WEBVTT

00:00.163 --> 00:05.223
This is ComfyUI for Animation, part
three: Background and Asset Generation.

00:06.803 --> 00:10.073
This workflow lets you input a
production frame as style reference

00:10.363 --> 00:14.383
and applies the style to either to a
background or to a foreground subject.

00:15.823 --> 00:17.803
Method one is background generation.

00:18.323 --> 00:21.193
Load a rough background sketch
and an existing production

00:21.193 --> 00:23.203
frame as the style reference.

00:23.873 --> 00:28.353
GPT Image two renders the sketch into
a fully painted background, matching

00:28.353 --> 00:31.983
the color palette, lighting, and
atmosphere of the production frame.

00:32.613 --> 00:35.413
The output is saved
alongside on comparison view.

00:37.153 --> 00:39.673
Method two is prop and subject generation.

00:40.193 --> 00:43.933
Load a prop or subject sketch
alongside the same production frame.

00:44.533 --> 00:48.753
The model renders it in the show style,
and the output automatically goes through

00:48.773 --> 00:53.943
background removal, giving you a clean
alpha masked asset ready for compositing.

00:54.603 --> 00:58.503
Optionally, you can disable the
background removal if it's not needed.

01:00.073 --> 01:03.023
For the two independent
methods, use whichever fits

01:03.023 --> 01:05.023
your task and bypass the other.

01:05.683 --> 01:08.653
Both methods use the production
frame as the style anchor.

01:08.983 --> 01:12.893
It's how the workflow keeps backgrounds
and props visually consistent with each

01:12.893 --> 01:15.103
other without any additional prompting.