Seen the Jensen Huang dancing video and wondered how to put yourself into a similar concert scene? With faceswap, you can upload a reference music video, choose a portrait for each scene, and generate a performance inspired by its movements.
Start with the AI dance video and motion transfer tool, or follow the preparation steps below. This tutorial explains our photo-and-video workflow. It does not reconstruct the original creator's production process or promise identical results.
What is the Jensen Huang AI world tour video?
The creative reference is “Jensen Huang World Tour” on Bilibili, published by 王与也行. It places a recognizable public figure in an AI-created concert setting, with stage entrances, close-ups, and performance shots.
The reference video is the creator's work, not a faceswap-generated example. Its finished appearance does not establish which models, software, or editing techniques the creator used. Our tool page embeds the original player and links to its source for inspiration.
If you are searching for “how to make a Jensen Huang dancing video,” the practical starting point is a reference performance and a clear portrait. Use your own footage or material you have permission to edit.
What you need to make your own AI dance video
The reference video supplies the motion, shot sequence, and original audio. The photos supply an appearance reference for the scenes you choose to generate.
| Input | Current support | What to choose |
|---|---|---|
| Reference music or dance video | Standard MP4, up to 100 MB | Start with a short clip and one clearly visible performer |
| Portrait photos | JPG, PNG, or WebP, up to 10 MB each | Clear facial detail with minimal blur or obstruction |
| Scenes you want to leave intact | Keep original footage | Useful for audience shots, titles, scenery, or complex group scenes |
Try a short video first so you can judge whether the portrait and movements work together before processing a longer MV. Our face swap photo quality checklist can help you compare portraits, but a good still image does not guarantee consistent video results.
How to make an AI world tour video, step by step
1. Upload your MP4 and analyze the scenes
Open the MV Motion Transfer page, sign in, and upload your reference video. You only need the video at this stage; you can choose the portraits after scene analysis.
Once analysis finishes, review the time range and original preview for each segment. Look for a clearly visible subject, rapid turns, overlapping people, and objects such as microphones or glasses that obscure the face.
2. Choose a portrait for each segment
Add your photos to the portrait library. Click a scene on the timeline to preview it, choose a portrait, and select Apply to this scene. To reuse one face, select several scenes and use Apply to selected scenes, or apply the portrait to all generated scenes. For example, use portrait A for the opening performance and portrait B for a later shot. Wait for the assignments to save before generating.
Select Keep original for any segment that should stay as it is. These segments remain in the final video without character generation.
Choosing a portrait per segment does not let you select a specific person within a group shot. The current tool cannot target only the person on the left or right in a scene with multiple people. Use single-person footage, or keep a difficult group scene unchanged.
The generation button becomes available once every segment selected for generation has a portrait.
3. Submit all segments together
Start generation once you have reviewed the whole sequence. The service queues the selected segments, processes them, and then assembles the video. You do not need to start each segment separately.
The page shows the task status and completed segments. Processing time depends on the amount of footage being generated, the scene content, and the queue. There is currently no fixed completion-time guarantee.
4. Review the results and download the full video
Watch the individual segment previews and the assembled result. Pay particular attention to head turns, hands crossing the face, glasses, and fast movement. Check for identity changes, flicker, and abrupt transitions between scenes.
If one generated segment needs another attempt, use its regenerate option without uploading the entire MV again. Failed tasks also have a retry option.
The service joins the segments in their original order and combines them with the source video's audio. You can then preview and download the complete MP4. This workflow does not compose a new song or generate a new singing performance.
What to look for in an AI dance video generator
For a project inspired by the Jensen Huang AI world tour video, check whether the tool supports the choices your footage requires:
- Your own reference video: uploading an MP4 lets you work with a chosen performance rather than relying on a fixed dance template.
- Scene-by-scene portraits and original footage: a music video may mix solo performances, group shots, and scenery. Review each scene before deciding what to generate.
- Segment previews and regeneration: inspect the whole sequence and redo individual scenes before downloading the final video.
faceswap provides these controls, but results still depend on your inputs. Motion transfer can change more than the face: clothing, body details, and backgrounds may also vary. Review the generated footage rather than assuming every detail will be preserved.
Trial pricing
As of October 15, 2026, video generation, failed-task retries, and individual segment regeneration do not deduct credits during the trial.
The displayed proposed rate is 1 credit per generated second. The estimate adds together the durations of the segments selected for generation, then rounds that total up to a whole second. Segments kept as original footage are excluded. The actual trial charge remains 0 credits; the tool page will explain any future billing changes.
For example, if a 30-second MV contains 12 seconds to generate and 18 seconds to keep, the proposed estimate is 12 credits. The current trial charge is 0 credits. Photo editing has its own pricing, so check the video tool's current trial notice before starting.
Frequently asked questions
Can I make a dancing video from my own photo?
Yes. Upload a reference MP4, then assign your portrait to the segments you want to generate. The video provides the movement reference; a portrait alone is not enough for this workflow. Start with a short, clear solo performance.
Can I generate an entire world tour by typing a prompt?
The current tool requires a reference video and portrait photos. It does not turn a single text prompt into a complete concert production, and it does not automatically reproduce all the editing choices in the original world tour video.
Can I change only one person in a group dance?
The current workflow does not support choosing a particular person within a multi-person shot. You can assign different portraits to separate segments, use clearer single-person footage, or keep the group scene as original footage.
Can I reuse the original Bilibili video?
Viewing the reference or its embedded player does not grant permission to download, modify, or republish it. Use footage you own or have permission to edit. When sharing a result, identify it as AI-generated or AI-edited so viewers can understand what they are seeing.
Why do glasses, side profiles, and fast movements cause problems?
These scenes provide changing or partially hidden facial information. Try a clearer portrait or a shorter solo shot, and keep a difficult segment unchanged if needed. The face swap artifact troubleshooting guide explains visible problems to look for around the eyes, hairline, jaw, and skin.
Ready to try? Open MV Motion Transfer, upload a short MP4, and set up your first AI dance video scene by scene.
For a reference video of your own, use the music video face swap workflow to plan generated scenes, original clips, and source audio.
