What Photos Work for NEON AI Dance? A Two-Photo Guide

What photos should I use for a NEON duo dance?

Use two separate photos with one consenting adult in each. Choose a clear face, even lighting, and enough of the outfit to understand the person's appearance. You do not need a photograph of both people together. The tool brings the two inputs into one requested 15-second AI dance performance using a prepared setup.

JPG, PNG, and WebP are accepted, up to 3 MB per image. The short side must be at least 340 pixels, and the long side must be no more than 2.5 times the short side. Those are upload requirements. Lighting, framing, and clothing visibility are practical suggestions that can make the source easier to interpret; they do not guarantee an exact likeness or a particular quality of motion.

Start with the face you want the dancer to resemble

Choose a recent-looking, recognizable image of the adult who agreed to the edit. Look for a face that is in focus and large enough to inspect without guessing at the eyes, nose, or mouth. A straightforward angle is usually easier to evaluate than an extreme profile or a head turned away from the camera.

Avoid masks, dark sunglasses, hair across the eyes, and objects covering the lower face. Strong beauty effects or filters may change the features you actually want represented. If the source has already replaced facial detail with smoothing, compression, or blur, the generator cannot be relied on to recover the original appearance. Select another authorized photo when the face is unclear instead of repeatedly enlarging a small crop and expecting new detail to appear.

Use even light and keep the person distinct

Pick a photo where you can see the face and outfit together. Deep shadow on one side, a bright window behind the person, or colored lighting that hides facial features can make the image harder to judge. A simple background can help you identify which details belong to the person, although matching backgrounds are not required.

Before selecting a photo, view it at a normal size and then zoom in once. If the face becomes a block of pixels, look for a better original. If the image is sharp but too dark, use a better-lit source where possible. Heavy editing can create an attractive still while removing useful identity cues. The goal is a readable reference, rather than a heavily stylized portrait whose features are already partly invented.

How much of the body should be visible?

A knees-up or full-body image gives more information about clothing, proportions, and the relationship between the face and body. Leave some space above the head and around the shoulders. Avoid a crop that cuts through the face or makes an important clothing detail sit directly on the edge.

A close portrait provides less information about the rest of the outfit. An image can pass the file checks without specifying all the details a dance scene needs. The model may interpret those missing areas, and even visible details can change during movement. If keeping an outfit recognizable matters to your project, choose a photo that shows it clearly, then inspect the completed video. Do not treat a full-body input as a promise that every accessory, logo, or fabric pattern will remain unchanged.

Can we use photos from different places?

Yes. One photo can be yours and the other your friend's, taken in a different location or on a different day. Separate photos are the intended workflow, so there is no need to create a collage before upload. Place one image in each dancer slot and check the previews individually.

Similar camera angles and lighting can make it easier to compare the pair before generating, but the site does not require an identical setting. The current form does not offer a custom background selector. Do not assume either source background will be copied exactly into the result. If your main requirement is to place people over a particular backdrop with precise control, consider whether an editing workflow better matches that project before spending generation credits.

What if my only picture is a group photo?

Start from an image you have permission to edit, then crop it to one adult before uploading. Remove other faces and leave enough room for the intended person's head and outfit. Check the exported file again: cropping can make a previously small face clearer on screen without actually adding any detail to it.

The final cropped image still has to meet the size and shape limits. If the subject occupies only a tiny part of the original photograph, a different source is usually a better choice. Do not upload a full group and rely on the system to identify the person you mean. A first and second upload slot can identify two separate inputs, but those slots do not give the generator a reliable way to choose one person from a crowd inside either image.

File size, dimensions, and format examples

File size and pixel dimensions measure different things. A photo can be under 3 MB but still have too few pixels, or have plenty of pixels while exceeding the file-size limit. Check both after exporting. A 600 by 900 pixel image meets the minimum dimension and shape requirements; a 300 by 900 image fails the short-side minimum and is too narrow for the allowed ratio.

A panoramic 500 by 2,000 pixel picture has enough pixels on its short side but exceeds the 2.5 ratio limit. Crop it to a less extreme rectangle if the person remains clearly visible. Export JPG, PNG, or WebP using an image tool rather than just changing the filename. GIF, SVG, video files, and HEIC are not accepted as dancer photos in this form.

How do I prepare a large phone photo?

Export an accepted copy from your device's photo or image tools. For HEIC, choose an actual JPG, PNG, or WebP conversion. If the file is over 3 MB, reduce its dimensions or use compression while keeping enough visible detail in the face. Recheck the saved copy, because the file selected in the browser must itself meet the requirements.

Keep the original separately so you can compare it with the smaller version. Avoid sending your image through a chain of repeated screenshots and messaging apps merely to reduce the size. Each step can alter the crop or degrade details. If the exported photo still looks unreadable, choose a different source before generating. The signed-in setup creates a compressed preview, which gives you another chance to inspect the image before submission.

Permission and private information matter too

Ask each adult whether they agree to being used in an AI dance edit and where the finished video may be shared. Permission to keep a photo in a personal album is different from agreement to generate and publish a new performance. A publicly visible image does not establish either kind of permission.

Use photos you own or are authorized to upload. Do not use minors, pets, celebrity impersonations, identity documents, or images intended to harass or deceive. Crop out addresses, badges, screens, and unrelated private details. You do not need those details to make a duo dance. Read Safety and consent if you are unsure whether the proposed use fits this workflow, and settle the question with both people before submitting.

Check the draft before using credits

Sign in and open the generator, select one photo for each person, and check that the previews are assigned correctly. Use Swap if you want to exchange the two positions. Choose landscape 16:9 or portrait 9:16, then 480p for 10 credits or 720p for 30 credits. Both resolutions request a 15-second video; neither guarantees perfect likeness.

Photo setup is free. The draft stays in this browser and can be restored for up to 24 hours. Use Clear draft to remove it when finished; other devices must be checked separately. Generate submits the images for processing and creates a submitted video request, which is separate from the local draft. Continue with the generation tutorial, or read Privacy before deciding whether to submit.