aenvoker

6 ms·

31.
by aenvoker 3y ago
Completely correct. They did a lot of testing of a variety of user interfaces before choosing Discord. With a web interface, new users would come in a prompt “dog”, “funny dog”, “two funny dogs”, get bored and leave. But, when a bunch of ne
32.
by aenvoker 3y ago
Stability, Midjourney, Dall-e, etc don’t want their products to be used for misinformation. They really do want AI generated images to be trivially identifiable. There are papers around describing how to bake a watermark into a model. Don’t
33.
by aenvoker 3y ago
They combine multiple custom algorithms that are designed to run on high-RAM (>24GB) GPUs. And, they incorporate a lot of user-provided feedback into the training of new models. If you have more time on your hands than cash, you can get
34.
by aenvoker 3y ago
Camera wonking is great, but if you don’t like any AI art, that suggests you are rejecting the concept out of hand. Being so low-barrier, there is naturally a flood of images from beginners who are just futzing around. But, I hear a commo
35.
by aenvoker 3y ago
FYI: MJ is in the process of introducing a smarter word filter that recognizes the difference between 'Wang Jian' vs. 'suck my fat wang'
36.
by aenvoker 3y ago
There is also an issue in many Standard Diffusion model files that makes them quite chaotic with short prompts. With some of those, I've found I get much better results by simply copy-paste-paste-paste-paste-paste-pasting a short posi
37.
by aenvoker 3y ago
I handle most of the moderation of violence and gore. That painting would get a user a message about the violence and gore guidelines, but that’s all. The guidelines are all about avoiding visceral reaction from the viewer. Not visceral li
38.
by aenvoker 3y ago
SD can be better than MJ if you dig really deep into the open source nature of SD. Open source UI and plugins, community-made models of many types and uses, tribal knowledge of techniques... It’s very complicated and requires a lot of searc
39.
by aenvoker 3y ago
> It's an archivists nightmare. I’m guessing you are referring to archive.org style public archiving. But, I’ll mention anyway that for your own generations there is a web interface to search and bulk download your images. And, a we
40.
by aenvoker 4y ago
Hi. I’m a volunteer Midjourney content moderator. They don’t pay me. I’m under no obligation to make them look good. The team behind MJ are a bunch of overly-nice people. The work very hard night and day to help people bring art into the wo
41.
by aenvoker 4y ago
David believes this is culturally sensitive to the Chinese users because for them, parody of politicians is taboo. Conspiracy theories abound. But, that’s all there is to it. David’s already made all the money he wants. From here he’s just
42.
by aenvoker 4y ago
Midjourney isn’t trying to be a commercial success. They just want to bring creative power to as many people as possible. Keep the servers scaling to match the rate of userbase growth is the primary financial concern. Meanwhile, the amount
43.
by aenvoker 4y ago
In this case, that’s not how it went down. There was a lot of hand-wringing about “We want to give users as much freedom as possible, but…” the team are a bunch of overly-nice people who don’t want to work on certain things. And, we want an
44.
by aenvoker 4y ago
David Holz has said they don’t want to be in the API business. Their goal is to bring creative power to individuals. Squeezing margin out of API calls is more about negotiating with corporations.
45.
by aenvoker 4y ago
This is the reason. Before Midjourney, www.eleuther.ai’s Discord had an image generation channel. There the benefits of generating socially were made obvious. People help each other, learn from each other, riff off each other. It accelerat
46.
by aenvoker 4y ago
Well, there’s also the question of what kinds of acts do you want to enable and participate in. As a content moderator at Midjourney, I get to think about this a lot :) People are free to do whatever they want on their own machines. But, t
47.
by aenvoker 4y ago
With either tech, this thread is a very cynical take. If you give generic prompts, you get generic results. Want something personal, get personal and you will get personal results out of the AI. The barrier to creation has been lowered. But
48.
by aenvoker 4y ago
Yep. It was literally a PS1 inside the PS2. It was used for PS1 back-compatibility in the early models. It had its own memory space, and exclusive access to all of the non-GPU I/O. CDROM, audio, controllers, USB, memory cards. You coul
49.
by aenvoker 4y ago
Again, naming Rutkowsi invokes an impression of his style. But, copies none of his paintings. Read the paper. What I found is that a random sampling of the database naturally found a small subset of images that are highly duplicated in th
50.
by aenvoker 4y ago
The training process looks at pixels. Gets an impression of the relationships between words and curves in images. But, to say it “remembers pixels” is pretty loaded language that implies copying pixels into the model file. I’d like to see e
51.
by aenvoker 4y ago
There was an article recently about how AAA games are requiring so much content lately that any development cycles starting today would need to target the PlayStation 6. I think there’s still some room for improvements in productivity.
52.
by aenvoker 4y ago
What you are talking about is Overfitting. It only happens with images that appear way too many times in too many forms in the training set. Usually with the most iconic images of all time, such as the Mona Lisa. And, naturally, hyper-iconi
53.
by aenvoker 4y ago
AI has reached a point where aesthetics and technique are no longer hurdles. You can literally prompt gibberish and get back a highly aesthetic image that is completely boring and irrelevant. So, the challenge is still to dig deep and make
54.
by aenvoker 4y ago
Making a cyclops was a long-running challenge among the users of Midjourney. Turns out multiprompting “photo of a girl:: big eye” so that the AI would attempt to make both concepts individually and simultaneously in the same image would do