---
tags:
  - reels
favourite: false
created: '2026-07-10T19:02:10+07:00'
modified: '2026-07-11T13:02:29+07:00'
draft: false
published: '2026-07-11T13:00:00'
published-url: https://www.instagram.com/reel/DapEPhGJgXb/
template: '[[Reels]]'
title: ChatGPT Sol Content
feature: '[[attachments/chatgpt-sol-content-cover.png]]'
published-cover-url: /2026/chatgpt-sol-content/attachments/chatgpt-sol-content-cover.png
published-cover-width: 640
published-cover-height: 1136
socialImage: /2026/chatgpt-sol-content/attachments/chatgpt-sol-content-cover.png
---

# ChatGPT Sol Content

![[attachments/chatgpt-sol-content-cover.png]]

# Goal
- To create another short clip 9:16 ratio content of myself talking toward the camera, but this time, the video part should completely built by you, that match the the audio and transcript i given to you.

# Generated Video
The video is a talking head video of myself in the condo room, a man talking and facing toward the camera. You can reference from my video
 make sure the generated video look like him, and must contain these signature
- Wearing a black t-shirt with graphic "anywhere-sidekick.png" (attached image)
- Holding microphone with wrench as a gimmick
- Keep the background like in referece video, it is his room
- (Ignore the freckles on his face, do not generated)

Below are some sample picture of myself from those video

![[main-avatar.png]]
![[cap-main.png]]

i put both into Heygen avatar
Let use Heygen avatar just to create a raw footage, a roll, my avatar talking to the camera
Start the vide with `main-avatar` and after shot 1, use `main-avatar-2nd-look-propeller-hat`
let use `Avatar V` to generate, i think current codex plugins didn't have that option. Then, u can use my chrome session to choose `Avatar V` to generate final video. 
For draft, if different model use different credit, use whatever that use lesser credit.


![[anywhere-sidekick.png]]

## Video rules
- The most key important part is that it must LOOK LIKE HIMSELF, a person in the video
- It MUST BE natural, realistic, clear facial skin, like it was recorded by himself , by the phone camera with small clear skin filter, like in other of his other video
- No camera movement, static camera, continuous, make fewest jump cut.
- Use the `content.m4a` in this folder and generate a video that match what he speak, make sure it look realistic, with emotion expression, lip, and moving, should match the tone of what the audio sound like
- There is the transcript below, but do not add subtitle in the generated video yet. just use it to generate srt file. Let output Genearted a clean footage first


# Shots
- Most of the video should be he talking to the camera 
- But let have these special shots 

## Shot 1
"แม้ แต่ วิ นา ที เดียว"
- Magically and instantly replace the hat on every spoken syllable.
- Each replacement must happen precisely on the matching audio beat.
- Do not use flashes, cuts, fades, or camera movement.
- The transformation should feel seamless, as though one hat magically becomes the next.
- Only the hat changes. Keep his face, head position, body, microphone, shirt, lighting, and background unchanged.
- Every hat must track his head naturally.
- We have Heygen avatar from all hats already
- Hey gen didn't allow to export/generate video less than 1seconds, but each syllable beat is less than a word

Hat sequence:
- **แม้:** 1st hat, **แต่:** 2nd hat, **"วิ"** 3rd hat, **"นา"** 4th hat, **"ที"** 5th Rainbow Bandana (not a hat), **เดียว**: last one,  main cap
On “เดียว,” transition seamlessly to the main-avatar-2nd-look-propeller-hat A-roll. Keep the propeller beanie for the remaining footage

![[cap-1st-strawhat.png]]
![[cap-2nd-m-cap.png]]
![[cap-3rd-pink.png]]
![[cap-4th-rabbit.png]]
![[cap-5th-bandana-headgear.png]]
![[cap-main.png]]

## Shot 2
"ไม่ต้องมานั่งหั่น Subtitle ภาษาไทยให้วุ่นวาย"
When he say these sentence, let that sentence
"ไม่ต้องมานั่งหั่น Subtitle ภาษาไทยให้วุ่นวาย" come to the screen 3D , like special effect, 
and he literraly use his hand to break those sentence into pieces

## Shot 3
"แค่สั่งหุ่นยนต์ ด้วย Prompt แค่ Prompt เดียว"

Let mascot avatar `N'KaoJai` floating in, and stay until the end of the clip
When flying in, let a man look at it, wink to him, while he speaking "แค่สั่งหุ่นยนต์ ด้วย Prompt แค่ Prompt เดียว"

after the wink, let the mascot then floating small, less movement, elegant, no expression


# Transcript 
Video ที่เห็นอยู่ที่ ไม่มีคนจริงๆ อยู่
"แม้ แต่ วินาที เดียว"

ที่คุณเห็นทั้งหมดตอนนี้ เป็นภาพของหุ่นยนต์
ที่สร้างโดย ChatGPT Sol Ultra
Model ล่าสุดจาก OpenAI ที่เพิ่งเปิดตัวไปเมื่อวานนี้

ไม่มีการตัดต่อ ไม่มีการเปิด Capcut
ไม่ต้องมานั่งหั่น Subtitle ภาษาไทยให้วุ่นวาย
แต่แค่สั่งหุ่นยนต์ ด้วย Prompt แค่ Prompt เดียว
อันนี้

ไม่รุ้อนาคตจะเป็นยังไง 
ที่คุณเห็น เป็นความจริง หรือจ้อจี้
แล้วเรื่องที่ผ่านมาอะ เป็นเรื่องจริงปะวะ
ความจริงคืออะไร มันยังน่าค้นหาอยู่ไหม
ก็... ไม่รู้เหมือนกัน

![[content.m4a]]