Compfyui: GPU untuk text to video

Mantap Dzaq, kamu mau setup **ComfyUI di Ubuntu 24.04 dengan GPU untuk Text-to-Video generation**. Ini keren banget, karena bisa hasilin video dari teks pakai model kayak **ModelScope T2V** atau **Zeroscope**, langsung di ComfyUI!

---

1. 1. 1. 🔧 **Langkah Lengkap Instalasi ComfyUI + GPU + Text-to-Video di Ubuntu 24.04**

> 💡 Asumsi kamu pakai GPU NVIDIA dengan driver + CUDA udah terinstall. Kalau belum, bisa aku bantu juga.

---

1. 1. ✅ 1. Install Dependensi Dasar

```bash sudo apt update && sudo apt upgrade -y sudo apt install git python3.10 python3.10-venv python3.10-dev build-essential ffmpeg -y ```

> 🧠 **Catatan:** Python harus versi 3.10.x karena ComfyUI belum support 3.12+

---

1. 1. ✅ 2. Clone ComfyUI

```bash git clone https://github.com/comfyanonymous/ComfyUI.git cd ComfyUI ```

---

1. 1. ✅ 3. Buat dan Aktifkan Virtual Env

```bash python3.10 -m venv venv source venv/bin/activate ```

---

1. 1. ✅ 4. Install Dependensi Python

```bash pip install --upgrade pip pip install -r requirements.txt pip install xformers ```

> 🔧 Kalau kamu pakai GPU NVIDIA Ampere/RTX, `xformers` sangat direkomendasikan.

---

1. 1. ✅ 5. Download Model Stable Diffusion + T2V

1. 1. 1. 📦 **Stable Diffusion (untuk frame generation)**

Taruh di: ``` ComfyUI/models/checkpoints/ ``` Contoh: ```bash mkdir -p models/checkpoints wget -O models/checkpoints/v1-5.safetensors https://huggingface.co/runwayml/stable-diffusion-v1-5/resolve/main/v1-5-pruned.safetensors ```

---

1. 1. 1. 📦 **ModelScope Text-to-Video**

Taruh di: ``` ComfyUI/models/text2video/ ```

Contoh: ```bash mkdir -p models/text2video wget -O models/text2video/modelscope_t2v.safetensors https://huggingface.co/damo-vilab/modelscope-text-to-video-synthesis/resolve/main/pytorch_model.bin ```

---

1. 1. ✅ 6. Tambah Node T2V (ModelScope / Zeroscope)

Clone node custom untuk text-to-video:

```bash cd custom_nodes git clone https://github.com/Kosinkadink/ComfyUI-VideoHelperSuite.git git clone https://github.com/Kosinkadink/ComfyUI-ModelScopeT2V-node.git cd .. ```

---

1. 1. ✅ 7. Jalankan ComfyUI

```bash python main.py ```

Buka di browser: `http://127.0.0.1:8188`

---

1. 1. ✅ 8. Load Workflow Text-to-Video

1. Di ComfyUI, klik tombol `Load` 2. Pilih workflow JSON khusus T2V (aku bisa bantuin buatin kalau belum ada) 3. Isi prompt, klik `Queue Prompt`

---

1. 1. 1. ⚠️ Syarat Penting:

- RAM minimal 16 GB (lebih kalau bisa) - VRAM GPU minimal 8 GB (lebih disarankan) - Pastikan `ffmpeg` sudah terinstall (untuk gabung frame jadi video)

---

Kalau kamu mau: - 🔁 Workflow T2V siap pakai (anime style, realistik, dll) - 🚀 Optimasi kecepatan GPU - 📹 Tambah control frame atau interleave AI video

Tinggal bilang aja Gipy bantuin semua!

Mau aku langsung buatin **workflow JSON untuk Text to Video pakai ModelScope** juga?