- This topic has 43 replies, 14 voices, and was last updated 1 week, 1 day ago by
.
-
Posts
-
yes, but its totally fixable, very easily. That was a quick 1-minute generated test. The problem areas can easily be masked, then re-generate the image only altering the masked area, the rest of the image remains unchanged. Plus, lora models come out all the time to specialize fixing details such as that, so it happens even less often on the first attempt.
@3demon, compared to mine they're perfect.
It's fine that you can edit the toes in seconds. Wonderful, even.
Yet you posted the image without editing the flaw. Even though it would only have taken you seconds to fix it. Perhaps you didn't even notice it, expecting that AI would have generated them correctly.
But the 3D models I use don't need editing to correct the number of toes. They begin with having the correct number of toes.
I suspect the AI's models also had the correct number of toes. Every single one of them. And yet...
So I suppose the point of this entire exercise is...why? Why do you evangelize AI image generation? What is your goal? Why do you think we need to use AI image generating?
Actually, I didn't notice the toes problem, and still don't. You said they look "ugly" so I assumed it was a preference thing. If they are not to your liking, you can mask the area, change the random seed, and spin again, or use a refiner to specialize in that area. Didn't think it was such a big deal it needed to meet anyone's expectations.
Evangelizing? Lol. Just pointing out inaccuracies about the current state of AI models, its not 2022 anymore. There are tools that many will find useful for their needs, and figured I would show some of the things now possible.
Everyone is free to choose whatever tools they like, I don't have any "goal" here, I don't give a flying fuck if you create your art with your mouse or if you fart on the keyboard to produce it.
I didn't say the toes were "ugly". I said they were "goofy", in an effort to draw your attention to them. Generally speaking, humans (with the noted exception of Ethiopia above) have 5 individual toes; not toes melded together. That you didn't notice, and still don't, is unsettling. I'm merely drawing attention to it. Which seems to bother you, given your snark. There's no need to be defensive. Not all criticism is hatefilled. And your goal, "There are tools that many will find useful for their needs, and figured I would show some of the things now possible." is fine; admirable even.
Nonetheless, a sincere thank you for introducing these AI image generating options. I hope it helps more people have a better understanding of what current AI image generation can bring to the table.
Ai is good for many things. Like pr0n, pr0n, more pr0n, and funny video's.
Kids are off limits. That push failed as the noticing gatherted pace. Now sad individuals are salivating over the latest AI image generating fad..Teen females!See Civit for more eye burning filth than you can handle. Of course you can create cats in pretty outfits if you're tired of pr0n, or dump AI entirely and do something of real value 🙂
Yeah, I use DAZ for the same thing as y'all do. To me, DAZ Studio is like a sandbox, it's really good and useful for rough drafts and using for one's own imagination too, and I absolutely love the fact that I can easily make OCs without the need to sculpt everything myself.
I'm wasn't defensive about the AI tools whatsoever (I could care less whether someone likes the tools or not)! I was slightly defensive toward the accusatory attack! ( of having some "goal" or objective, as if i'm trying to pull a fast one on you or something?) I don't get it. Then accusing me of evangelizing it like a salesman, but then complaining that I didnt create perfection to sell it to you ?? (make it make SENSE!!).
It was a fast demo of a few capabilities of modern tools, take 'em or leave 'em.
That said, I know there are some people who have some hatred toward anything AI-related, no matter what. That seems kind of bonkers to me, that it drives people crazy. I don't understand that. If it's something that can help you achieve great result quicker, it seems crazy to me to ignore it. It's like a stubborn old man trying to work on his car to get a stubborn bolt off, and refusing to use air-impact wrenches because back in his day all they had was a box wrench. It seems crazy to me.
A question as to your motive, which you explained, isn't an attack. That's why I said you were defensive, which you admit to, slightly or not. How you interpret a question, devoid of bold capitals, exclamation points, or profanity, is on you.
Expecting the requisite number of toes on a human (in the process of your demonstration of the evolutionary advances in AI image generation capacities) isn't asking for perfection. It's the standard. Ethiopia notwithstanding.
No one in this forum has ignored AI. And if you were of the belief that you were going to soothe their seething "hatred", I don't know what to tell you. But you are under yet another misapprehension that people only want a quicker result. As I and others have explained above, it is the process we enjoy; a process Stable Diffusion, ComfyUI, etal, cannot approach. Not yet anyway. That you don't understand, despite it being made clear, is only because you can't understand. But you will...in time.
@inversekinematics I hear ya
You keep harping on the toes, when I still don't see the issue. I count 10 toes on the seated image, and on the standing image the pinky toes are occluded (as they would be in many occasions when someone is wearing those type of shoes, since they are shorter toes and don't extend far enough! they would be pressed against the side of the 'band' around the shoe's upper).
Expecting the requisite number of toes on a human (in the process of your demonstration of the evolutionary advances in AI image generation capacities) isn't asking for perfection.
The same can be said for ANY tool, including DAZ. I guess those 'imperfections' are acceptable to you. All the pokethoughs and D-Farce BS, those are all totally fine for you and escapes critique (even though they are more of a challenge to correct when they arise).
There's a reason why even software such as Photoshop are adding AI edit models to their products: some people find them useful. If you don't like them, don't use them. It's really that simple.
You'd have to be guessing, because you have no knowledge of what is acceptable to me. You aren't aware of my critiques of Studio because it isn't the topic. That's a hackneyed fallacious attempt to shift to a topic we aren't discussing. But just to be clear, I don't consider pokethrough acceptable. I'd consider evident pokethrough to be the careless oversight of the user, such as yourself not even noticing your model has melded toes during an enthusiastic demonstration of the wonderous capacities of AI image generation. And then pretending it's natural. But notice this: I'm not evangelizing Studio.
No one at all cares that you prefer AI. No one at all cares that I don't. I too believe it is a valuable tool. And it will get even better than it already is. I believe I've made that clear.
Thanks again for the demonstration. Let it rest and go enjoy making images. As I will. It's really that simple.
No worries, it's all good. And to be clear, I also DO NOT prefer AI, but I *DO* prefer it as a tool in conjunction with many other tools. It's just one spoke on the wheel. I use it with various other software, DAZ included, and utilize each program to their own strengths. DAZ is great for setting the scene, and there are many other things far better suited to handle stuff that DAZ sucks ass at doing, and that is where other tools come into play, AI edit models included.
right now we have an exponentially increasing range of options catering to all kinds of preferences.
i am currently returning back to my interest in dabbling with customization of 2d fighting games with pixel art - street fighter, martial masters, king of fighters, kaiser knuckle et al
i want to expand animations, i.e. add more to the original sprites and started with generating a 3d model from a reference image in order to process it and apply npr shaders to it in blender.there is a site called assethub which i tried out, and it nailed the pose and the art style, it separates the mesh into body, head, clothes, accessories etc. quite expensive, so after i used up the free credits i won’t return for now, but these are early results and they are already impressive.
doing this manually would have taken weeks, likely longer. this happened in 10 minutes total. taking the grind out of the creation process is quite an upgrade, i’d say.
- You must be logged in to reply to this topic.
