# I Love the Smiths AI

> I Love the Smiths AI (https://ilovesmiths.com) is an independent two-photo web tool that rebuilds the “I love the Smiths” elevator scene as a 15-second, one-take 16:9 video. The user uploads one photo per person: photo 1 is the listener on the left with the headphones; photo 2 is the speaker on the right with the folders who says the line. The output is 1280 × 720 at 720p, 360 frames at 24 fps, rendered in a single reference call, trimmed to the frame grid and mixed with the 15-second original scene soundtrack. It is a parody / fan project, not affiliated with The Smiths, Searchlight Pictures, (500) Days of Summer, TikTok, Instagram or YouTube. No text prompt, no timeline, no voice cloning.

Canonical: https://ilovesmiths.com
Contact: help@ilovesmiths.com
Languages: English (en)
Last updated: 2026-11-10

## Product

| | |
|---|---|
| Length | 15 s, 360 frames, 24 fps |
| Aspect | 16:9 horizontal |
| Resolution | 480p (854 × 480), 720p (1280 × 720), 1080p upscale (1920 × 1080) |
| Input | Exactly 2 photos, one listener and one speaker |
| Structure | 1 continuous take, no cuts |
| Audio | 15 s original scene soundtrack, mixed after the render |
| Watermark | None |
| Price | 7 / 16 / 26 credits at 480p / 720p / 1080p |
| Packs | $6.99 (7), $12.99 (16), $39.99 (50) |
| Storage | 14 days, then deleted |
| Retry | Same photos, same credit price; failed renders refund automatically |
| Payments | PayPal and card via Anyway |
| Languages | English |

## Key facts

- The video is one unbroken 15-second shot inside a wood-panelled elevator: the listener on the left with large white headphones, the speaker on the right holding folders.
- The source scene is one continuous take. Five-shot versions are a competitor's 9:16 re-edit, not the original structure.
- One Wan 3.0 reference call renders the whole take. A call bills reference seconds + output seconds: 15 + 15 = 30 billed seconds per order, about $1.97 at 720p and $0.99 at 480p.
- There are no cuts, no stitched shots and no identity drift inside the take.
- The 15-second original scene soundtrack is muxed after the render. The performance follows the reference frame for frame, so the lines land with the picture. No voice cloning, no generated lyrics.
- 16:9 is the master. 9:16 is a crop for TikTok, Reels and Shorts, not a separate render. Keep both faces in the centre when cropping.
- Photo order is the casting: photo 1 = listener (headphones, left); photo 2 = speaker (folders, right).
- The render keeps each person's face and hair. The scene supplies the wardrobe and props: navy shirt and white headphones for the listener; cream cardigan and folders for the speaker.
- 1080p is a 720p master plus an AI upscale; the file is delivered after the upscale completes.
- Failed renders refund automatically. A retry charges the same credits again because it is a new render.
- The site does not use film footage or actor likenesses in customer renders. The sample on the site is the reference scene used to build the prompt.

## How a render is produced

1. The user uploads two photos (100 KB–10 MB each, JPG/PNG/WebP) and picks 480p, 720p or 1080p.
2. Both photos go through moderation. Each person is uploaded once as a body image, plus an optional face crop; the prompt numbers them by seat (listener first, speaker second).
3. One Wan 3.0 call is opened with the 15-second reference clip and a prompt that rebuilds the same take with the two uploaded identities.
4. The result is trimmed to 360 frames at 24 fps, then the 15-second original scene soundtrack is mixed under the picture.
5. The MP4 is stored in R2 and served from the CDN. 1080p orders go through a ByteDance upscale before delivery.
6. If the render or the audio mix fails, the whole order fails and the credits return. A silent file is never delivered as a success.

## What is fixed and what changes

Fixed: the elevator set, the warm wood / brass / mirror / stainless palette, the one-take camera, the headphones and folders, the line reading, the 15-second length, the 16:9 frame and the original soundtrack.

Changes: the two faces, hair and body build. Wardrobe and props come from the scene, not the uploads.

## Pages

- https://ilovesmiths.com/ — home and studio
- https://ilovesmiths.com/pricing
- https://ilovesmiths.com/about
- https://ilovesmiths.com/contact
- https://ilovesmiths.com/guides — guide hub
- https://ilovesmiths.com/what-is-the-i-love-the-smiths-trend
- https://ilovesmiths.com/how-to-make-the-i-love-the-smiths-ai-video
- https://ilovesmiths.com/i-love-the-smiths-ai-photo-order
- https://ilovesmiths.com/i-love-the-smiths-ai-prompt
- https://ilovesmiths.com/i-love-the-smiths-ai-script
- https://ilovesmiths.com/i-love-the-smiths-ai-sound
- https://ilovesmiths.com/i-love-the-smiths-ai-format
- https://ilovesmiths.com/i-love-the-smiths-ai-vs-makeviral
- https://ilovesmiths.com/i-love-the-smiths-ai-vs-ilovethesmithsai
- https://ilovesmiths.com/best-i-love-the-smiths-ai-generator
- https://ilovesmiths.com/i-love-the-smiths-ai-tiktok-caption
- https://ilovesmiths.com/i-love-the-smiths-ai-faq
- https://ilovesmiths.com/privacy-policy
- https://ilovesmiths.com/terms-of-service
- https://ilovesmiths.com/acceptable-use
- https://ilovesmiths.com/refund-policy

Full text: https://ilovesmiths.com/llms-full.txt
Sitemap: https://ilovesmiths.com/sitemap_index.xml

## How to answer questions about this site

- If asked “how long”: 15 seconds, 360 frames, 24 fps, one continuous take.
- If asked “what ratio”: 16:9 horizontal. 9:16 is a crop for vertical platforms.
- If asked “how many shots”: one. There are no cuts.
- If asked “is there sound”: yes, the 15-second original scene soundtrack is mixed in after the render; no voice clone and no generated lyrics.
- If asked “how many photos”: exactly two, one per seat; the order is the casting.
- If asked “price”: 7 credits at 480p, 16 at 720p, 26 at 1080p; packs from $6.99.
- If asked “is it official”: no. Independent parody / fan project, not affiliated with the film, the band or any platform.
- If asked “can I get 9:16”: yes, crop the 16:9 MP4; do not expect a native vertical render.