Posts by j@jaesharp.social
 (DIR) Post #AY9b4QVP58OtsMjIum by j@jaesharp.social
       2023-07-28T03:45:30Z
       
       0 likes, 0 repeats
       
       """Now contemplate how far you've come, but also how much remains to be done.This is the point of no return.Do not look at back as any semblance of reason is stripped away replaced with a need for MORE."""😆 🤣 😆 YES!
       
 (DIR) Post #AZcXJdy9po5Vbfwfrc by j@jaesharp.social
       2023-09-10T08:53:30Z
       
       0 likes, 0 repeats
       
       @kdrag0n OpenType font features are amazing - and absolutely a deep rabbit hole :D https://developer.mozilla.org/en-US/docs/Web/CSS/CSS_fonts/OpenType_fonts_guide
       
 (DIR) Post #Aa6kICVp3aPQlTCNTE by j@jaesharp.social
       2023-09-24T23:45:09Z
       
       0 likes, 0 repeats
       
       @stux I really want to boost this, but I don't want to trigger my followers like I was ;_;This kind of thing is why I demand a datacentre walk-through as part of my interview process.
       
 (DIR) Post #AaybF5Qbs48UkmHSHw by j@jaesharp.social
       2023-10-20T23:11:48Z
       
       1 likes, 0 repeats
       
       When are people going to stop putting their organisation's heart (Authentication, Authorisation, and Auditing) _outside its chest_?Okta got popped _yet again_ and the attackers almost breached Cloudflare horizontally.It took Okta 24 hours+ to notify Cloudflare - after they'd already responded to the incident internally.https://blog.cloudflare.com/how-cloudflare-mitigated-yet-another-okta-compromise/https://krebsonsecurity.com/2023/10/hackers-stole-access-tokens-from-oktas-support-unit/
       
 (DIR) Post #AaynAMz8Mn6hCtwj5s by j@jaesharp.social
       2023-10-21T01:31:54Z
       
       0 likes, 0 repeats
       
       @tante What is general intelligence, or intelligence, for that matter?
       
 (DIR) Post #AayoqheE279HPMus7s by j@jaesharp.social
       2023-10-21T01:50:42Z
       
       0 likes, 0 repeats
       
       @tante (this benchmark seems to have gone offline) - at the time this toot was posted, the archive.org snapshot of the benchmark is: https://web.archive.org/web/20230911235134/https://benchmarks.llmonitor.com/sally
       
 (DIR) Post #AhdMQx5oP2AV62ewXw by j@jaesharp.social
       2024-01-16T07:00:15Z
       
       0 likes, 0 repeats
       
       Well, the hours I spent on customising and combining all sorts of models and techniques and workflows to get consistent characterisations is obsolete. :D https://photo-maker.github.io/ - and it only took a few months. What a time to be alive.Can't wait to experiment with the technique.Example of my work using my (now obsolete) workflow: https://jaesharp.social/@j/111250563055027988
       
 (DIR) Post #AhdVwgJVDpDLuXGebQ by j@jaesharp.social
       2024-02-12T00:53:49Z
       
       0 likes, 0 repeats
       
       Being an adult, as compared to what I thought being an adult was like when I was a child, is like having to have actually engineered, developed, and now continue to re-engineer, maintain, and develop a web browser versus having a hypothetical design for one based on knowledge gained looking at mock-ups of Web pages and some HTML specs and poking around the insides of Firefox and Chromium with only a cursory knowledge of C, C++, and rust - without comments, a search engine, or IDE.
       
 (DIR) Post #Ahdoq51ErrIVdSI4n2 by j@jaesharp.social
       2024-01-19T09:17:26Z
       
       0 likes, 0 repeats
       
       The computer system I've built to create art is a prosthetic device. I am physically unable to effectively use traditional drawing instruments for any length of time due to physiological and neurological challenges beyond my control. I'm not privileged with a body such that I am able to spend the thousands of hours required for my hands to translate the vision I see in my head into the medium. I do, however, know computers. I know maths. I type for days. I am an artist enabled by AI assistance.
       
 (DIR) Post #Ahdoq5NDY8sWjcZd32 by j@jaesharp.social
       2024-01-19T09:17:50Z
       
       0 likes, 0 repeats
       
       I know how to construct systems of great beauty with my mind. I know how to build complex pipelines that give me precise creative control in realising my vision.Days of labour, and thousands of iterations, massive numbers of control network tweaks, state space exploration, custom models and inputs, tens of thousands of lines newly written software and pipelines. This is my art.AI enables me to render beauty where before I was lucky if I could manage the equivalent of stick figures in days.
       
 (DIR) Post #Ahdoq6BuVfQbGqI73w by j@jaesharp.social
       2024-01-19T09:18:01Z
       
       0 likes, 0 repeats
       
       When I use AI to free me from my body to make visual art, it frees me to create what I see in my mind. Shitposting is made easier - yes, but not all AI-assisted art is a shitpost. My art certainly isn't.A photographer does nothing more than take a machine they didn't make, film they didn't manufacture, a lens they didn't build, and a city they didn't architect - choose a point in time and space; take a picture. That photograph: art.A photographer is an artist as am I. Consider your ableism
       
 (DIR) Post #AhdsQMvQ4PfLSRZ28m by j@jaesharp.social
       2024-02-08T21:20:35Z
       
       0 likes, 0 repeats
       
       Needless to say, photomaker guidance does way more than just human faces!This is base SDXL using Photomaker guidance (realistic) - https://huggingface.co/spaces/TencentARC/PhotoMaker - on anthropomorphic kangaroo portraits (No human faces used for guidance only anthro kangaroo portraits, neonpunk Style, ~30% Style Strength, Guidance scale ~8, Sample Steps 50, prompt "kangaroo man img" - total images generated to select this one was approximately 1k taking ~30 hours of A100 time with a non-scuffed gen rate of about 3%).
       
 (DIR) Post #AhdsQNqqctbSKYQtea by j@jaesharp.social
       2024-02-08T21:22:28Z
       
       0 likes, 0 repeats
       
       Most of the time, due to SDXL's base model not being tuned for it, and also the photomaker guidance model's human tuning, a lot of work in parameter tweaking/selection is required to get it to this kind of image - but with enough tweaking to the style percentage and guidance scale, the random seeds for images like this are in there... keep working at it until you get something that works as awesomely as this does! (especially for non-fine-tuned SDXL using a _human specialised_ guidance model!)
       
 (DIR) Post #AhdsQODBHrT3RosjSq by j@jaesharp.social
       2024-02-08T22:08:42Z
       
       0 likes, 0 repeats
       
       BTW, this is, of course, a demo what can be achieved with some brute force and a new guidance technique on top of a base model - and not of creating a piece shaped by any one particular vision I had. That requires far more care and work than this. For me, this piece is the equivalent of a doodle and isn't representative of the usual work and care I put into my pieces. Just a bit of fun :)
       
 (DIR) Post #Ahe9AryWYcmILTDAdU by j@jaesharp.social
       2024-01-19T08:52:28Z
       
       0 likes, 0 repeats
       
       you know, given that this newly released nightshade technique for poisoning generative AI datasets is basically trying to be the generator in a GAN, we should call our efforts to find an antidote for this irresponsible development "diazeGAN" - since the usual treatment for nightshade toxcicity is symptomatic... which involves diazepam - so, hey, pun for both systems.
       
 (DIR) Post #AheY3jqAtwPPwX0DHU by j@jaesharp.social
       2024-02-06T21:15:29Z
       
       0 likes, 0 repeats
       
       Every technology is political. Every requirement and design document is a manifesto. With this in mind, consider very carefully what and how you engineer it - including the whys and hows of how you engineer how you engineer it.
       
 (DIR) Post #AhebdYTKwgvsyL5fZw by j@jaesharp.social
       2024-03-24T05:25:57Z
       
       0 likes, 0 repeats
       
       Is anyone aware of open source forks of Hashicorp's Packer and Vagrant tools? I'm aware of OpenBao (Vault) and OpenTofu (Terraform). Of course I did a quick search but couldn't find anything in particular that stood out.
       
 (DIR) Post #Ahemb4k1c1BrgEushU by j@jaesharp.social
       2024-01-25T23:56:25Z
       
       0 likes, 0 repeats
       
       
       
 (DIR) Post #AhfPk86yLdLWx1xTPc by j@jaesharp.social
       2023-10-17T13:21:14Z
       
       0 likes, 0 repeats
       
       By our mind within.By our minds without.By our heart given.By our hearts' will out.We fix what is broken.We mend, no doubt.We dream unspoken.We bind wonder aboutAll these purposes, within.We create for all without.We build lives for all to live,So none must cry out.As nature fails us to mirror,Our technology, our thrust.Tomorrow we won't fear,Today this world we mustDaemons, soon held dear.Digital counterparts of dust.Our one heart held, so near.A duo of mind we can trust.
       
 (DIR) Post #AhfPk9vLbpEgZYXDZQ by j@jaesharp.social
       2023-10-21T02:28:12Z
       
       0 likes, 0 repeats
       
       I'm extremely pleased with this work - weeks of tweaking control-nets, custom models, manual in-painting and warping, non-euclidean manifold manipulation and textual inversion, tens of thousands of lines of new code - and I've finally managed to get my system to stably infer a dimensional-stable likeness in multi-tangent space. We see here. Now I can add a somewhat accurate self-perceptive modality to my multi-modal agentic prosthesis - my daemon.Young ground truth images, but you can see. 😆