click.
Renoise: the honest recreation
Renoise is the one site that explicitly does not touch the original clip: it generates a brand-new scene in the trend's look from your photo — red-lit entrance, friend holding a speaker, you freestyling — 4 seconds, 480p, silent by design, 20 credits per clip, with the first clip free if you follow the project on X and YouTube. Its own FAQ is candid about why: the original footage "belongs to the people who filmed it." If you want the real moves and the real sound, this is not your tool; if you want to sidestep the likeness question, it is the only one designed for that.
vswp.app: the Telegram route
vswp's guide offers a free ~5-second 480p preview with a watermark, then delivers the full clip inside Telegram paid with Telegram Stars — a payment path none of the Western tools use. Dollar pricing is never stated.
How to make your Bulin 47 AI video
The 30-second version of these steps is at the top of this article — this section is the longer one, with the choices that save a wasted generation.
- Pick a recast tool from the table above. The lowest per-render rates come with plans (Genjutsu yearly, or Popliqo's 480p tier); for a single video with zero commitment, Starrd or Seadanse are the simplest.
- Prepare one strong photo. Front-facing, sharp, evenly lit, eyes visible (take the sunglasses off — the original performer wears them, and the AI needs to see your eyes), chest-up or waist-up so your outfit carries over. Group photos, celebrity photos, and minors' photos are rejected by every serious tool.
- Upload and choose quality deliberately. The 480p tier is typically half the price of 720p; on TikTok and Reels, phone-screen quality differences are small.
- Generate and wait. Stated render times range from about 9 minutes (Popliqo estimate) to 5–20 minutes (bulin47ai.com); failed renders are refunded automatically on most platforms, including Popliqo.
- Post with disclosure. TikTok, Instagram and YouTube all require AI-generated content to be labeled (per Starrd's guide), and the audio may be matched against the original recording when you post.
The mistakes people hit most, per Starrd's DIY notes: muting the reference clip breaks the mouth timing (the model reads lip sync from the audio); swapping a profile frame gives the model too little face to hold onto; and generating an invented crowd fights the real camera movement — which is exactly why the dedicated templates keep the original crowd.
