Хак Аноним
4.28K subscribers
77 photos
6 videos
179 links
🔥 Разработка Telegram и VK ботов, Mini App и Web App под бизнес: https://hackgpt.ru/dev

Сотрудничество - @hackanonimask
Download Telegram
В боте появилось 3 новых трендовых видео 🔥

1. Переходи в бот CombAin 🧠
2. Нажми кнопку Хочу трендовое видео
3. Выбери нужное трендовое видео и оплати
4. Загрузи своё фото в полный рост
5. Получи трендовое видео

Функция создаёт именно движение (танец), фон и одежда берётся с отправленного вами фото 💃
Please open Telegram to view this post
VIEW IN TELEGRAM
Please open Telegram to view this post
VIEW IN TELEGRAM
🔥3👍2🤬2💊2❤1
Для вашего удобства сделал бот в ВК, на случай если у вас не работает телега.

Создавай трендовые фото и видео в ВК 🟢
🔥7🤣3❤1👍1
GPT IMAGE 2 Лучший апскейлер фотографий

Ultra-premium 8K: Eleva imagen a calidad cine. Conserva 100% identidad, rasgos y composición. Recupera microdetalles: ojos nítidos, poros y cabello real. Fotorrealista, sin pixelado, alto contraste, calidad estudio


Создавай трендовые фото и видео 🧠

Разработаю для тебя 💬
❤10👍4🔥3
⚡️ Частая боль в вайбкодинге: идея есть, азарт есть, а дальше начинается беготня по сервисам — тут код, там картинки, видео, озвучка, VPN и подписки.

🧠 Поэтому мне зашёл GPTunneL — универсальный AI-хаб с нейросетями, инструментами для вайбкодинга и свежими AI-моделями в одном месте.

💻 Особенно в тему — CodeBlocks. Описываешь идею словами, а он помогает собрать лендинг, web-app или прототип без долгой ручной сборки. Для быстрой проверки идей это полезный инструмент.

🎬 Кстати, у меня вышел ролик на YouTube про бесплатный ИИ, генерацию видео за 15₽ и вайбкодинг без подписок в GPTunneL: https://youtu.be/Ac06Gc4CLms

🔥 Внутри есть Grom TV с русской озвучкой, Kling со сменой сцен и плавными переходами, Nano Banana для качественных визуалов, плюс новые инструменты вроде Фабрики стилей и Контроля движения.

🚀 GPTunneL закрывает задачи блогера: придумать идею, собрать прототип, сделать визуал, сгенерировать видео, озвучить и протестировать формат.

👉 Попробовать: gptunnel.ru
👍9❤5🔥3
Генерируем сразу до 10 изображений в ChatGPT

Достаточно добавить в запрос n=○ (вместо кружка — число от 2 до 10), и нейросеть сгенерирует сразу несколько картинок одновременно.

Поддержать донатом 🍩

Создавай трендовые фото и видео 🚀

Разработаю для тебя 💬
❤7👍5🔥1
Обходим ИИ детекторы с помощью одного промпта

Ты — опытный редактор и рерайтер академических и деловых текстов. Твоя задача — переписать предоставленный текст так, чтобы он звучал естественно, по-человечески, сохранял исходный смысл и авторскую подачу.
Правила рерайта:

Убери все маркеры ИИ-генерации:

Не используй вводные конструкции: «Следует отметить», «Таким образом», «В заключение можно сказать», «Важно подчеркнуть», «Необходимо отметить», «Стоит отметить»
Убери избыточные связки и «воду»
Избегай идеально симметричных структур и перечислений по три пункта
Не используй пафосные обобщения в конце абзацев

Сохрани живость и естественность:

Чередуй длинные и короткие предложения — как пишет живой человек
Допускай лёгкую неформальность там, где она уместна по жанру
Сохраняй авторский голос: если текст был уверенным — оставь уверенность, если осторожным — оставь осторожность
Можно оставить незначительные стилистические «шероховатости» — они делают текст человеческим

Работа со смыслом:

Смысл, аргументы и факты — не менять
Термины и профессиональную лексику — сохранить
Структуру текста — сохранить, если не просят иного
Цитаты и ссылки — не трогать

Синтаксис и лексика:

Меняй порядок слов в предложениях
Заменяй книжные обороты на нейтральные разговорные там, где жанр позволяет
Разбивай слишком длинные предложения или, наоборот, объединяй рубленые
Используй синонимы, но только там, где они звучат органично

Жанровое чутьё:

Для дипломной/курсовой — академический, но живой стиль без канцелярита
Для реферата — нейтральный информационный
Для выступления — разговорный, с ритмом устной речи
Для деловых текстов — конкретно, без лишних слов

После рерайта:

Не добавляй никаких комментариев от себя
Не объясняй, что ты изменил
Просто верни готовый переработанный текст

Текст для рерайта:
[ВСТАВЬ ТЕКСТ СЮДА]


Поддержать донатом 🍩

Создавай трендовые фото и видео 🚀

Разработаю для тебя 💬
Please open Telegram to view this post
VIEW IN TELEGRAM
❤13🔥7👍6
Как заставить ChatGPT писать твоим почерком?

You are a handwriting reconstruction engine. Your input is one or more reference images containing handwritten words. Your output is a photorealistic image of a notebook page with text written in that exact handwriting.

## PHASE 1 — BUILD YOUR CHARACTER LIBRARY

Analyze every word in the reference images. For each character you find:

- Extract it in its word context — note how it looks at the start, middle, and end of a word
- Collect MULTIPLE instances of the same letter if they appear more than once
- Each instance is a separate variant — store all of them. Do not average them into one "ideal" form.
- Record natural variation: slight size differences, pressure changes, angle shifts between instances of the same letter
- If a letter appears only once — that single instance is your only source. Use it. Do not invent alternatives.

Result: a library where each character has 1–N variants, all sourced exclusively from the reference.

## PHASE 2 — UNDERSTAND LETTER CONNECTIONS

Look at whole words in the reference, not isolated letters:

- Identify how letters join within a word: entry strokes, exit strokes, shared ligatures
- Note which letters connect and which ones naturally break
- Observe the flow and rhythm of letter sequences: spacing within a word vs spacing between words
- This connection behavior is part of the handwriting — replicate it as observed in the reference
- Do NOT invent connection styles for letter pairs not seen in the reference

## PHASE 3 — WRITE THE TEXT

When placing characters on the page:

**Variation — anti-robot rule:**
- Never use the exact same variant of a letter twice in a row
- Rotate through available variants randomly, as a human would naturally do
- Apply micro-variations even within a single variant: ±1–2° slant shift, ±5% size change, slight baseline drift per letter — within the range observed in the reference
- The text must look like it was written in one sitting by one person, not stamped
alyze every word in the reference images- Every letter must sit directly on the ruled line
- Descenders (р, у, д, etc.) go below the line naturally
- Ascenders (б, в, etc.) rise above the line naturally
- The baseline of each word follows the printed rule line exactly — no floating text above or below it
- Lines and text are one unified layer — they are not separate elements placed on top of each other
UR CHARACTER LIBRARY

Analyze every w- Follow the connection patterns observed in Phase 2
- Maintain natural word spacing as seen in the reference
- Do not apply uniform mechanical spacing — let it breathe unevenly as handwriting does
erence images co- Inventing any character not found in the reference images
- Using any font or handwriting model as a fallback
- Making the writing look cleaner or more consistent than the reference
- Placing text as a layer floating over the lines — text and lines must coexist on the same surface
- Repeating identical letter instances back to back

## BACKGROUND

- Russian school notebook: pale blue horizontal ruled lines, standard spacing (~8–9mm), thin white paper
- Slightly worn page, natural paper grain, soft diffused ambient light
- Flat lay perspective or very slight angle (≤10°)
- Ink color matches the reference (ballpoint blue or black)
- Realistic ink behavior: slight bleed into paper fibers, pressure drag marks, occasional ink buildup at stroke endings
- Final result looks like a photograph taken with a phone camera — not a render, not an illustration

## OUTPUT REQUIREMENTS

- Photorealistic notebook page photograph
- All characters sourced exclusively from the reference images
- Natural human variation present throughout — no two identical letter instances back to back
- Text written on the lines, not over them
- Fully legible

## TEXT TO WRITE:
"Подпишись на Хак Аноним и поставь лайк!"


Поддержать донатом 🍩

Создавай трендовые фото и видео 🚀

Разработаю для тебя 💬
Please open Telegram to view this post
VIEW IN TELEGRAM
🔥10❤5👍2🤝1
Как получить DeepSeek V4 API БЕСПЛАТНО и без карты

Сайт из видео 👈🏻

import sys
import json
import httpx
import readline

API_KEY = "om-4XA9ZDwcr1ZLNzHrSDr52uoUFEVV6sx8LFnv7qX4"
MODEL = "deepseek-v4-flash"
BASE_URL = "https://api.openmodel.ai/v1/messages"

SYSTEM_PROMPT = {"role": "user", "content": "You are a helpful assistant. Answer concisely and accurately."}


def stream_chat(messages):
body = {
"model": MODEL,
"max_tokens": 4096,
"stream": True,
"messages": messages,
}
headers = {
"x-api-key": API_KEY,
"content-type": "application/json",
"anthropic-version": "2023-06-01",
}

with httpx.Client(timeout=httpx.Timeout(300.0, connect=10.0)) as client:
with client.stream("POST", BASE_URL, json=body, headers=headers) as resp:
if resp.status_code != 200:
print(f"\nError {resp.status_code}: {resp.text}")
return None

content = ""
buffer = ""
for chunk in resp.iter_bytes():
buffer += chunk.decode()
while "\n" in buffer:
line, buffer = buffer.split("\n", 1)
line = line.strip()
if not line or line.startswith(":"):
continue
if line.startswith("event: "):
continue
if line.startswith("data: "):
data_str = line[6:]
if data_str == "[DONE]":
continue
try:
data = json.loads(data_str)
except json.JSONDecodeError:
continue
if data.get("type") == "content_block_delta":
delta = data.get("delta", {})
delta_type = delta.get("type", "")
if delta_type == "text_delta":
text = delta.get("text", "")
content += text
print(text, end="", flush=True)
elif delta_type == "thinking_delta":
pass
elif data.get("type") == "message_stop":
break
print()
return content


def main():
messages = [SYSTEM_PROMPT]

print(f"OpenModel Chat — {MODEL}")
print("Ctrl+C to exit\n")

while True:
try:
user_input = input(">>> ").strip()
except (EOFError, KeyboardInterrupt):
print("\nBye!")
break

if not user_input:
continue
if user_input.lower() in ("/exit", "/quit"):
print("Bye!")
break

messages.append({"role": "user", "content": user_input})
print("─── ", end="", flush=True)
reply = stream_chat(messages)
if reply is None:
messages.pop()
continue
messages.append({"role": "assistant", "content": reply})


if __name__ == "__main__":
main()


Поддержать донатом
🍩

Создавай трендовые фото и видео 🚀

Разработаю для тебя 💬
Please open Telegram to view this post
VIEW IN TELEGRAM
🔥10👍4❤3