Theme editor

  • RequestStream Movies, TV shows and anime streaming • 1 week trial
  • LewdCorner Site Cleanup Update
    A new cleanup update has been posted covering the recent Vault rework, rank changes, policy cleanup, and theme polish. The goal is to make LC cleaner, easier to understand, and safer for the site going forward. - Jack Of Blades
    Read More

AI Art Recommendations

  • Thread starter Thread starter Dean
  • Start date Start date
  • Replies Replies 110
  • Views Views 12K
  • Tags Tags
    ai cg
Is one better than the other or is it more a matter of what packs are you using? I cant run either without setting my laptop on fire but I'm interested in how the AI art works and is developing. I know MidJourney is a bit more 'artistic' in its renders but I like the pics made with stable diffusion better either way I see a job in working AI becoming big in the future. If I was 20 years younger I'd be learning that instead of trying to translate manga like I did back as a teen.
 
prety sur it's not too late, at least for fun!
Unfortunatly I do not have a NVIDIA card so with stable diffusion it takes 5mn to generate a picture. Does MidJourney also works with NVIDIA CPU ?
 
It honestly depends what you want and what your hardware capabilities are. Stable Diffusion, whilst I believe it has a browser client, is really built to be downloaded and run on your device/laptop/computer. From there, you can tweak it, grow it, build upon it. That's why Stable Diffusion images tend to be more complex - they have a lot more input, and therefore a lot better output.
MidJourney, on the other hand, is just done through Discord - you write a prompt, it takes place in the cloud, it comes back to you. For most peoples AI needs, MidJourney works great - it's what I use in my day-to-day. I also use DALLE which is now in-built into ChatGPT. It's getting better by each roleout.
Honestly though, if you're looking to just dabble? Try . Totally free and browser based, don't even need a log in.
 
It honestly depends what you want and what your hardware capabilities are. Stable Diffusion, whilst I believe it has a browser client, is really built to be downloaded and run on your device/laptop/computer. From there, you can tweak it, grow it, build upon it. That's why Stable Diffusion images tend to be more complex - they have a lot more input, and therefore a lot better output.
MidJourney, on the other hand, is just done through Discord - you write a prompt, it takes place in the cloud, it comes back to you. For most peoples AI needs, MidJourney works great - it's what I use in my day-to-day. I also use DALLE which is now in-built into ChatGPT. It's getting better by each roleout.
Honestly though, if you're looking to just dabble? Try . Totally free and browser based, don't even need a log in.
Thank's dude!
 
Thank's dude!

No problem! Happy to help.
Any images I past on this site, except for Memes/Gifs are all made in Perchance! So, it's quite capable
 
Stable for control & the ability to tweak.

Midjourney for decent results with ease of use.
 
Stable for control & the ability to tweak.

Midjourney for decent results with ease of use.
I cant run stable on my laptop but Midjourney isn't worth subscribing to for me at this moment. I hope they make specifically designed AI designed GPUs before too long AI and crypto mining keep driving up the costs and makes getting good cards hard to get for actual gaming. I'd run a duel GPU setup if the architecture and software was set up better to optimize different kinds of tasks.
 
I cant run stable on my laptop but Midjourney isn't worth subscribing to for me at this moment. I hope they make specifically designed AI designed GPUs before too long AI and crypto mining keep driving up the costs and makes getting good cards hard to get for actual gaming. I'd run a duel GPU setup if the architecture and software was set up better to optimize different kinds of tasks.

If you cannot run Stable, you might try Fooocus. Its similar, but is slightly more forgiving on the system.

Honestly though, unless you have 8gb or so Vram (don't listen to the people who say you need 16/24, while true for video, pictures work just fine at 8gb), I wouldn't consider local rendering.

Not sure what kind of Laptop you are running, but I would have a look and see if the videocard can be upgraded. Even if its to an 'older' (note: still perfectly fine) 3070m or something similar. As long as its 3X gen you should be ok for running pretty much any local generative method.
 
I have tried a few text to image ai generators. Same results ... mediocre!

Has anyone else had better luck? 🤔
 
I think you can use the onsite one for Civitai for free. Haven't tried tho. I use stable diffusion locally.
 
If you want good results, you almost need one that you run from your own pc.
Or spend much time to tweak the prompts of the online ones and go thru hundreds of generations to get some good images.

I think you can use the onsite one for Civitai for free. Haven't tried tho. I use stable diffusion locally.
As far as I know it doesn't allow much NSFW stuff. :unsure:
 
If you want good results, you almost need one that you run from your own pc.
Or spend much time to tweak the prompts of the online ones and go thru hundreds of generations to get some good images.


As far as I know it doesn't allow much NSFW stuff. :unsure:
Good to know. Thanks
 
If you want good results, you almost need one that you run from your own pc.
Or spend much time to tweak the prompts of the online ones and go thru hundreds of generations to get some good images.


As far as I know it doesn't allow much NSFW stuff. :unsure:
How to run from PC?
 
How to run from PC?
It's too complicated for a single post. It's way easier and more helpful for you if you google for installation guides for it. 😅
 
Anyone is using SD 1.5 ? any help on that? i feel like my prompt sometis just don't work. i am trying to use lora i make but sometimes it just don't do what i write. any help with models to use too? and what is the best way to create a Lora for nsfw use?(does regularization images do something)
 
It's likely you'll need to disable the NSFW checker which automatically blocks some of your outputs.
 
Back
Top Bottom