STABLE DIFFUSION 3 NEW LICENSING!
ฝัง
- เผยแพร่เมื่อ 4 ก.ค. 2024
- Stability AI updates their licensing terms for SD3. Better? Yes.
Prompt styles for Stable diffusion Automatic1111, Forge, ComfyUI & Vlad/SD.Next: / sebs-hilis-79649068
Get early access to videos and help me, support me on Patreon / sebastiankamph
Chat with me in our community discord: / discord
Stable Diffusion for Beginners Playlist • Stable Diffusion Begin...
My Weekly AI Art Challenges • Let's AI Paint - Weekl...
My Stable diffusion workflow to Perfect Images • Revealing my Workflow ...
ControlNet tutorial and install guide • NEW ControlNet for Sta...
Famous Scenes Remade by ControlNet AI • Famous Scenes Remade b...
![[LIVE] : ONE ลุมพินี 71 วันนี้!! คู่เอก "ทรงชัยน้อย vs รักษ์"](http://i.ytimg.com/vi/_AGPnq2p6Js/mqdefault.jpg)
"We acknowledge that our latest release, Stable Diffusion 3 Medium, didn’t meet our community’s high expectations,".
So being able to render a person laying on grass without looking like some mutated accident is a high expectation now?
" and recognize that the commercial license originally associated with SD3 caused some confusion and concern. "
Yes, it must have been the community that were confused and nothing to do with the fact that the license was abysmal.
Overall a typical insulting corporate response blaming the community.
It's crazy how allergic these companies are to a simple, honest "we messed up." It should be the norm, but it's so rare than any company willing to genuinely admit fault instantly gets goodwill for it, but the vast majority still won't do it.
I mostly agree with you, but the important thing is they're trying to correct it and improve now. So let's give them a chance to.
Stable Diffusion, Adobe, Figma…
Their responses are so corporate, it's like the community is talking to Adobe. Their discord is worse but in a different way. No official responses on anything at all, just ignoring any questions from the community.
Getting kicked off CivitAI must have weighed heavily in their walking that license back. All the free finetuning they get from the community would have evaporated, and they need it. They needed it for SDXL, they need it for SD3, the base models are not good enough as-is to win people over. So they surrendered. Nice!
Yeah that must be it. They also lost a big client since civitai is used to run their models so they must have some kind of contract with civitai so they must have lost some revenue from it
I wouldn't say it was the only reason, the negative press and other factors would play a part but yeah, I'd definitely agree the loss of CivitAI and effectively the community with it was a major factor.
Whether it's too late or not is another question.
Hopefully this mean Civitai will restart to allow SD3 Based Models Soon.
The SD3 2B is still being worked on by the community, even if the old license took out a lot of wind compared to SD1.5 and SDXL.
But I honestly hope the next release is the beginning of a redemption Arc for both Stability and SD3.
SD3 has the most potential so far as long as we're allowed to go nuts with it.
a lot of people will "nuts" on the fine tune model.... -__-
How long you are in this bubble of SD? Many words I don’t understand and I wonder how much time it would take to speak your language ;)
@@Vanderful_life ohooho if you look at the civitai model and turn off the nsfw filtrer you Will know what I am talking about...
I think it's better if we avoid Stability AI and check out some other dope alternatives that are also pretty solid. I'm currently messing around with stuff like the new Pixart model, and it's blown my mind more than once already.
Of course, the massive tools available for SD aren't there yet, but we'll see.
The open-source alternative from the community will also be interesting if they can pull it off.
I don't think Stability AI is really learning from their mistakes. The new CEO, Prem Akkaraju, isn't exactly known for being super open-source friendly. He's more focused on balancing corporate needs with community projects, which might not sit well with the open-source crowd. And with the critical remarks from Comfy, it's clear that many interests are clashing.
Just my two cents. Take it with a grain of salt.
@@TheDrumtracks how much vram do I need for it? Can my poor 6gb card handle it? Is it harder to run than sdxl?
@@DaniDani-zb4wd I tried the google colab demo. It loads two models 9GB and 10GB. So it won't work with your 6GB card.
That surprise equine pun hurt me to my core. Very nice.
😅🤣
Thanks for the Update :)
You bet! 🌟🌟
@@sebastiankamph what a mellow dude cheers brother
+1
I heard forge and automatic1111 are stopping updates and everything will go to comfy. Is this true? Can you make a video about this
No way. Too many people using them. Comfyui is too complicated for most users
@@TheCynicalNihilist Sort of. I want to see where it's heading first. But eventually, yes!
Sad if true
@@NotY0urHeroo why so? comfy is more powerfull. I was using a1111 myself and was skeptical about switch to comfy, but once i tried i wont ever go back, it's just everything in a1111 + so much more.
While this is a good and positive move, it shouldn't have needed to happen in the first place. You also have to ask is it too late?
The community saved SDXL but it didn't save SD2, so it's not a guarantee that will happen. A lot may depend on this 'new' version of the model.
16 channel vae will save everything! Hopium
This is similar to Unreal & Unity's licensing, though Unity also went through an unpopular license a few months ago.
true!
Can't wait for SD156 when they finally fix hands lol
Do they have enough of the team left to fulfill those promises?
Isn’t section 4 of the new license still really bad? It’s saying it can’t use SD3’s output to finetune anything other than list of allowlisted models (and its derivatives).
Hopefully they'll release pony for sd3 now
What is pony?
@@mik3lang3lo I think it's p0rn
@@mik3lang3lo hmmmmm nothing xD
@@mik3lang3lostay pure. Don’t get swayed in by the pony 😢
@@mik3lang3lo Pony is a trained checkpoint that will forever change your world view of image generation... Like... FOREVER...
Just be a little nice? Stability AI didn't make a "mistake", they slapped the community in the face. They deserved all the ridicule they got and more. Does this new mea culpa hold any weight? No. Do we trust them moving forward despite this new back-tracking? No. Who is even in charge at the company anymore? This whole thing stinks of a ploy. Let's follow Comfy and see where he and his folks lead us.
Is this license fixed? I have no idea how I would effectively distribute "this license" with my derivative works. Github is not an issue as I can certainly throw a file up. With image generation I have no idea. Meta data maybe? No idea for prints if I am using a third party.
An image created via AI is not considered a derivative work as far as I understand. Selling a derivative would be like selling a fine-tuned version of the model itself.
Is stable Diffusion 3 better than 1111. How is anatomy now is it working properly now? What's your opinion?
a1111 is just a user interface to load stable diffusion models. SD3 is better than 1.5 and SDXL in SOME areas, but it's still lacking. If the human anatomy gets fixed in next release, it will most likely surpass 1.5 and SDXL in quality, not counting 3rd party models or extensions like controlnet.
@@sebastiankamph Thanks for the answer
So if someone fine-tune SD3, and releases that one, does company than need a license for that one, if they earn over 1 mil, or who has the rights for a fine tuned model?
So, a fine tuned model is similar to modifying a vehicle. You can put bodykits on it, buy aftermarket upgrades for the engine and bigger wheels etc. But you're still driving the same car made by the same manufacturer. I think this analogy is best.
Another way to look at it is like roblox or unreal engine. You can create a game within their environment, but your game cannot exist or run without the underlying engine code.
There was a perturbed version of sd3 that was on civitai that did better with anatomy and the reason seemed to have been a built in censorship later in the mod
Still won’t be using it, they showed their true colours and what they actually want… don’t think they won’t eventually go after it again
Really glad they listened. Now they wont die like adobe should
Now I just wish they would remove the censorship on the models.
At this rate, you will die of old age before that happens. No company currently on the AI race is going to risk their necks for our entertainment. They will keep the uncensored versions for themselves. Our best shot is to fine-tune and that will have its limits because they will intentionally continue to release the worst versions of their models until they can ensure their best versions are not exploited for adult content.
Just wait for Pony SD3. SDXL was also censored and now we can prompt titties and other kinky stuff. 😏
Du har en så trevlig röst. 😊
Tack, väldigt snällt! 😘
they realized they are nothing without the community
Its getting somewhere
too little too late, I guess.
how to mix two faces to get a new unique face and not just generate according to a model or lore. To create a new face that can already be swaped or made lore
Try using 2 ipadapters
Is the only thing that has changed is the license? what about the sensor? can we make nsfw images with sd 3
The base model hasn't been trained on any NSFW material, so it doesn't know how. It simply doesn't know what people look like without clothes since it's never seen anybody without clothes.
When/if we get fine tunes it'll probably be ok.
@phizc thanks for the explanation man!
@@phizc That can't be true. They need billions of images to train, which means a giant dataset that wasn't manually put together, which quite simply can't completely avoid certain kinds of content reliably. I'm sure there was nudity in the initial training set.
@@peterlewis2178 maybe there was. But it wasn't told what it meant.
CLIP works by associating text tokens with an image. For example, you can show it hundreds of images of dogs along with descriptions such as "A brown dog in the field" and "A Labrador dog on the sofa" and the model will learn what a dog is by what's common-ish in the image.
I'm not 100% sure how the diffusion model uses the CLIP guidance exactly, but the main point here is that it needs the description during training (or at least booru style tags).
So, if you don't want the model to know how to create images of naked people, celebrities, copyrighted materials, artists styles by name, etc., you simply filter out all images that contains any of the "censored" words. So while it's probable that it's seen naked people during training, they have not been tagged or described as being naked. If you ask the model to create an image of a naked person, it doesn't know how.
It's like showing an artist a thousand photos of people without saying their names, and afterwards asking the artist to draw an image of one of them by name. Of course that would be impossible since it could have been any one of the people in the photos, but the artist might try to use statistics if their memory is great. If the name is female, ignore the photos of males. What's the average look of the females shown, e.g. skin color, age, clothes, etc. So they would be able to draw an image of someone, but it probably won't look anything like the correct person.
Coming here for the SD news, staying for the Dad Jokes!
So what if we don't have public numbers of the yearly revenue? On the other hand, the owner, of the holder, of the mother company does have, but that would mean way more than a million dollar, but on the other hand, the small sub company does barely make any money. How would that 1 million dollar limitation make any sense. So if the mother company makes 1 billion a year but the small sub company makes +-0 money, they now need to pay for an enterprise license? I think that's so strange.
Now waiting that CivitAI approves it, then we get custom SD3 models.
Still optimistic, huh :) It depends on how interesting SD 3.1 will be for the community, we don't know it yet.
So your statement means that you are going to do it? A lot of big players in the customising game already stated on Reddit that they won't touch SD3 because anatomy is too broken.
@@bxl2012 Lets see what kind of fix they release..
The new license still gives them nearly the same powers as before for people using it commercially. They can revoke the license.
Hallelyerya! 🎉
Am skeptical, but hopeful. Broken trust is hard to mend.
If the internal testing models are better, then why release a half-baked thing like this and not just wait a couple more months until it was all done? Sure the license issue might have also exploded, but it might not have been as bad if the product was good to go. idk
Is it too late, though? Image generation, while impressive, is still relatively niche. With LLMs and text-to-video models on the rise, StabilityAI was already behind the eight-ball in terms of relevancy. Hopefully the community quickly forgives and then breaths life into this new model that was otherwise DoA.
Will it come out for automatic1111?
definately. A1111 is just an interface for SD models to load into.
@@brootalbap Yes, but SD3 is currently only avail for Comfy
Go 🎉🎉
0h shocker so it ended up being EXACTLY like every other one they have lol. shocker the ai community hyped again
👋
🤩
Too late too little. They ignored the community for weeks, nobody trusts them anymore and moved on to other alternatives. They should've did that the first week.
What are those alternatives?
@@tomreiner8676 Currently SigmaArt but at the moment half of the team that left StabilityAI including the creator of ComfyUI are in pre-production phase of OpenDiffusion (a new model)
There are no other good open source alternatives. The only alternatives are closed source
Maybe playground model. @@titanmars9229but for me ,SD is the only thing I use.
@@titanmars9229 This is subjective opinion, not universal truth. There are a lot of good open source models. BTW Stable Diffusion was NEVER OPEN SOURCED, only open WEIGHTS, which is completely different :)
Lada hruska
looking kinda young ...
I just want to know who's the actual burger were so crazy and hyping so much on SD3 before releasing to public to use and reviewing ?They always telling people what this or that's so and so good compared to an old one and bla bla bla.....I remember everyone been told loudly to hear that SD3 is freaking GOOD like superman right? Because the reality is none of the latest release is better than the last one! It's always like this. Now I found that they are actually the one who threw the shit to thier own faces.... that's what I know!
SD3's launch isn't the first bad move from Stability AI. There are way nicer people to talk to (and way better open source licences) with Pixart and Lumina. SAI sees us getting ready to leave SD and go for better options, and now they remember we exist because they discovered (one more time) that without the community finishing the job for them they are dead. Especially after not having been able to find someone to buy 100 millions of debts growing by 7 millions of deficit each month.
Get lost SAI, you're bankrupted anyway.
BTW "open source" means you release the source, where is it ?
Well as far as i understood what i have heared from a video of the ceo from invoke ai, i would disagree with some of your statement..
the license told last time , that if you stop paying for SD3, all of your works, all of your derivative works, including build loras would have to be destroyed...
and the new version isnt better here in any way..
So... its still a no go...
th-cam.com/video/23tEjvCNZHA/w-d-xo.html
Oh as far as i see, my linked film now seems to be outdated right? so this would be not perfect but an improvement
Can someone go over the licensing again. If its free, does it still mean they have full control over your work such as image output with sd3 and custom models? The new ai ryzen CPUs were suppose to be able to use sd3. Completely useless model. Better just get an nvidia GPU.
I'm sorry, I don't think I will ever trust them again. My opinion.
Here always coming back to check Sebastian’s dad jokes moment 😅. I hope Stability AI notices that they are not just an entity in AI development/competition, but is also a unique & powerful engine in creating the AI “culture”, just like the anime/manga in Japan being supported by millions of secondary, tertiary creators built upon its open-source ecology. I don’t even mind paying for the models and the licenses (so long as they are within a reasonable/affordable range.) Curious about how they will build a sustainable monetizing model. What’s Sebastian’s insight?
@@hh1162 honestly, I think it's going to be tough. They have built their product around an audience that generally likes their stuff free and open source.
been a while since Ive checked in on AI news since I kinda abandoned this tech in favor of 3d modelling an animating, but it's interesting to come back here and see what's going on, thank you for that.
I've heard that the human generation problems are at least in part due to their NSFW filters?
That's what people are saying, but I don't know if it's true.
Never stop the dad jokes bro.... fuckin makes me wanna subscribe with my alt accounts I love it so much.
@@user-ne7ok9sm4i thank you, that's very kind of you ☺️
Lawyers fucked up, like they always do. Same story with Adobe.
so sd3 is taking lessons from Unity game engine on policy. thick it would be best not to use sd3 for a long time. and this explanation sounds more like a smoke screen to cover up the BS.
It's still very concerning that they attempted the license model they did. It's a shame there aren't yet any good alternatives.
Seb. Beard where
What's up with the hat?
FIRST?
Woooh! 🌟
SECOND ?
Almost!
SD3 Medium is AWFUL.
fireship can explain anything in 100 seconds, but this guy needs eight and a half minutes to go over this. It's 2024. You need to step it up. Just being honest.