# This took 12 hours to make and nobody cares... — retained record

> Retained reddit record from reddit:comfyui.

Canonical: https://highsignal.app/data/records/9d309e0fd7d69b5a

[← sources](https://highsignal.app/data)


source: [reddit](https://highsignal.app/data/reddit) (reddit:comfyui)  published 2026-10-30 23:41 ingested 2026-10-31 02:39

[original source ↗](https://www.reddit.com/r/comfyui/comments/1xjv2d9/this_took_12_hours_to_make_and_nobody_cares/)

## retained text



```
Original 4K video without compression [Youtube](https://youtu.be/5BPbBxeT-_4)

This is pinacle of what open source models can do realism wise! Took 12hrs to make, quality takes time. I am amazed that people are not impressed by how much you can pull out of open source models, but in the end, it doesn't even matter...  


This is my workflow that is based on basic available templates in comfyui.

Workflow is basically no loras minimax h3 workflows with 3.3MP 20 steps, then DLSS5 Enhancer, Then Seedvr or Nvidia Upscaler. Input images are most important, they need to be minimum 3.3 MP or higher!

So feel free to shit on it...


```



document fetched 2026-10-31 00:44 · canonical https://reddit.com/r/comfyui/comments/1xjv2d9/this_took_12_hours_to_make_and_nobody_cares
