this post was submitted on 21 Sep 2024
12 points (100.0% liked)

[Moved to Piefed] Animation

1243 readers
1 users here now

This community has moved to [email protected]

Our community is a place for enthusiasts and fans to discuss all things related to animated media, cartoon news, etc.

Rules

1. Be RespectfulTreat everyone with respect. No harassment, bullying, or hate speech.

2. Stay on TopicGenerally ensure all posts and comments are relevant to the theme of animation. Leeway will of course be given for the natural flow of conversation

3. No SpamAvoid spamming the subreddit with repetitive messages, self-promotion, or off-topic links. Excessive posting or commenting to dominate conversations is not allowed.

4. Be Mindful of SpoilersEnsure spoilers are tagged correctly. See here.

5. No Linking Pirated MaterialJust to be safe, don't directly link to pirated content. General discussion is allowed within limits.

6. No PornSome shows/films feature nudity or erotica. These are usually allowed but must be tagged as NSFW.
Basically: Don't be an asshole

On Anime

While Anime is a beloved animation style with a rich history and extensive following, our focus here leans towards other forms of animation. This is due to the immense popularity of Anime in online spaces, which is amply covered by other dedicated communities.

This doesn't mean that Anime related conversation is not allowed here, only that it's not our focus. For a dedicated Anime instance, check out ani.social.

On Technical Discussion

While our primary goal is to discuss animated media, conversations about animation techniques and the process of creating animation are also welcome. Whether you're a budding animator seeking to understand more about the industry, or a seasoned professional sharing your experiences, you're welcome here.

Related Communities

founded 2 years ago
MODERATORS
 

Abstract

We propose the first video diffusion framework for reference-based lineart video colorization. Unlike previous works that rely solely on image generative models to colorize lineart frame by frame, our approach leverages a large-scale pretrained video diffusion model to generate colorized animation videos. This approach leads to more temporally consistent results and is better equipped to handle large motions. Firstly, we introduce Sketch-guided ControlNet which provides additional control to finetune an image-to-video diffusion model for controllable video synthesis, enabling the generation of animation videos conditioned on lineart. We then propose Reference Attention to facilitate the transfer of colors from the reference frame to other frames containing fast and expansive motions. Finally, we present a novel scheme for sequential sampling, incorporating the Overlapped Blending Module and Prev-Reference Attention, to extend the video diffusion model beyond its original fixed-length limitation for long video colorization. Both qualitative and quantitative results demonstrate that our method significantly outperforms state-of-the-art techniques in terms of frame and video quality, as well as temporal consistency. Moreover, our method is capable of generating high-quality, long temporal-consistent animation videos with large motions, which is not achievable in previous works. Our code and model are available at this https URL.

Paper: https://arxiv.org/abs/2409.12960

Project Page: https://luckyhzt.github.io/lvcd

Code: (coming soon)

Supplementary Demo clips: https://luckyhzt.github.io/lvcd/supplementary/supplementary.html

you are viewing a single comment's thread
view the rest of the comments
[โ€“] Even_Adder 1 points 9 months ago (1 children)

Let me know if this kind of post isn't allowed here. It looked alright to post per the rules.

[โ€“] [email protected] 2 points 9 months ago

It's not against the rules and is relevant to the community so you're fine.