AI Tiny DeskGuide
Can I Make a Tiny Desk AI Video With My Friends?
Learn how to add friends, choose band roles and check photos, credits and expected results before generating.
What changed — First publication: role-by-role instructions, photo requirements, costs and the existing example.
The Short Answer
Yes. AI Tiny Desk lets you upload a lead-singer portrait and optionally add up to four friends as the saxophone player, keyboard player and left/right backing vocalists. Put a separate portrait of one consenting adult in each role you want to replace. You do not need to write a prompt or upload a reference performance.
Two or three portraits are enough; you do not need to fill every slot. Unfilled roles follow the reference band rather than disappearing. Each video is 15 seconds, with no extra credits for adding friends at the same resolution. Faces, movement and audio can vary.
Article updates
- First publication: role-by-role instructions, photo requirements, costs and the existing example.
Watch the existing video example
This is the existing homepage result with one uploaded lead-singer portrait. The still is captured from that same video. It illustrates the setting, movement and sound; it is not a verified multi-photo example or a guarantee of group results.
Choose your five band roles
- Lead singer — required
- Saxophone player — optional
- Keyboard player — optional
- Left backing vocalist — optional
- Right backing vocalist — optional
Upload directly into the slot for the person’s role. Empty optional roles follow the reference performers; generated faces may vary. Two portraits replace two roles, not the entire cast with just two people.
Prepare separate portraits
Use clear, evenly lit portraits with one adult per image. Avoid group photographs, hidden faces and heavy filters. JPG, PNG, WEBP and HEIC/HEIF are supported, up to 20 MB each. Get each friend’s permission for uploading and AI processing. Minors and public figures are not supported.
Make your video step by step
- Upload the required lead-singer portrait.
- Expand the friends section and upload a separate portrait into each desired role.
- Select resolution and aspect ratio; the default is 720P with adaptive framing.
- Sign in or register, complete required email verification and buy credits if your balance is insufficient.
- Review photos, settings, credits and permissions, then confirm generation. Buying credits alone does not start the video.
- Check task status in your creations or My Videos. Play the complete result with sound and download the MP4.
Credits and output
Each 15-second video costs 240 credits at 480P, 480 at 720P or 960 at 1080P. Adding friends does not change the price at the same resolution. Credit packs are one-time purchases; free previews are disabled. Generation usually takes 5–15 minutes and can take longer.
PricingCheck the finished result
Check each person’s face across cuts, role placement, hands, instruments, movement and sound. Higher resolution does not guarantee better resemblance. Original recordings, exact lyrics and precise lip sync are not guaranteed. Confirmed technical failures release reserved credits; dissatisfaction with a delivered result is reviewed separately under the refund policy.
Refund policyFrequently asked questions
Can I use just two or three people?
Yes. Supply the lead singer and one or two optional portraits. Other band roles remain based on the reference.
Separate portraits or one group photo?
Use separate portraits, one adult in each image. The tool does not assign multiple people from one group photograph.
What happens to empty roles?
They follow the reference band’s appearance and placement, with possible generated variations. They are not removed.
Does adding friends cost extra?
No. The current credit cost depends on resolution, not the number of uploaded roles.
Can I choose who sings and plays each instrument?
Yes, within the five fixed role slots. You cannot add arbitrary instruments or redesign the performance.
Will everyone appear in the finished video?
The model is instructed to follow your assignments, but clear visibility, equal screen time and consistent identity are not guaranteed.
Does it include the original Outstanding song?
The model generates audio using the reference performance. Original recordings, lyrics and voices are not guaranteed. Buying generation does not grant rights to the original song.