← sources

New to ComfyUI, looking for recommendations based on my specs

source: reddit (reddit:comfyui)published 2026-11-11 12:50ingested 2026-11-12 02:42

retained text

Hey guys, I'm currently trying to create a YouTube shorts series based on whimsical characters (which I haven't really decided on yet). I've just been experimenting with different templates, settings and prompts so far. My specs are:

  
RTX 3080 12GB

i7-12700

32GB RAM

  
First I used Wan 2.2 5b with many different settings and prompts but the results were disappointing. I was unable to generate even the simplest movements as everything would become distorted and surreal. Then I switched to 14b and the results were a huge improvement. I don't expect photorealistic output, cartoony and childish stuff would work well as long as I can create a consistent and engaging story.

  
I haven't got to voiceovers yet but almost everyone suggests ElevenLabs for it, might give it a shot. My current workflow is basically image z image (text to image) --> wan 2.2 14b (image to video). If you could recommend some templates or give general tips, I would really appreciate it!

document fetched 2026-11-12 00:43 · canonical https://reddit.com/r/comfyui/comments/1xtgxpn/new_to_comfyui_looking_for_recommendations_based