- Choose a stack for image, voice, and video without getting lost among models.
- Distinguish open models, commercial use, and cloud services.
- Keep a record of which model, license, and parameters you used.
Aulafy base stack
- ComfyUI: visual node-based interface for images, video, audio, and reproducible workflows.
- Diffusers: Python code to generate, compare, and automate images with versionable pipelines.
- FLUX: powerful family for images; review each variant because they don't all share the same license.
- Whisper: local transcription and subtitles.
- Piper: local text-to-speech with lightweight models.
- Wan: video generation for short clips, usually with higher VRAM requirements.
Record sheet you should keep for each resource
recurso: tipo: modelo | lora | voz | workflow | dataset nombre: version_o_commit: fuente: licencia: uso_permitido: personal | comercial | investigacion | revisar restricciones: parametros_clave: fecha_revision: 2026-07-02
Quick decision
- I want an image now: ComfyUI with a simple workflow.
- I want to repeat a hundred variations: Diffusers with seed, prompt, and manifest.
- I want voice for a lesson: Whisper to transcribe and Piper to narrate.
- I want video: start with 3 to 5 second clips and a single scene.
- I want to sell it: check the license before generating, not after.