Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Diffusion Bee: Stable Diffusion GUI App for M1 Mac

github.com

Post reply on HN
Print view View original

131–140 of 199 posts

Re: Diffusion Bee: Stable Diffusion GUI App for M1 Mac

#131

Post by OkGoDoIt »

fibonacc wrote: post #17
thrill wrote: post #11

Earlier quoted context omitted.

It works, well, incredibly.

how is this possible without a dedicated gPU? thought stable diffusion would require far more horsepower

Horsepower can make it go faster, but the major limitation is graphics memory. Graphics cards with under 12 GB of memory can’t handle the model (although I believe there are lower-memory optimizations out there), which means you need a pretty high-end dedicated graphics card on a PC. But because Apple Silicon chips have reasonably fast on-chip graphics with integrated memory, it can run pretty efficiently as long as your Mac has 16gb or more of RAM.

Re: Diffusion Bee: Stable Diffusion GUI App for M1 Mac

#132

Post by vermilingua »

hsn915 wrote: post #82

This requires macOS 12.5 but it doesn't seem like it's released? At least for my region? System preferences "Software updates" tab says I'm on the latest (12.4) and there are no updates for me to install. How am I supposed to try this?

I had the same issue (12.2.1), something seems to be borked with software update. Resolved by booting into safe mode, update showed up as expected.

Re: Diffusion Bee: Stable Diffusion GUI App for M1 Mac

#133

Post by paulgb »

gpm wrote: post #27
simondotau wrote: post #21

Earlier quoted context omitted.

And the real questions come out. Seriously though, I imagine this is less a case of whether this specific implementation permits pornography, but whether any porn was included in the dataset it was trained on. No matter how good AI is, it only knows what it knows.

"Stable Diffusion" is referring to one specific set of models, which were trained on a dataset including some porn. The official implementation has a second model that detects pornography, and replaces outputs including it with a picture of this dude https://www.youtube.com/watch?v=dQw4w9WgXcQ (Not kidding). Removing that is a really simple a one line change in the official script.

> Not kidding

I was amused by this when reading the source. Here’s the function that loads the replacement.

https://github.com/CompVis/stable-diffusion/blob/69ae4b35e0a...

It looks like removing line 309 of the same file would disable the check, but I haven’t tried it.

Re: Diffusion Bee: Stable Diffusion GUI App for M1 Mac

#134

Post by artdigital »

paulgb wrote: post #133
gpm wrote: post #27

Earlier quoted context omitted.

"Stable Diffusion" is referring to one specific set of models, which were trained on a dataset including some porn. The official implementation has a second model that detects pornography, and replaces outputs including it with a picture of this dude https://www.youtube.com/watch?v=dQw4w9WgXcQ (Not kidding). Removing that is a really simple a one line change in the official script.

> Not kidding I was amused by this when reading the source. Here’s the function that loads the replacement. https://github.com/CompVis/stable-diffusion/blob/69ae4b35e0a... It looks like removing line 309 of the same file would disable the check, but I haven’t tried it.

This is the line to change to disable the nsfw filter: https://github.com/CompVis/stable-diffusion/blob/69ae4b35e0a...

Re: Diffusion Bee: Stable Diffusion GUI App for M1 Mac

#135

Post by NonNefarious »

aivoyeur wrote: post #71
ihateolives wrote: post #68

Earlier quoted context omitted.

> This really is a one-click installer, which is excellent. As far as I know it's the first and only one for M1. There's https://www.charl-e.com/

That's a very pretty landing page, but it entirely avoids describing any features, or showing what it looks like beyond a thumbnail-sized image. Do you have any insights?

I've tried it. The first time it did generate an image, albeit missing one of the few elements I mentioned.

Now it only generates black frames, even after being restarted. Not impressed at this point.

The options are sizes up to 768 x 768, "steps," and "guidance scale." You can do text-to-image or image-to-image.

Re: Diffusion Bee: Stable Diffusion GUI App for M1 Mac

#137

Post by barrkel »

josephg wrote: post #34
metadat wrote: post #24

Earlier quoted context omitted.

Based on the upvotes, inquiring minds _really want to know_!

It can make porn images, but I haven't been able to get consistently good results. There's probably ways to tweak the parameters to improve things, but I haven't figured that out. It mostly understands what naked people look like, but the images I've generated involve a lot of accidental body horror. You get a lot of people with extra arms, weird eyes, or body parts in the wrong places. The fact that they explicitly…

In particular it is very confused about genitalia, it tends towards hermaphrodite.

Re: Diffusion Bee: Stable Diffusion GUI App for M1 Mac

#138

Post by barrkel »

TillE wrote: post #59
XorNot wrote: post #38

Earlier quoted context omitted.

I mean I don't know why we have to talk around this point: yes, you can definitely generate pornography with SD. Be prepared to see a lot of weird things, because that super-realism really means it just drives right into the uncanny/body horror valley at full speed. But the second point here is also wrong: the whole reason these models are interesting is because they can generate things they haven't seen before - the…

> But the second point here is also wrong: the whole reason these models are interesting is because they can generate things they haven't seen before No, it's essentially generating mashups of its training data, which can be very interesting. So a model that hasn't been trained on a lot of porn will of course do a very bad job at generating porn.

Mashups is the wrong way to think about it. It's generalizing at a higher level than texture / image sampling and it can tween things in latent space to get to visual spaces that haven't been explored by human artists before.

It requires a good steer and prompting is a clumsy tool for fine tuning - it's adequate for initialization but we lack words for every shade of meaning, and phrase weighting is pretty clumsy too, because words have a blend of meaning.

Re: Diffusion Bee: Stable Diffusion GUI App for M1 Mac

#140

Post by drcongo »

andreasley wrote: post #26
indigodaddy wrote: post #19

How long is the generation taking on say a 2021 MacBook M1 Pro?

55 seconds on a M1 Pro MacBook with 16GB RAM to generate a picture (with default settings, so 512x512, 25 steps). On a M1 Ultra, it takes 12 seconds with 64GB RAM (same settings). While computing, the Mac Studio is pulling 100 watts of power.

M1 Pro MBP with 32GB RAM here, takes 30 seconds on the default settings.
Post reply on HN