---
title: 'What ChatGPT Image 2 Does Is Scary'
source: 'https://youtube.com/watch?v=XnNyZf37Kao'
video_id: 'XnNyZf37Kao'
date: 2026-08-09
duration_sec: 67
---

# What ChatGPT Image 2 Does Is Scary

> Source: [What ChatGPT Image 2 Does Is Scary](https://youtube.com/watch?v=XnNyZf37Kao)

## Summary

The video demonstrates the impressive image editing capabilities of ChatGPT's Image 2 feature, showing how it can seamlessly replace a person in a photo while preserving the scene's details and consistency. The creator highlights the tool's ability to maintain continuity and change angles, contrasting it with Gemini's limitations.

### Key Points

- **Image Replacement** [00:01] — The creator copies an image and uploads it to GPT, then adds a picture of themselves and prompts GPT to place the boy from image two into the place of the man from image one. The result is described as 'really good,' with all original elements preserved.
- **Object Addition** [00:15] — The creator prompts GPT to place a MacBook on the table, and the tool successfully adds the object while maintaining the scene's realism, resulting in a more professional-looking photo.
- **Angle Change** [00:28] — The creator asks GPT to take a picture as if it were taken from behind the person sitting in the chair. GPT successfully changes the angle while keeping the scene's continuous flow, such as the shelf, hands, and books in the same positions.
- **Comparison with Gemini** [00:54] — The creator notes that GPT's error rate is much lower than Gemini's for this type of task, based on their observations. They invite viewers to comment with their own experiences.

### Conclusion

ChatGPT Image 2 demonstrates advanced image editing capabilities, particularly in maintaining scene consistency and handling angle changes, outperforming competitors like Gemini in these tasks.

## Transcript

found this image here, and I liked it. I'm going to copy this image and bring it here into GPT.  And now I'm going to add a picture of myself.  I'm going to put this prompt here.  Place the boy from image two in the place of the man
from image one.  GPT already did it, and man, it turned out really good.  So, he kept exactly everything that was in the image here and just really added it.  Look how fantastic!  Place the MacBook on the table. Look how cool!  So here you
have a much more professional photo. Take a picture as if it were taken from behind the person sitting in the chair.  So here I'm change the angle, which is something that Gemini has some difficulty
changing.  Look, he did it here, you can see that he kept the scene much a continuous flow.  See? Shelf here going there.  You can see that the hands are in the same position, and the books are still up here.  In other
words, the error rate is much lower, from what I've seen there, when you do this kind of thing with GPT than with Gemenai.  Comment below what you think, if you've already tried it.