How to Make a NYC Rats AI Video With Your Own Photos

Updated

Short answer

You need two photos, one per person: the first person runs at the camera, the second throws the big bluff kick. Both must be adults who agreed to appear. Add them on the NYC Rats AI page and generate; each video uses 10 credits.

Make yours from your own photos

13 s · 720p · 9:16 · with music · 10 credits per video

Pick your photos

Before you start

Generation is open on the NYC Rats AI page. You sign in, add two photos and spend 10 credits per video, so do the preparation below first and the credits go to a clip you will want to keep.

The steps

  1. Pick two different adults who agreed to appear. Photos of children are not accepted.
  2. Decide who is who. The first person starts on the left, runs at the camera and throws the kick that lands near the lens. The second starts on the right and throws the big, exaggerated sideways kick from far away.
  3. Choose one clear, front-facing, well-lit photo of each person. Use one person per photo and avoid sunglasses, masks and heavy filters.
  4. Add the two photos to the cards on the page. Each card checks the file type, size and shortest side before accepting it.
  5. Press generate. The clip is rebuilt from a first frame in one continuous pass, then you download the vertical 9:16 video.
  6. Post it and say it is AI-generated.

Photo checklist

  • JPEG, PNG or WebP, under 10 MB.
  • The short side at least 256 pixels.
  • Face clearly visible, no one else in the shot.
  • Two people who look clearly different from each other. In our test, one person in a yellow hoodie and one in an olive jacket stayed distinct through the whole clip.

How your two photos are used

The clip starts from a single first frame: your two photos and a street layout are combined into one still that shows who stands where. Each person’s face, hair and clothes are meant to come from their own photo, and the background comes from the street layout, not from your photos.

The video is then generated from that first frame in one continuous pass, with no cuts and no stitching, so the shot stays unbroken. The camera moves and the order of kicks are described in the prompt. We keep about 12.7 seconds of it and add a rhythmic beat.

Getting the best from your photos

  • Use daylight or even indoor light. A clear, evenly lit face gives the first frame the most to work with.
  • Stand about an arm’s length from the camera so the face fills a good part of the photo, but do not crop in so tight that hair and shoulders are cut off.
  • Keep the expression natural. In close-ups the generated faces tend to look serious and focused, so a photo with an extreme expression may not carry over.
  • Make the two people easy to tell apart: different clothes and hair help the two identities stay separate.
  • The background of your photos does not matter. The first frame takes the person from each photo and puts them on the street layout.

What the result looks like, and what it does not

The moves follow the same order as the original, but we cannot promise exact seconds. The pace can be a little faster than the original, mouths open and close without matching the music, and street details differ from the original location. The sample was made with two AI-generated adult stand-ins: one in a yellow hoodie and one in an olive jacket stayed distinct through the whole clip. Real user photos are not tested yet.

Check the clip before you share it

  • Watch the whole clip once: the two identities should not swap or blend at any point.
  • Look at hands and feet during the kicks. That is where we saw the most variation in our tests.
  • Make sure nothing in the frame could be mistaken for real footage of a real event.
  • Say in the caption that it is AI-generated.

What to expect

The sample runs about 13 seconds and shows the moves in a fixed order: the big bluff kick, the run at the lens, a camera turn, close-up hops, a second approach, and a swing away past a yellow sign. The setting is a daytime street. The aim is the same order of moves, but we cannot promise the exact seconds, and faces may differ from your photos. Real user photos are not tested yet, so check the clip before you share it.

Questions

Can I generate the video today?

Yes. Sign in on the NYC Rats AI page, add your two photos and press generate. Each video uses 10 credits from a one-time pack, and the finished MP4 is kept for 7 days for you to download.

Can I use one photo for both people?

No. The scene needs two different people, one photo each, so the two identities stay apart.

Can I use a pet?

We tested two stand-in animal characters that already stand upright (a hippo and a crocodile) and the result held up. We have not tested real pet photos such as a cat or dog on four legs, so we do not promise them.

How long is the finished clip?

About 13 seconds in our sample, vertical 9:16.

Will the faces look exactly like my photos?

The aim is that each face, hair and outfit comes from its own photo, and that held in our test with two sample photos. Real user photos are not tested yet, so check the clip before you share it.

Sources

Facts on this page come from these sources. Links open in a new tab.

  1. 01Instagram, @aliaboutine: Never Take The Shortcut in NYC (Reel, credited to Higgsfield) (November 10, 2026)
  2. 02Xiaohongshu, repost by DearKwok: The original street clip, reposted (post in Chinese)

Keep reading

Make your own NYC Rats AI video

Pick two photos and generate your own version. Each video uses 10 credits from a one-time pack. Not affiliated with the original creators or Higgsfield.

Pick your photos