Upload a front-facing selfie, check the night preview and press Create video; for two people, start in Image.
Seedance 2 Mini: text, frame and media reference video generation
Google's latest model with exceptional quality output
With one image, Grok follows the image aspect ratio.
Seedance and MiniMax also bill the actual duration of input videos. Documents: maximum 50 pages.
When this option is enabled, the output video may be selected by AI Generate and published to the Explore.
Photo 1 on the LEFT and Photo 2 on the RIGHT; keep both faces, hair and clothes. Both stand close side by side in a late-night house-party kitchen with string lights, a crowded counter and friends blurred in the background; the left one holds a phone up like a microphone, the right one grins. Close handheld framing as if a friend is filming. Vertical 9:16. No text, no logos, no real celebrities.
The credit cost shows on the Image button before you generate.
When Image finishes, tap “Animate in Video”: the still goes in as the first frame with the prompt.
Open VideoA solo at a bus stop and two friends in a party kitchen, remixed from Seedance 2.0 Fast and Seedance 2 Mini.
Solo at the bus stop starts from one front-facing selfie, which became a night still and then the clip. It is a remix of a Seedance 2.0 Fast clip (720p, 9:16, 6 seconds): trimmed, pushed in slightly, graded sage green and stamped ghibliia.com top left, never mirrored.
Two friends in the party kitchen starts from two separate portraits that went straight to video with no still. It is a remix of a Seedance 2 Mini clip (480p, 9:16, 10 seconds) with the same trim, grade and stamp. Both people are invented adults, and no credits were spent on either sample here.
A woman in a mustard hoodie films at arm's length under a rain-wet bus shelter at night; friends laugh behind her and she ends by pointing at the lens.
Remix of a Seedance 2.0 Fast clip Use this photo
He raps into a phone held like a mic under string lights; she laughs beside him, then joins in with rounded lips.
Remix of a Seedance 2 Mini clip Do it with 2 photosUpload a front-facing selfie with the mouth visible, check the night preview, press Create video.
One face, looking straight at the camera, evenly lit, mouth uncovered, framed from head and shoulders to waist. Side angles, sunglasses and heavy beauty filters make the face slide around or the mouth turn rubbery.
The effect is already loaded when the page opens: vertical 9:16, a phone held close with a wide lens, a street at night and people behind you. Pick a scene above the maker and look at the preview; if your face sits small or turned, swap the photo first.
Press Create video. The credit cost shows on that button before every run. The clip opens on the mic check, with head nods on the beat and a finger at the lens. Download it and take it to the app where you'll add the sound.
JME opens his 2017 GRM Daily verse with check one, check two; the AI puts your face in that moment.
On TikTok it also goes by the JME rap filter, and the CapCut template is listed as “JME Rap check 1 check 2”. The AI version took off around October 19, 2026, when a post built on that template reached about 100K likes; most of the early clips came from UK creators.
“Check one, check two” is what you say when testing a microphone. In the trend it is the opening line: someone raps right into a phone camera at night, crowd behind, nodding and pointing at the lens.
JME is Jamie Adenuga, a grime MC and producer from Tottenham, North London, active since the early 2000s. He is Skepta's younger brother and co-founded the crew and label Boy Better Know. The sound comes from his part in the Boy Better Know freestyle with Skepta, Frisco and Jammer, posted by GRM Daily on January 12, 2017.
Luke Dean, Locky and JME released a track called “Check One, Check Two” on July 10, 2026. It is a different recording, which is why a sound search shows two results; the filter runs on the 2017 freestyle.
Combine two photos on Image: one raps at the lens, one reacts behind; animate it on Video.
The original shot already has a mate at the rapper's shoulder hyping him up. To split the roles, each person brings their own photo so every face keeps its place. Build the scene first in Image with both photos as references in one request, then open it in Video as the first frame. Both steps sit in the two-photo card on this page, and the party kitchen sample shows the split.
A front-facing photo, mouth visible, waist-up. This person stands in front, close to the lens, and opens with check one, check two; in the sample it's the guy in the denim shirt holding the phone.
A second, separate photo, also front-facing and well lit. This person stays beside or behind and laughs, nods or points at the rapper, like the woman in the red sweater. Two separate photos keep the faces from blending or swapping spots.
The clip has no JME recording; add the sound on TikTok and line up the first word.
AI generators don't hand over JME's recording: most export a silent clip or make new audio. The original sound gets added in the app you post from, through its sound library; Instagram Reels works the same way through Audio.
Download the video from Ghibliia and start a new TikTok post with that file.
Tap Add sound and search “check one check two” or “JME”. Pick the 2017 freestyle, not the 2026 single by Luke Dean, Locky and JME.
Trim the start of the sound until “check one” lands on the first mouth movement. If the mouth runs early or late, move the trim point rather than changing the speed. For another take from your own image as the first frame, use Video.
JME is the stage name of Jamie Adenuga, a grime MC and producer from Tottenham in North London who has been releasing music since the early 2000s. He's Skepta's younger brother and one of the founders of Boy Better Know.
It took off around October 19, 2026, after a TikTok made with a CapCut template passed roughly 100K likes. Most of that first wave came from UK accounts.
No. The single by Luke Dean, Locky and JME came out on July 10, 2026; the trend runs on JME's 2017 GRM Daily freestyle. Pick the freestyle when you search for the sound.
One person facing the camera, evenly lit, mouth not covered, cropped somewhere between head-and-shoulders and waist-up. Profile shots, sunglasses and strong beauty filters tend to make the face drift.
Usually a different recording was added or the speed was changed afterwards. Trim the start of the sound so the first word lands on the first mouth movement.
The solo is a remix of a 6-second Seedance 2.0 Fast clip and the duo is cut from a 10-second Seedance 2 Mini clip. Both were trimmed, reframed and color graded, and this page spent no credits on them.
