---
title: "Reconstructed Slides: How fast are LLM inference engines anyway? — Charles Frye, Modal"
category: "slides"
video_id: "DeFF3J8T5Pk"
sourceLabels: ["Cropped public YouTube video frames", "Local OpenCV slide-region detection", "Local RapidOCR"]
---

# Reconstructed Slides: How fast are LLM inference engines anyway? — Charles Frye, Modal

## Source Video
[How fast are LLM inference engines anyway? — Charles Frye, Modal](https://www.youtube.com/watch?v=DeFF3J8T5Pk)

## Method
This deck is reconstructed from the existing video frame captures by detecting likely slide regions with OpenCV, cropping/upscaling those regions, deduplicating similar crops, and OCRing the cropped slide images locally. It is a cleaner companion to the full-stage frame deck.

## Reconstructed Slides
![[assets/reconstructed-slides/DeFF3J8T5Pk/slide-001.jpg]]

- Source frame: `slide-001.jpg`
- Crop: `full` `[0, 0, 960, 540]` score `160.12`
![[assets/reconstructed-slides/DeFF3J8T5Pk/slide-002.jpg]]

- Source frame: `slide-002.jpg`
- Crop: `full` `[0, 0, 960, 540]` score `166.1`
![[assets/reconstructed-slides/DeFF3J8T5Pk/slide-003.jpg]]

- Source frame: `slide-003.jpg`
- Crop: `full` `[0, 0, 960, 540]` score `166.89`
![[assets/reconstructed-slides/DeFF3J8T5Pk/slide-004.jpg]]

- Source frame: `slide-004.jpg`
- Crop: `full` `[0, 0, 960, 540]` score `162.14`
![[assets/reconstructed-slides/DeFF3J8T5Pk/slide-005.jpg]]

- Source frame: `slide-005.jpg`
- Crop: `full` `[0, 0, 960, 540]` score `159.38`
![[assets/reconstructed-slides/DeFF3J8T5Pk/slide-006.jpg]]

- Source frame: `slide-006.jpg`
- Crop: `full` `[0, 0, 960, 540]` score `176.23`
![[assets/reconstructed-slides/DeFF3J8T5Pk/slide-007.jpg]]

- Source frame: `slide-007.jpg`
- Crop: `full` `[0, 0, 960, 540]` score `176.58`
![[assets/reconstructed-slides/DeFF3J8T5Pk/slide-008.jpg]]

- Source frame: `slide-008.jpg`
- Crop: `full` `[0, 0, 960, 540]` score `162.74`
## Dense Scene-Detected Slide Candidates
- [[youtube-DeFF3J8T5Pk-dense-slides]]