What a template video provides
The site supplies one fixed template video and its poster. That source defines the bodies, clothing, instruments, movement, camera, and timing. Photos provide replacement faces for three labeled targets: Lead singer, Keys, and Sax. Lead singer is required. Leaving Keys or Sax empty keeps that performer’s existing face. The tool uses this same video throughout the setup and does not ask you to choose or upload another source clip.
Set up a template video face swap in 3 steps
Step 1 Review the fixed template video
Watch the configured source and locate Lead singer, Keys, and Sax. Check face visibility, head turns, camera cuts, and any microphone in front of a mouth. These are recorded features of the fixed performance. The video is already available in the tool; there is no template-selection step or pending source-video placeholder.
Step 2 Upload and map face photos
Upload one clear adult portrait to Lead singer. Add separate Keys and Sax photos only for the supporting faces you want changed. Use the role labels to assign each photo. Leave optional slots empty to keep the source faces. Sax remains the third target even when no Keys portrait is uploaded.
Step 3 Preview and check your face mapping
Use Preview my setup to review the fixed video, mapped role names, portrait count, and frame preference. Photos remain in your browser and are not saved with the draft. Video processing and finished downloads are not connected; a future download will require a real successful edit of this source video.
The fixed template video
Solo Soul Concert
- Source scene
- The fixed office-concert source video shown on the homepage, with targets labeled Lead singer, Keys, and Sax.
- Source framing
- Keep the recorded positions of Lead singer, Keys, and Sax. Preserve the source framing and keep each replacement assigned to its labeled performer.
- Face photo
- One clear adult portrait for Lead singer is required. Add separate portraits for Keys and Sax only when you want to replace those faces.
Use the fixed Solo Soul Concert template video as the source footage. Map portrait slot one to Lead singer (required), portrait slot two to Keys (optional), and portrait slot three to Sax (optional). Replace each face only with the portrait explicitly mapped to that role. If the Keys slot is empty but Sax has a portrait, apply that portrait to Sax and leave Keys unchanged. Never infer the target from the number or upload order of the photos. Leave every unmapped person unchanged. Preserve the source bodies, hairstyles, clothing, instruments, gestures, mouth movement, camera motion, lighting, background, and timing. Keep each replacement assigned to the same performer throughout the clip, including head turns and brief occlusions. Blend face edges and skin tone with the source lighting without redesigning the room. Retain cleared source audio only where supported; do not generate a song or clone a voice. Do not add people, logos, broadcast graphics, scenes, or new performance actions.
Map the three source performers
| Performer | Photo slot | How to map the face |
|---|---|---|
| Lead singer | Required · slot 1 | Upload a single adult face for the Lead singer before opening a settings preview. A photo in another slot cannot satisfy this requirement. Choose visible eyes, an unobstructed jawline, and a face angle reasonably close to the source performer. The portrait supplies facial identity; it does not replace the source body, hairstyle, clothing, microphone, or voice. Removing this photo makes the setup incomplete until a Lead singer portrait is added again. |
| Keys | Optional · slot 2 | Use this slot only when you want to change the Keys performer’s face. A portrait of that person’s face is enough; the reference does not need to show a keyboard, hands, or a performance pose. Leave Keys empty to preserve its original face. The slot keeps its own role identifier throughout the setup. Adding or removing a Sax portrait does not move the Lead singer or Sax reference into Keys. |
| Sax | Optional · slot 3 | Use this slot for the Sax performer’s face reference. It remains assigned to Sax when Keys is empty, so you can preview Lead singer and Sax without uploading a third person’s photo. Choose a clear adult face and avoid a reference hidden by an instrument or hand. The source clip supplies the instrument, recorded gestures, and timing. Leaving Sax empty preserves that existing face instead of removing the performer or creating a substitute. |
Check the mapping before saving
Before saving, read the three labels and check the uploaded preview in each occupied slot. One portrait should identify one intended person. If a face is assigned incorrectly, remove that photo from the wrong slot and upload it under the correct label. The count shows occupied roles out of three; it does not replace the Lead singer requirement. A setup with only Keys and Sax is still missing its required main portrait.
The saved preview lists the mapped performers so you can check the assignment again. It stores settings and role names, not the actual image files. Reopening a draft cannot restore the photos for a future processing request. When real face replacement becomes available, inspect the returned video through head turns and occlusions, compare unchanged people with the source, and listen to any retained audio before sharing it.
Video inputs and preservation settings
The fixed video must accompany the face references in a real processing request. A portrait plus a scene prompt does not identify the clip to edit. Keep explicit role-to-photo assignments for Lead singer, Keys, and Sax; optional gaps must not shift another photo into a different role. No numeric seed or identity-strength setting is promised because the supported controls depend on the processing provider.
Keep source timing and camera movement in the instructions. Treat 9:16 and 16:9 as output preferences to confirm against the fixed video, not as new viewpoints. Ask to retain cleared template audio only if supported. A face replacement prompt does not request new music, new lip-sync, or a cloned voice.
Match the face photo to the performer
Choose a sharp face with visible eyes and natural skin detail. Avoid sunglasses, masks, heavy beauty filters, and tiny faces inside group photos. Match the target’s general face angle where possible. The reference does not transfer your hairstyle, clothes, body, or background into the template. Review the edited hairline and jaw rather than assuming every surrounding detail will match your photo.
For the soulful template-video variation, use the outstanding tiny desk ai guide. To upload the Lead singer portrait and optional Keys or Sax portraits, return to Tiny Desk AI.
Template & Prompt FAQ
Is a Tiny Desk AI template a video or a text prompt?
It is the existing video to be personalized. The accompanying prompt tells the processor which faces to replace and what to preserve. The prompt cannot substitute for the source video input.
What happens when I leave a face slot empty?
Keys and Sax are optional: leaving either slot empty keeps that performer’s original face. Lead singer is required, so leaving it empty prevents a settings preview. Empty optional slots do not add or remove people. Real multi-face editing will depend on the connected processing provider.
Can I copy the face replacement prompts?
Yes. Copy the complete prompt for the fixed video and keep its Lead singer, Keys, and Sax assignments intact. Apply a face reference only to its labeled role. These are this site’s suggested instructions, not an official program asset or a guarantee of a flawless swap.
Do templates change the singer’s voice?
No. This workflow changes selected faces. It does not make a performer sing in your voice or supply a commercial recording. Retaining a source soundtrack depends on both processing support and the rights to that audio.