message_id int64 42 1.99k | datetime stringdate 2026-05-04 14:31:38 2026-07-22 07:10:01 | width int64 480 2.17k | height int64 675 2.56k | image unknown | text stringlengths 16 564 | __index_level_0__ int64 0 1.06k |
|---|---|---|---|---|---|---|
1,942 | 2026-07-20T15:09:51Z | 1,023 | 1,537 | [
255,
216,
255,
224,
0,
16,
74,
70,
73,
70,
0,
1,
1,
1,
0,
72,
0,
72,
0,
0,
255,
226,
1,
216,
73,
67,
67,
95,
80,
82,
79,
70,
73,
76,
69,
0,
1,
1,
0,
0,
1,
200,
0,
0,
0,
0,
4,
48,
0,
0,
109,
110,
116,
114,
82,... | 任务：生成一张高定时尚棚拍人像
主体：成年东亚女性，单人居中，气质自信
服装：深色高定礼服，丝缎和丝绸材质，轮廓优雅
场景：无干扰物的深 charcoal 背景，极简棚拍
光线：低调布光，一盏大软箱，脸部和礼服有柔和高光，边缘有轮廓光
镜头：DSLR、85mm、f/1.8、浅景深、时尚杂志封面质感
风格：luxury fashion editorial、photorealistic、cinematic lighting、RAW photo quality
约束：不要卡通、不要 CG 感、不要过度磨皮、不要文字、水印、杂乱道具
• 先锁背景和光线：深色无干扰背景 + 单软箱，是这条稳定出高级感的核心。
• 礼服材质要写清楚：satin... | 0 |
1,943 | 2026-07-20T15:09:51Z | 1,023 | 1,537 | "/9j/4AAQSkZJRgABAQEASABIAAD/4gHYSUNDX1BST0ZJTEUAAQEAAAHIAAAAAAQwAABtbnRyUkdCIFhZWiAH4AABAAEAAAAAAAB(...TRUNCATED) | "任务：生成一张高定时尚棚拍人像\n主体：成年东亚女性，单人居中，气质(...TRUNCATED) | 1 |
1,944 | 2026-07-20T15:09:51Z | 1,023 | 1,537 | "/9j/4AAQSkZJRgABAQEASABIAAD/4gHYSUNDX1BST0ZJTEUAAQEAAAHIAAAAAAQwAABtbnRyUkdCIFhZWiAH4AABAAEAAAAAAAB(...TRUNCATED) | "任务：生成一张高定时尚棚拍人像\n主体：成年东亚女性，单人居中，气质(...TRUNCATED) | 2 |
1,945 | 2026-07-20T15:09:51Z | 1,023 | 1,537 | "/9j/4AAQSkZJRgABAQEASABIAAD/4gHYSUNDX1BST0ZJTEUAAQEAAAHIAAAAAAQwAABtbnRyUkdCIFhZWiAH4AABAAEAAAAAAAB(...TRUNCATED) | "任务：生成一张高定时尚棚拍人像\n主体：成年东亚女性，单人居中，气质(...TRUNCATED) | 3 |
1,947 | 2026-07-21T01:04:27Z | 1,672 | 941 | "/9j/4AAQSkZJRgABAQAAAQABAAD/2wBDAAUDBAQEAwUEBAQFBQUGBwwIBwcHBw8LCwkMEQ8SEhEPERETFhwXExQaFRERGCEYGh0(...TRUNCATED) | "镜头：极限运动摄影机级别的超广角鱼眼 + 贴地仰拍 + 人物微微前倾\n反差(...TRUNCATED) | 4 |
1,948 | 2026-07-21T01:04:27Z | 1,672 | 941 | "/9j/4AAQSkZJRgABAQAAAQABAAD/2wBDAAUDBAQEAwUEBAQFBQUGBwwIBwcHBw8LCwkMEQ8SEhEPERETFhwXExQaFRERGCEYGh0(...TRUNCATED) | "镜头：极限运动摄影机级别的超广角鱼眼 + 贴地仰拍 + 人物微微前倾\n反差(...TRUNCATED) | 5 |
1,949 | 2026-07-21T01:04:27Z | 1,672 | 941 | "/9j/4AAQSkZJRgABAQAAAQABAAD/2wBDAAUDBAQEAwUEBAQFBQUGBwwIBwcHBw8LCwkMEQ8SEhEPERETFhwXExQaFRERGCEYGh0(...TRUNCATED) | "镜头：极限运动摄影机级别的超广角鱼眼 + 贴地仰拍 + 人物微微前倾\n反差(...TRUNCATED) | 6 |
1,950 | 2026-07-21T01:04:27Z | 1,672 | 941 | "/9j/4AAQSkZJRgABAQAAAQABAAD/2wBDAAUDBAQEAwUEBAQFBQUGBwwIBwcHBw8LCwkMEQ8SEhEPERETFhwXExQaFRERGCEYGh0(...TRUNCATED) | "镜头：极限运动摄影机级别的超广角鱼眼 + 贴地仰拍 + 人物微微前倾\n反差(...TRUNCATED) | 7 |
1,952 | 2026-07-21T03:04:24Z | 941 | 1,672 | "/9j/4AAQSkZJRgABAQEASABIAAD/4gHYSUNDX1BST0ZJTEUAAQEAAAHIAAAAAAQwAABtbnRyUkdCIFhZWiAH4AABAAEAAAAAAAB(...TRUNCATED) | "任务：生成一套风格统一的人像 / 写真 / 服装图提示词\n重点：先给风格分(...TRUNCATED) | 8 |
1,953 | 2026-07-21T03:04:24Z | 941 | 1,672 | "/9j/4AAQSkZJRgABAQEASABIAAD/4gHYSUNDX1BST0ZJTEUAAQEAAAHIAAAAAAQwAABtbnRyUkdCIFhZWiAH4AABAAEAAAAAAAB(...TRUNCATED) | "任务：生成一套风格统一的人像 / 写真 / 服装图提示词\n重点：先给风格分(...TRUNCATED) | 9 |
End of preview. Expand in Data Studio
Telegram AI Image Dataset — Cleaned for VLM LoRA Training
A cleaned dataset of 1,050 AI-generated images with their generation prompts, collected from a Chinese Telegram channel focused on GPT-Image-2 prompt engineering.
Each image is paired with a structured generation prompt in Chinese/English. Prompts have been cleaned of channel boilerplate — no bot instructions, hashtags, source credits, model name prefixes, emoji title lines, or channel footer ads.
Dataset Structure
| Column | Type | Description |
|---|---|---|
message_id |
int64 | Telegram message ID |
datetime |
string | Message timestamp |
width |
int64 | Image width in pixels |
height |
int64 | Image height in pixels |
image |
binary | Embedded WebP image bytes |
text |
string | Cleaned generation prompt (Chinese/English) |
Cleaning Applied
- Removed 18 noise columns — Telegram metadata, author info, file paths, entity data, etc.
- Batch grouping — Consecutive same-generation images grouped by matching dimensions; prompts propagated to all images in the batch.
- Noise removal — Stripped model name prefixes (
GPT-Image-2｜), emoji title lines, bot instructions (直接在 Bot 里输入提示词), hashtags, source credits, channel footer ads (VPN推荐,教程目录,邪修频道). - Ad filtering — Removed promotional posts, channel announcements, and non-prompt content.
- Remaining — 1,050 rows from 321 prompt groups, typically 2–4 images per prompt.
Usage
from datasets import load_dataset
ds = load_dataset("GCStream/telegram-channel-dataset", split="train")
print(ds[0]["text"]) # Clean prompt
print(ds[0]["image"]) # PIL image
Notes
- Images are embedded as binary WebP in the parquet
imagecolumn. - Prompts are primarily Chinese with English keywords, structured as field-value formats (任务, 主体, 场景, 光线, 镜头, 风格, etc.).
- Suitable for VLM fine-tuning (e.g., FLUX, SD3, DeepFloyd), prompt engineering analysis, and image-caption training.
- Downloads last month
- 80