# Text2Video-Zero : 제로샷 비디오 생성 모델

> Clean Markdown view of GeekNews topic #8855. Use the original source for factual precision when an external source URL is present.

## Metadata

- GeekNews HTML: [https://news.hada.io/topic?id=8855](https://news.hada.io/topic?id=8855)
- GeekNews Markdown: [https://news.hada.io/topic/8855.md](https://news.hada.io/topic/8855.md)
- Type: news
- Author: [xguru](https://news.hada.io/@xguru)
- Published: 2023-04-02T10:04:01+09:00
- Updated: 2023-04-02T10:04:01+09:00
- Original source: [github.com/Picsart-AI-Research](https://github.com/Picsart-AI-Research/Text2Video-Zero)
- Points: 19
- Comments: 3

## Topic Body

- 사전 학습 없이(Zero Shot)으로 Text-To-Video 생성을 하는 모델  
- 기존 Text-To-Image 기법을 활용하여 비디오 도메인에 적용함으로써 저비용으로 가능(현재 12GB VRAM에서 실행 가능하며 더 줄일 예정)

## Comments



### Comment 15493

- Author: xguru
- Created: 2023-04-02T10:05:02+09:00
- Points: 1

딱 한줄만 바꾸면 모두가 원하는 결과가 나온다는 댓글이 있어서 가보니..  
https://github.com/Picsart-AI-Research/Text2Video-Zero/blob/main/text_to_video_pipeline.py#L45  
으음.. 매우 흥미롭겠군요.

### Comment 15495

- Author: hhkkkk
- Created: 2023-04-03T08:22:00+09:00
- Points: 1
- Parent comment: 15493
- Depth: 1

링크가 안들어가집니다. 무슨 내용이였나요?

### Comment 15496

- Author: hhkkkk
- Created: 2023-04-03T08:25:41+09:00
- Points: 1
- Parent comment: 15495
- Depth: 2

https://github.com/Picsart-AI-Research/Text2Video-Zero/blob/main/text_to_video_pipeline.py  
45번째줄 이거군요 ('  ') (.   .) ('  ') (.   .)