Compfyui: simple workflow
Oke Dzaq, ini dia **workflow visual paling dasar di ComfyUI** buat **Text-to-Image** (generate gambar dari teks) supaya kamu ngerti alurnya dulu sebelum masuk ke yang lebih kompleks kayak video π₯
---
- π§ **Basic Workflow: Text to Image di ComfyUI**
- π Node-by-Node Flow:
``` [CheckpointLoaderSimple] β Load model (misal: sd15.safetensors)
β
[CLIPTextEncode] β Encode prompt teks kamu (misal: "anime girl in cyberpunk city")
β
[EmptyLatentImage] β Ukuran canvas (misal: 512x768)
β
[KSampler] β Proses sampling/generasi gambar
β
[VAEDecode] β Decode latent jadi gambar asli
β
[SaveImage] β Simpan hasilnya ke file ```
---
- π Diagram Alur Sederhana:
```
ββββββββββββββββββββββββββββββ β CheckpointLoaderSimple β β Load model .safetensors ββββββββββββββ¬ββββββββββββββββ β ββββββββββββββββββββββββββββββ β CLIPTextEncode β β Masukin prompt kamu ββββββββββββββ¬ββββββββββββββββ β ββββββββββββββββββββββββββββββ β EmptyLatentImage β β Ukuran gambar (W x H) ββββββββββββββ¬ββββββββββββββββ β ββββββββββββββββββββββββββββββ β KSampler β β Sampling step pakai CFG, Seed, Steps ββββββββββββββ¬ββββββββββββββββ β ββββββββββββββββββββββββββββββ β VAEDecode β β Convert latent β gambar ββββββββββββββ¬ββββββββββββββββ β ββββββββββββββββββββββββββββββ β SaveImage β β Simpan gambar ke disk ββββββββββββββββββββββββββββββ
```
---
- π Contoh Parameter Dasar:
- **Prompt**: *"a futuristic city, glowing lights, night, cyberpunk style"* - **Width x Height**: `512 x 768` - **Steps**: `20` - **CFG Scale**: `7.0` - **Sampler**: `Euler a` atau `DPM++ 2M` - **Seed**: `12345` (atau random)
---
- π Lokasi File Output:
- Default: `/ComfyUI/output/` - Nama file biasanya mengandung timestamp atau hash
---
Kalau kamu mau, aku bisa: - Kirim **file `.json` workflow dasar** ini buat langsung load di ComfyUI - Tambahin fitur kayak **Negative Prompt**, **LoRA**, atau **Upscaler** - Atau ubah jadi **Text-to-Video** alur ModelScope / Zeroscope
Mau sekalian file `.json` workflow ini aku buatin buat kamu drag langsung ke ComfyUI?