TE
TechEcho
Home24h TopNewestBestAskShowJobs
GitHubTwitter
Home

TechEcho

A tech news platform built with Next.js, providing global tech news and discussions.

GitHubTwitter

Home

HomeNewestBestAskShowJobs

Resources

HackerNews APIOriginal HackerNewsNext.js

© 2025 TechEcho. All rights reserved.

PixelNN – Example-Based Image Synthesis

497 pointsby pentestercrabover 7 years ago

25 comments

otp124over 7 years ago
I used to roll my eyes at crime television shows, whenever they said &quot;Enhance&quot; for a low quality image.<p>Now it seems the possibility of that becoming realistic are increasing with a steady clip, based on this paper and other enhancement techniques I&#x27;ve seen posted here.
评论 #15328628 未加载
评论 #15329257 未加载
评论 #15328634 未加载
评论 #15330806 未加载
评论 #15328862 未加载
评论 #15332694 未加载
评论 #15328863 未加载
评论 #15333608 未加载
评论 #15329059 未加载
nlover 7 years ago
To paraphrase Google Brain&#x27;s Vincent Vanhoucke, this appears to be another example where using context prediction from neighboring values outperforms an autoencoder approach.<p>If 2017 was the year of GANs, 2018 will be the year context prediction.
mahoover 7 years ago
I hope some day this will generalize to video. I don&#x27;t care about the exact shape of background trees in an action movie - with this approach, they could be compressed to just a few bytes, regardless of resolution.
评论 #15329538 未加载
评论 #15329430 未加载
评论 #15328760 未加载
laytheaover 7 years ago
I wonder if this could be applied to &quot;incomplete&quot; 3D models and the work shifted to the GPU!?
joostersover 7 years ago
I don&#x27;t understand how the edges-to-faces can possibly work. The inputs seem to be black &amp; white, and yet the output pictures have light skin tones.<p>How can their algorithm work out the skin tone from a colourless image. Perhaps their training data only had white people in it?
评论 #15330774 未加载
评论 #15329610 未加载
评论 #15329612 未加载
imronover 7 years ago
Seems to have a thing for beards.
jokoonover 7 years ago
I have a large collection of images, many being accessible through google image search.<p>I wonder if there could be a way to &quot;index&quot; those images so I can find them back without storing the whole image, using some type of clever image histogram or hashing-kind function.<p>I wonder if that thing already exist, since there are many images, and since most images have a lot of difference in their data, could it be possible to create some kind of function that describe an image in a way that entering such histogram redirects to (or the closest) the image it indexed? I guess I&#x27;m lacking the math, but it sounds like some &quot;averaging&quot; hashing function.
评论 #15330075 未加载
评论 #15329936 未加载
评论 #15331906 未加载
ChuckMcMover 7 years ago
Is anyone in the FX business playing with this stuff? I&#x27;m thinking generational backdrops with groups of people&#x2F;stuff&#x2F;animals in them without a lot of modelling input.
XorNotover 7 years ago
So is there an analagous process that would apply to audio I wonder?
评论 #15330722 未加载
评论 #15328838 未加载
评论 #15328727 未加载
tinyrick2over 7 years ago
This is amazing. I especially like how the result can somewhat be interpreted by showing from what image the part of the generated image is copied (see Figure 5).
deevolutionover 7 years ago
Apparently you grow a beard after using their nn model?
评论 #15330443 未加载
throwaway00100over 7 years ago
No code available.
评论 #15330486 未加载
Wildgooseover 7 years ago
Very clever. I wonder if something like this could be used for other forms of sensor data as well?
评论 #15329275 未加载
verytrivialover 7 years ago
A pair of the inputs in the edge-to-edge faces are swapped. I have nagged an author.
评论 #15336726 未加载
the8472over 7 years ago
All those examples are fairly low-resolution. Does this approach scale or can it be applied in some tiled fashion? Or would the artifacts get worse for larger images?
tke248over 7 years ago
Does anything like this exist on the web would like to send a blurry license plate picture through this and see what it comes up with..
kensaiover 7 years ago
OMG, now the &quot;enhance&quot; they say in investigative TV series and movies will actually be reality! :p
nathan_f77over 7 years ago
This is cool, but in the comparison with Pix-to-Pix, it seems like Pix-to-Pix is the clear winner.
smrtinsertover 7 years ago
&quot;Enhance&quot; is real. When will this stuff trickle into lower level law enforcement?
评论 #15330668 未加载
评论 #15330666 未加载
评论 #15330756 未加载
评论 #15330662 未加载
ScoutOrgoover 7 years ago
Can we use this to identify the leprechaun and find where da gold at?
yazanatorover 7 years ago
Is there a GitHub repository link?
avianover 7 years ago
I found the title somewhat misleading. I was expecting some clever application of the nearest-neighbor interpolation. But this seems to involve neural nets and appears far from &quot;simple&quot; to me (I&#x27;m not in the image processing field though).
评论 #15330547 未加载
评论 #15329319 未加载
评论 #15328890 未加载
评论 #15331305 未加载
评论 #15329863 未加载
imaginenoreover 7 years ago
It almost looks like they mixed training and testing data in some of the examples. The bottom-left sample in the normals-to-faces is extremely suspicions.
评论 #15329833 未加载
sgtAtomover 7 years ago
Enhance.
debuggerpkover 7 years ago
Hollywood got it right!!!