Getting a measly $21 USD from patreon every month from people who want to support my art feels profoundly more life-affirming than getting a whole month's pay from a full time job...
I just got unfriended by someone on Facebook for mentioning on their Dr Seuss fake outrage post that Dr Seuss wrote 6 books that incorporated racial stereotypes that were often depicted in subservient roles to white people, including one case of a white person who wanted to put a coloured person in his Zoo.
This cancel culture is getting out of control
I got canceled in a similar way by one of my right-wing uncles because, after he was harassing my aunt(his sister) about how "abortion is murder", I posted a story about a baby that died in ICE custody because the mother was denied access to medical care. I've never really interacted with him that much anyway, so I don't view it as a loss.
Software rave
Posted: Thu Mar 11, 2021 8:39 pm
by Limey
New AI powered, machine learning software from China that recreates a person's face from fuzzy, grainy, blurred photos.
I'm just blown away by some of the results.
The software is called Remini, you can get it at the Google play store for Android. Not sure (or give a shit) if it's on Apple.
I have it running on my desktop via the Android emulator, Bluestacks 4.
It's not 100% - it can sometimes just make people look retarded, but when it works, it really works. I've seen it work magic.
I am eating crow, because for years I've ranted about "computer, enhance!" algorithms in TV and Sci-Fi and yet, here it is. We live in the future.
I've had pictures I've been restoring and colorising for years and that work was not in vain, it's been complemented by this system mapping skin texture, hair, lashes etc. I just had to get the light and colour right, which is my specialty.
I drew my avatar as a self-portrait from a photo and yet, sometimes, restoring can involve reconstructive surgery (I've drawn a few hands!) - the new software is able to add the final touches I've needed.
There's one of my grandmother that has really been brought to life.
Myheritage.com have licensed the software and may have added better animation (I think the animation is from Israel) - but that's a bit uncanny valley to me. It won't be long until we make movies on computer, using actors in their prime who are long dead, like puppets. It will look real.
It won't be long until we make movies on computer, using actors in their prime who are long dead, like puppets. It will look real.
Deep fakes:
I was impressed by how fast Deep fake went from Ted Talks to real-world application and I've been following faceswap on youtube for a while now. Their alternate Luke Skywalker for the Mandalorian was a lot less 'uncanny valley'.
That's the mapping algorithm that will be paired with the restore AI process I'm talking about, so you can rebuild these people from just a few poor quality images. It's nuts how a pixel just a few shades darker can become a mole and realistic skin, better than hollywood CGI.
So far, this remini app is the only real one doing it out there. The rest are nowehere near as well developed in their AI. It probably won't be long until Adobe try to monopolise it.
Rave 2 f5 VPN client Mac build
Posted: Thu Mar 11, 2021 10:39 pm
by Limey
Got given a Mac book at work to test. I get to pretend I still break into banks.
It did the same weird thing as iPhones used to do when you start ARP route poisoning them. It goes back to the DHCP server and goes 'one up' in IP. If you keep spoofing the default gateway to same MAC and old IP, it still somehow routes, but is now not triggering the same protections. The shitty Safari underpinned VPN client will then happily ask the end user if they want to trust a self-signed certificate (spoof of the real, signed one) and if/when the end users clicks "continue" because sure, trusting a certificate sounds like a great idea! - Then it's game over, there are your domain credentials in clear text.
Fuck your couch!
*drops the mouse*
Re: Little Raves
Posted: Fri Mar 12, 2021 1:40 am
by last_caress
Neural net generated music is pretty interesting too.
An album in the style of The Beatles, generated by OpenAI Jukebox
I also saw something about ai regeneration of missing parts of an image that was cool.
We find that, just as a large transformer model trained on language can generate coherent text, the same exact model trained on pixel sequences can generate coherent image completions and samples. By establishing a correlation between sample quality and image classification accuracy, we show that our best generative model also contains features competitive with top convolutional nets in the unsupervised setting.
Re: Little Raves
Posted: Sat Mar 13, 2021 8:49 am
by MoneyJungle
If you’re older than God and running hurts your knees, run-walk splits actually work. Jeff Galloway is da man (he also doesn’t recommend stretching).
