DrSiemer

18 ms·

211.
by DrSiemer 3y ago
What is the relationship between (I assume you mean generative) AI, which is the foundation for many useful new technologies, and NFT/crypto, which usually amounts to nothing more than a pyramid scheme style investment opportunity?
212.
by DrSiemer 3y ago
You just KNOW when an image is from SD No, you know when a beginner generated an image in Stable Diffusion. With enough skill and attention, you will not. Sure, there is a learning curve and it takes more time to get to a good result. But
213.
by DrSiemer 3y ago
Runs just fine on my 8GB card. You probably did not swap the VAE. Without that it will only produce junk.
214.
by DrSiemer 3y ago
Original link is blocked for me: https://web.archive.org/web/20230924005317/https://www.villa... Related: https://youtu.be/2Zqq7nv2FYU?si=Fcx9ryn1YsXlWQtT
215.
by DrSiemer 3y ago
That's probably because they use "clean" output from their own base models. What you get from those is a far cry from what a custom model like DreamShaper together with negative embeddings can do.
216.
by DrSiemer 3y ago
Don't remember what it was exactly, but I did run into some obscure scenario (something with tab focus I believe) where Edge had a bug that prevented my code from working. Just spin up a VM for testing Edge, the differences are small.
217.
by DrSiemer 3y ago
Midjourney is a limited set of models and (to me at least) much easier to spot than Stable Diffusion output, with it's giant array of custom models, LoRA's, textual embeddings and tricks like HiRes fix and upscale script.
218.
by DrSiemer 3y ago
The control that offers is extremely limited versus SD in A1111 with all it's different models, LoRA's, embeddings, extensions and ControlNet types.
219.
by DrSiemer 3y ago
Isn't it one of the more likely answers to the Fermi paradox? "Transcending" could also be "everyone got really into VR and stopped caring about actual reproduction".
220.
by DrSiemer 3y ago
Some pretty clever friends of mine are firmly on the "fake news" fence with regards to climate change (and some other similar issues). I've given up on changing their minds. No matter how deep I analyze their data and point o
221.
by DrSiemer 3y ago
For that to happen this flickering animation style has to become popular first. I don't think it will be. It can only be applied to very niche situations. It will never be a mainstream thing like old cameras were.
222.
by DrSiemer 3y ago
Not going to bother with this until the temporal cohesion issue is solved. The results are cool, but the variety in frames makes it look like a very specific and distracting art style instead of true animation.
223.
by DrSiemer 4y ago
Not to mention the video style itself, which appears to have been portalled in straight from the 90s. In most examples the Imagen examples are actually closer to what I'd expect to see as well. Less detailed, yes, but a better and more
224.
by DrSiemer 4y ago
Just rename it in the tokenconfig.json
225.
by DrSiemer 4y ago
While watching Kubo and the Two Strings I thought it was a bit odd how persistently they tried to make this 3d movie look like stop motion. I was blown away when most of it turned out to be actual stop motion. There is some 3d though. Perso
226.
by DrSiemer 4y ago
Set & setting?
227.
by DrSiemer 4y ago
The people I personally know who got over depressions and processed traumatic experiences from their childhood would disagree.
228.
by DrSiemer 4y ago
A counterargument to this would be that llm's are also an incredible learning tool. If you don't understand something, just ask. I envy the speed with which a younger version of myself will be able to pick things up.
229.
by DrSiemer 4y ago
Exactly this. To finally be able to unlock the fridge with countless projects abandoned for lack of time and field knowledge!
230.
by DrSiemer 4y ago
So which is better, running 7B without quantization or running 13B with? They both require about the same amount of vRAM (10gb).
231.
by DrSiemer 4y ago
Not obvious. A simple "mostly" would have fixed the statement. Speaking as someone who still occasionally needs a print-out and will not ever see that need go away unless some kind of amazing e-ink technology becomes the norm.
232.
by DrSiemer 4y ago
This reminds me of when I made a parody "game" once, where all you had to do was click to get a point. I had included a message at 10k, for a friend that I had expected to check the js source, but to my surprise a non-programmer f
233.
by DrSiemer 4y ago
ZZT was created by Tim Sweeney, CEO and founder of Epic Games. It was a huge contribution to my deeper understanding of programming at the time.
234.
by DrSiemer 4y ago
Obviously those are not questions that you can trust an AI language model with. You have social media or real life connections to discuss that. In theory it is possible that sponsored content will creep in, but that does not invalidate the
235.
by DrSiemer 4y ago
Often when looking for an answer I just need to confirm what I already know, or look up something I forgot. Google offers me a bunch of sponsored links first and then sites with the best SEO, which will smother me with all kinds of unrelate
236.
by DrSiemer 4y ago
No, it is an accurate evaluation of what ChatGPT does. The model is not trying to explain something to you, it is trying to convince itself that what it is writing looks like it could have been written by a human. There is no logical thinki
237.
by DrSiemer 4y ago
The biggest advantage is not in finding the answer. It's in being able to ask questions about that answer and getting immediate and pretty accurate responses.
238.
by DrSiemer 4y ago
Please feel free to voluntarily limit turmoil on the work floor today
239.
by DrSiemer 4y ago
Referring to Twitter not paying rent for some of their office space I assume, which was allegedly part of a renegotiation tactic.
240.
by DrSiemer 4y ago
All articles that do not answer the question they pose should get a disclaimer like that. Or maybe a "Crowd sourced TLDR" browser addon could work.

More ›