41–50 of 519 posts
Seems the 768-v model, if used properly, can substantially speed-up the generation, but not exactly sure yet. Seems straightforward to switch to 512-base model for my app next week.
They apparently tried to combat NSFW generation by filtering the training dataset not to include any.
You can generate all the bloody violent gore you like, but god forbid anybody see a human body in its natural state
I am a solo dev working on a creative content creation app to leverage the latest developments in AI. Demoing even the v1 of stable diffusion to the non-technical general users blows them away completely. Now that v2 is here, it’s clear we’re not able to keep pace in developing products to take advantage of it. The general public still is blown away by autosuggest in mobile OS keyboards. Very few really know how far…
I am a solo dev working on a creative content creation app to leverage the latest developments in AI. Demoing even the v1 of stable diffusion to the non-technical general users blows them away completely. Now that v2 is here, it’s clear we’re not able to keep pace in developing products to take advantage of it. The general public still is blown away by autosuggest in mobile OS keyboards. Very few really know how far…
What are you building?
They apparently tried to combat NSFW generation by filtering the training dataset not to include any.
In practice, it's unclear how well avoiding training on NSFW images will work: the original LAION-400M dataset used for both SD versions did filter out some of the NSFW stuff, and it appears SD 2.0 filters out a bit more. The use of OpenCLIP in SD 2.0 may also prevent some leakage of NSFW textual concepts compared to OpenAI's CLIP. It will, however, definitely not affect the more-common use case of anime women with v…
At least if people have to finetune the model on that shit, then you can argue that it's not your fault because someone had to do extra steps to put stuff in there.
I am a solo dev working on a creative content creation app to leverage the latest developments in AI. Demoing even the v1 of stable diffusion to the non-technical general users blows them away completely. Now that v2 is here, it’s clear we’re not able to keep pace in developing products to take advantage of it. The general public still is blown away by autosuggest in mobile OS keyboards. Very few really know how far…
What kind of stuff does your app do that blows people away?
768x768 native models (v1.x maxed out at 512x512)
a built-in 4x upscaler: "Combined with our text-to-image models, Stable Diffusion 2.0 can now generate images with resolutions of 2048x2048–or even higher."
Depth-to-Image Diffusion Model: "infers the depth of an input image, and then generates new images using both the text and depth information." Depth-to-Image can offer all sorts of new creative applications, delivering transformations that look radically different from the original but which still preserve the coherence and depth of that image (see the demo gif if you haven't looked)
Better inpainting model
Trained with a stronger NSFW filter on training data.
For me the depth-to-image model is a huge highlight and something I wasn't expecting. The NSFW filter is a nothing (it's trivially easy to fine-tune the model on porn if you want, and porn collections are surprisingly easy to come by...).
The higher resolution features are interesting. HuggingFace has got the 1.x models working for inference in under 1G of VRAM, and if those optimizations can be preserved it opens up a bunch of interesting possibilities.
They apparently tried to combat NSFW generation by filtering the training dataset not to include any.
You can generate all the bloody violent gore you like, but god forbid anybody see a human body in its natural state
I am a solo dev working on a creative content creation app to leverage the latest developments in AI. Demoing even the v1 of stable diffusion to the non-technical general users blows them away completely. Now that v2 is here, it’s clear we’re not able to keep pace in developing products to take advantage of it. The general public still is blown away by autosuggest in mobile OS keyboards. Very few really know how far…
I don’t know anybody that is blown away by keyboard auto suggest. It’s wrong as often as it is right. Not saying it isn’t useful, but let’s not oversell it.
Let's take a deterministic algorithm that predictably corrects your typos and build it on AI. It will offer you no benefits, but it will completely destroy the utility since it will never work predictably or accurately.
They apparently tried to combat NSFW generation by filtering the training dataset not to include any.
A confrontation is inevitable, though. Right now it costs moderate sums of money to do this level of training. Not always will this be so. If I were an AI-centric organization, I would be racing to position myself as a trustworthy actor in my particular corner of the AI space so that when legislators start asking questions about the explosion of bad actors, I can engage in a little bit of regulatory capture, and have the legislators legislate whatever regulations I've already implemented, to the disadvantage of my competitors.
For people who say "people can make whatever images they like in photoshop," I will remind you of this: https://i.imgur.com/5DJrd.jpg
