The Tiny Desk AI video trend is taking over TikTok, with AI-generated versions of celebrities, football coaches, and fictional characters performing "Outstanding" in an NPR-style Tiny Desk concert.

The trend is inspired by Charlie Wilson's 2023 NPR Tiny Desk performance, particularly the moment when he and his band perform The Gap Band's "Outstanding." Creators are using AI to replace the original musicians with their own characters while recreating the recognizable camera movements, singing, and instrumental performances.

One of the most popular early examples featured college football coaches whose teams had just lost, reimagined as members of a Tiny Desk band.

@theburnerfromhe11 College Football’s Week 8 All-Loss Tiny Desk Concert 😭🎤 #alabama #collegefootball #indiana #georgia #texas ♬ original sound - The Burner

How to Make the Tiny Desk AI Trend

• What You Need to Make the Tiny Desk AI Trend
• Step 1: Create Five Character Reference Sheets
• Step 2: Turn the Tiny Desk Performance Into Four Motion References
• Step 3: Generate Each Tiny Desk Clip With Seedance 2.5
• Step 4: Combine the Four Clips in the Tiny Desk AI Template
• Frequently Asked Questions

What You Need to Make the Tiny Desk AI Trend

To recreate the trend, you'll need:

• Five character reference images: One for each person or character appearing in your Tiny Desk performance.
• Four motion-reference clips: Short sketch-style videos showing the movements, camera angles, and timing you want to recreate.
• An AI video generator: We're using Seedance 2.5 through Kai.
• A video editing template: To combine the four clips and synchronize them with "Outstanding."

Rather than asking AI to recreate the entire performance in one generation, we'll generate each section separately. Here's how the four clips are organized:

Step 1: Create Five Character Reference Sheets

Start by choosing the five people or characters you want to feature in your Tiny Desk concert. These can be real people, fictional characters, celebrities, or original AI-generated characters. The important thing is that each character has a clear reference image.

For the best results, choose images where:

• The subject's face is clearly visible.
• Their hairstyle and facial features are easy to identify.
• Their clothing and accessories aren't obscured.
• The image isn't heavily filtered or distorted.

If you're using photos with different lighting, angles, or image quality, it helps to standardize them before generating the videos. You can use the following prompt to create a consistent reference portrait.

Next, open a Character Sheet Generator and create an AI character sheet for each of the five subjects. Character sheets provide multiple views of the same person, helping the AI maintain their appearance when the camera angle changes.

It's also worth deciding what each character should wear at this stage. For example, if you're making a version featuring football coaches, you might want each character to wear a recognizable team jacket or hoodie.

Including those outfits in the character references is more reliable than asking the AI to invent clothing during video generation.

Step 2: Turn the Tiny Desk Performance Into Four Motion References

Next, we need to recreate the camera movements and lighting from the original performance.

Instead of feeding the AI the original concert footage directly, we're using a sketch-style motion-reference technique. This involves converting the source footage into simple black-and-white outlines.

These outlines preserve the information we want the AI, while reducing the visual detail that might cause the model to reproduce the original performers rather than the characters we provide.

For this trend, we've divided the sequence into four sections.

• Clip 1: Group performance
  The first reference features three performers together. This establishes the setting and introduces the band.
• Clip 2: Solo close-up
  The second reference focuses on a single character performing the prominent solo moment.
• Clip 3: Instrumental solo
  The third reference features one character performing the instrumental section.
• Clip 4: Final group performance
  The fourth reference returns to the same three characters from Clip 1.

Keeping the clips separate means we can assign different character references to each generation without confusing the AI with too many identities at once.

It also means that if one performance doesn't look right, we only need to regenerate that section rather than starting the entire video again.

Step 3: Generate Each Tiny Desk Clip With Seedance 2.5

Now that we have our character sheets and motion references, we can generate the individual videos.

Open Kai, select Seedance 2.5, and upload the character references and motion-reference video for your first clip. We'll use slightly different prompts for the group performances and solo shots.

Generate Clip 1: Three-Person Group Performance

For the first clip, upload:

1. Character sheet 1
2. Character sheet 2
3. Character sheet 3
4. Motion-reference video 1

Copy the following prompt.

Generate Clip 2: Solo Performance

For the second clip, upload:

1. Character sheet 4
2. Motion-reference video 2

This section focuses on one character, so we can simplify the prompt:

The most important thing to check here is facial consistency.

Because the character is featured in a close-up, any distortion around the eyes, mouth, or face will be more noticeable than in the wider group shots.

Generate Clip 3: Instrumental Solo

The third clip follows the same process, but this time we're recreating the instrumental performance.

Upload:

1. Character sheet 5
2. Motion-reference video 3

AI video models can struggle with musical instruments, especially when fingers move quickly or the instrument crosses in front of the face.

If the instrument becomes distorted, regenerate this clip with a stronger instruction to preserve its shape and positioning.

Generate Clip 4: Final Three-Person Group Performance

For the final clip, we'll return to the same three characters from the opening.

Upload:

1. Character sheet 1
2. Character sheet 2
3. Character sheet 3
4. Motion-reference video 4

Reuse the group performance prompt from Clip 1, replacing the motion reference with the fourth clip.

Step 4: Combine the Four Clips in the Tiny Desk AI Template

Once all four generations are finished, the final step is to combine them into a single video and synchronize the performance with the song.

Open the Tiny Desk AI template. The template contains four video placeholders arranged in the correct sequence, with the audio track positioned underneath.

To assemble your video:

1. Select the first video placeholder and click Replace.
2. Upload your generated Clip 1.
3. Repeat for Clips 2, 3, and 4, replacing the corresponding placeholders.
4. Play the entire video and check that the transitions line up with the music.

Because the clips were generated individually, some may be slightly longer or shorter than their intended positions in the template.

If a clip is too long, trim the beginning or end rather than changing its playback speed. This helps preserve the natural movements and performance timing.

Once you're happy with the result, click Export Project and download the video.

You can then share it on TikTok, Instagram Reels, or YouTube Shorts. Use the platforms' AI-generated content labels where required, and make sure you have permission to use the images and audio.

Frequently Asked Questions