Back to blog

Talk to Me Doja AI Guide: Make the Cheek-Cupping Sway Video from One Photo (2026)

Nov 17, 2026Talk to Me Doja Editorial TeamTalk to Me Doja Editorial Team

Complete Talk to Me Doja AI guide for 2026: turn one portrait photo into a 12-second 9:16 cheek-cupping sway video from $4.99 with no subscription.

This Talk to Me Doja AI guide walks you through every technical and creative detail behind the November 2026 cheek-cupping sway trend. Instead of filming ten takes in front of a ring light or keyframing anime overlays by hand, you can upload a single front-facing portrait to the Talk to Me Doja AI Video Generator and receive a full 12-second, watermark-free 9:16 vertical MP4 in three to eight minutes. Below, we cover the anime scene that started the format, the five choreography beats an authentic clip must hit, how single-photo motion transfer compares to CapCut templates, and how to publish your finished video on TikTok, Instagram Reels, and YouTube Shorts.

Who Is Urara Shiraishi and Why Is Everyone Cupping Their Cheeks?

Before generating your clip, it helps to know why the gesture looks so specific. The choreography comes directly from Episode 1 of Yamada-kun and the Seven Witches (Yamada-kun to 7-nin no Majo), which aired on April 12, 2015 and was animated by Liden Films from Miki Yoshikawa's manga. In that opening episode, rough-edged high school delinquent Ryuu Yamada accidentally swaps bodies with top honors student Urara Shiraishi after a fall down the school stairs.

When student council recruiter Toranosuke Miyamura corners Urara (who is actually Yamada inside her body) and asks her to join the Supernatural Studies Club, Yamada tries to mimic how he thinks a bashful honors girl behaves. He places both palms softly against his cheeks, tilts his head, pouts, and sways left to right.

On November 7, 2026, TikTok creator @lovingbalanci paired that exact 2015 animation loop with a sped-up cut of Doja Cat's "Like That" featuring Gucci Mane (originally released November 7, 2019 on Hot Pink) under the on-screen caption "Speak to me Doja". That first upload earned 229,000 views in five days. Two days later, on November 9, 2026, live-action creators jumped in: @the_real.ryan pulled 3.1 million views in three days by acting out the shy cheek-cupping sway next to the anime reference, and @bunnicutieluv added another 744,000 views. By November 10, 2026, creators like @twin.cities.mattress and thousands of AI video remixers had turned "Talk to me Doja" into one of the month's biggest short-form formats, documented on Know Your Meme. For a deeper look at the anime plot and creator timeline, read our Talk to Me Doja Meme origin breakdown.

The 5 Storyboard Beats Every Authentic Talk to Me Doja AI Video Hits

Generic image-to-video prompts fail on this trend because they make the subject wave randomly or spin around. A recognizable Talk to Me Doja clip runs 12 seconds and follows five timed beats mapped to the rhythm of the sped-up track. In the Talk to Me Doja AI Studio, these five beats are pre-wired into our motion template:

BeatTimestampMotion NameWhat Happens on Screen
Beat 10–2sCool Pose

Starts facing the lens with a calm, unbothered expression before the beat drops.

Beat 22–5sCheek Cup

Brings both hands up softly under the chin and cheeks, framing the face cleanly.

Beat 35–8sBashful Sway

Tilts the head and sways gently left to right with Urara's shy pout.

Beat 48–10sShy Glance

Looks away for a beat, then glances back into the camera with a flustered blush.

Beat 510–12sCamera Lean

Leans forward toward the lens with a warm, playful smile on the final count.

Because the motion starts with a two-second unbothered pause before the hands come up at the 2-second mark, you have plenty of room at the top of the frame to place on-screen text like "Speak to me Doja" or "POV: she actually replied to your DM".

Comparing the 3 Ways to Make a Talk to Me Doja Video

Creators joining the trend usually pick between three workflows: using our dedicated single-photo motion generator, cutting slides inside a CapCut template, or writing raw prompts in a general-purpose video tool. Here is how those three methods stack up in practice:

FeatureTalk to Me Doja AICapCut TemplatesRaw AI Video Prompts
Input Needed1 portrait photo (JPG, PNG, or WebP)3–6 self-recorded clips or still slides1 photo + manual multi-beat prompt tuning
Actual Motion

Full 5-beat cheek-cupping sway performed by your photo

Screen shakes, zooms, and anime picture-in-picture

Frequent extra fingers or missed cheek-cup timing

Duration & Format12-second 9:16 vertical MP47–15 seconds depending on creator cutOften limited to 4s–5s segments
WatermarkZero watermark on every downloadTemplate badge unless subscribedOften stamped on entry tiers
Cost Structure

One-time packs from $4.99; credits never expire

Free basic or recurring monthly Pro plan$15–$95/month subscription commitment

If you already filmed yourself doing the dance, CapCut works fine for adding beat flashes. If you only have a still photo and want the person in that photo to actually bring both hands to their cheeks and sway for 12 full seconds, a tuned motion-transfer preset saves hours of trial and error.

Fast vs. Premium Models: Credit Costs, Resolutions, and Render Times

You can watch the sample video before using credits. If a task is marked failed or rejected after credits are deducted, those credits automatically return to your balance.

1. Fast Tier (bytedance/seedance-2-mini)

  • Fast · 480p Standard (15 credits): The $4.99 Starter pack covers one clip. Estimated render time: 3–5 minutes.
  • Fast · 720p HD (22 credits): Estimated render time: 4–6 minutes.

2. Premium Tier (bytedance/seedance-2-fast)

  • Premium · 480p Standard (30 credits): Estimated render time: 4–6 minutes.
  • Premium · 720p HD (45 credits): The $12.99 Creator pack covers one clip. Estimated render time: 5–8 minutes.

The pricing page lists all three one-time packs: Starter ($4.99 / 15 credits), Creator ($12.99 / 45 credits), and Studio ($29.99 / 120 credits).

Step-by-Step Summary and Export Tips for TikTok, Reels, and Shorts

For a granular screen-by-screen walkthrough, check out our tutorial on how to make a Talk to Me Doja AI video. Here is the fast checklist to go from photo to published post:

  1. Pick a well-lit half-body portrait: Choose a JPG, PNG, or WebP file under 20 MB (minimum 300 × 300 px) showing one adult facing forward. Because the AI dancer keeps the clothing from your photo, upper-body or waist-up shots where your arms and hands are not cropped out produce the cleanest cheek cup. Read our 7 photo tips for Talk to Me Doja AI videos before uploading.
  2. Confirm rights and choose a version: In the Studio, check the photo rights box, pick Fast or Premium at 480p or 720p, and choose whether to include AI-generated music.
  3. Add the song when posting: Turn audio off before generating or mute the clip in your social app. Search the app’s music library for “Like That.” See the song guide for tips on lining the song up with the cheek cup.
  4. Share links carefully: Anyone with a working media link can open the file. Download the videos you want to keep.

Ready to turn your favorite portrait into the 12-second cheek-cupping sway? Head straight to the Talk to Me Doja AI Video Generator or open the Studio workspace to start your render.