TE
科技回声
首页24小时热榜最新最佳问答展示工作
GitHubTwitter
首页

科技回声

基于 Next.js 构建的科技新闻平台,提供全球科技新闻和讨论内容。

GitHubTwitter

首页

首页最新最佳问答展示工作

资源链接

HackerNews API原版 HackerNewsNext.js

© 2025 科技回声. 版权所有。

I was messing around with img2img and I created this monstrosity

146 点作者 sanroot99超过 2 年前

11 条评论

knrz超过 2 年前
The redditor was using img2img to do style transfer frame by frame, which is why we have the jumping faces all the time (“instability”).<p>This isn’t a limitation of neural nets, as early as 2018 we’ve had stable style transfer for videos, see <a href="https:&#x2F;&#x2F;medium.com&#x2F;element-ai-research-lab&#x2F;stabilizing-neural-style-transfer-for-video-62675e203e42" rel="nofollow">https:&#x2F;&#x2F;medium.com&#x2F;element-ai-research-lab&#x2F;stabilizing-neura...</a>
mynegation超过 2 年前
In the next five years all high cost productions are going to be replaced by a neural style transfer from the child scribbles. I am only half-joking.
评论 #33104613 未加载
评论 #33104444 未加载
评论 #33104468 未加载
评论 #33106796 未加载
评论 #33106418 未加载
评论 #33104438 未加载
评论 #33106791 未加载
评论 #33107142 未加载
socialismisok超过 2 年前
Ghibli style is a real stretch. Even discounting &quot;style&quot; means more than just the colors on the frame (what shot compositions, what&#x27;s in frame, where does the camera linger, etc) the redditor didn&#x27;t really succeed in capturing even one frame that had the visual style of a Ghibli frame.<p>Praying stuff like this feels... Not crass, exactly, but like... Unrefined and ersatz in a way. It&#x27;s like buying a mall katana and displaying it like it&#x27;s something of cultural value.
评论 #33105239 未加载
dirtyid超过 2 年前
If I didn&#x27;t know context, looks almost stylistically intentional. Crazy to think a few years ago this would have taken untold tens of thousands of man hours to plan and animate, would have probably gone viral and won a bunch of awards. Now it&#x27;s just an interesting accident from fucking around.
评论 #33106482 未加载
hericium超过 2 年前
Those morphing faces gave me a disturbing flashback of looking into a mirror after LSD macrodose.
评论 #33104913 未加载
评论 #33104409 未加载
评论 #33105990 未加载
pushedx超过 2 年前
The random face style each frame is oddly captivating
评论 #33104620 未加载
mellosouls超过 2 年前
Original:<p><a href="https:&#x2F;&#x2F;youtube.com&#x2F;watch?v=biZwR6CzVzs" rel="nofollow">https:&#x2F;&#x2F;youtube.com&#x2F;watch?v=biZwR6CzVzs</a>
6510超过 2 年前
The future will need some bug fixes.
superchroma超过 2 年前
Not even close to right or stable, falls apart in the tighter shot on geralt when it no longer recognizes him as human, and AI of course has no idea about 3D context so can never do cell shading in the way it is being trained.<p>This was honestly a great waste of compute power.<p>To the downvoters, do what you want, but it doesn&#x27;t change the apparent reality of what we&#x27;re seeing. AI image generation is an exciting and growing technology but this isn&#x27;t a remarkable or successful application of it in any measurable dimension. The system lacks temporal and spatial context and that is going to be a hard blocker for this application.
评论 #33104731 未加载
评论 #33104586 未加载
davidkuennen超过 2 年前
Crazy. Just imagine what you could do with it. You could basically just do some amateur scenes in your garden and have SD make an anime masterpiece of it.
jesperlang超过 2 年前
Very dream like, trippy results! Reminds me alot of Waking Life movie