From One Portrait to a Reference-Guided Dance Performance
Quick Answer: How to Make a Talk Dirty Dance Video in Under 3 Minutes
The current generator follows a fixed, reference-guided workflow. It does not ask for a second photo, background upload, or custom production prompt. This guide explains the settings and checks that ensure high-quality results with the underlying Wan 3.0 generation model.
Table of Contents
What You Need Before Getting Started
You need one portrait of yourself or an adult who has explicitly agreed to appear in an AI-generated video. The file must be JPG, PNG, or WebP and no larger than 15 MB. A clear face, visible hair, and an unobstructed upper body provide more identity information than a distant full-body shot.
Choose the photo before thinking about prompt language. The standard generator does not require a prompt because its dance reference is already configured. What the model needs from you is a usable visual identity, a target aspect ratio, and enough credits for the selected resolution.
Check that the person is an adult and that you have permission to animate the photo. The service may perform content safety checks. Do not upload photos of minors or people who have not authorized the result.
Step 1: Upload a Clear Portrait
Open the generator and select Browse Files or drop a photo into the upload panel. The preview should show a recognizable face and clear hair silhouette. Avoid photos where the face is turned far away, covered by a mask or sunglasses, or compressed into a tiny area of the frame.
If the photo is too bright, the face can lose detail. If it is too dark, the model has less information for the replacement lead. Neutral indoor lighting or soft daylight is usually more useful than a dramatic colored light.
Step 2: Choose Aspect Ratio & Video Quality
The three aspect-ratio buttons use a phone, monitor, and square icon. Select the output for the final platform, not the shape of the source photo. Then choose the quality tier that matches the desired detail and credit budget.
| Setting | Best use | Credits |
|---|---|---|
| 9:16 | TikTok, Instagram Reels, YouTube Shorts | 10 / 20 / 40 |
| 16:9 | Landscape and widescreen edits | 10 / 20 / 40 |
| 1:1 | Square social posts and previews | 10 / 20 / 40 |
| 480P | Lower-cost first attempt | 10 |
| 720P | Balanced default for social video | 20 |
| 1080P | More pixel detail when available | 40 |
Do not confuse resolution with aspect ratio
Step 3: Generate Your AI Dance
Press Generate Video after confirming the photo and settings. The button changes to Generating Video and becomes disabled while the request is active. The upload preview then shows the single progress message, with a My Creations link if you want to leave the page.
Wan 3.0 uses the black-hat lead dancer as the replacement target. The model follows the reference for full-body movement, group coordination, rhythm, and camera energy. It does not guarantee a frame-perfect copy or an exact likeness. Fast turns, overlapping dancers, and hands are the most common areas to inspect.
Do not submit a second request while the first is active. The generator prevents duplicate submission and can reconnect to an active task when you return to the page.
Step 4: Preview and Download
When the task succeeds, the preview panel replaces the upload photo with the generated video. Play it once at normal speed and once while pausing on the turn, group crossing, and final pose. Check the face for unexpected changes, the body for sudden proportion shifts, and the background for unstable stage details.
Download the MP4 from the result panel. If the file is intended for a social platform, keep the original downloaded file instead of re-recording the preview. Use Create Another to clear the result and start a new generation.
A completed clip is 15 seconds. If your social edit needs a shorter cut, trim around the strongest movement rather than slowing the entire clip and weakening the rhythm.
Tips for Better Face and Body Consistency
Use one subject
A single adult face gives the model a clear identity target. Group photos create conflicting information.
Keep facial detail visible
Avoid masks, sunglasses, extreme shadows, and beauty filters that erase skin structure.
Preserve hair shape
Hair is one of the strongest identity cues during full-body movement and turns.
Prefer a half-body portrait
Shoulders and upper-body proportions help the lead dancer remain coherent in the group scene.
Inspect hands and profile frames
These are common drift points in generated dance video, especially when the lead is partly hidden.
Choose an appropriate quality
Higher resolution can preserve more detail, but a poor source photo will not become sharper simply because 1080P is selected.
Common Problems and How to Fix Them
The image will not upload
Confirm the actual format is JPG, PNG, or WebP and the file is under 15 MB. Re-exporting an image from a messaging app can help when the filename and file format do not match.
The face changes during the turn
Use a clearer, more front-facing portrait and inspect the result at the most extreme profile angle. No current image-to-video workflow can guarantee perfect identity at every frame.
The account has insufficient credits
The selected quality requires 10, 20, or 40 credits. Choose a lower resolution or review the current credit packs on the pricing page.
Talk Dirty Dance AI FAQ
Make Your Talk Dirty Dance Video
Start with the full-body workflow and inspect the result before sharing.