FreeCatFights

Hello everyone,

As some of you know and many of you can see, I've been a part of this community for quite a loooong time. My journey began as a young one lurking the internet, indulging in a new guilty pleasure. It kicked off in 2015 with my first digital artwork, and a lot has changed since then.

Over these 11 years, I've met incredible people, experienced highs and lows, completed amazing and unique commissions, and even faced challenges like being banned on DeviantArt and threatened.

In 2022, the landscape of digital art changed dramatically. The influx of AI-generated art made it difficult for my hard work to stand out. I tried to adapt by also creating AI art, but it never felt right to me. I crave control over my creations, and simple AI just doesn't allow that.

It was disheartening to see my meticulously crafted 3D art overshadowed by quick, AI-generated pieces. Creating a single image of detailed 3D art can take me up to a whole day, whereas AI art takes mere minutes. The Zeitgeist changed. Amazing long-time artists got sick of it and some even stopped creating all together, which further dampened my enthusiasm.

Despite these challenges, I've decided to not throw in the towel just yet!. I'm excited to announce that I'm reviving my artwork with a new approach!
I'll be combining my OG 3D skills with AI to create art that retains my desired control and level of detail and still gets the glamour and gloss of Ai.

I'm eager to share this new style with you all and might even start a series explaining my process if people want to get into it also!

Additionally, I want to let you know that I've heavily invested in upgrading my tools, including a high-end graphics card capable of outdoing NASA's. I am now fully open to commissions and ready to take on new projects. So feel free to hit me up here or on DeviantArt if you want anything made.

Thank you for reading my rant (& shameless plug) . I look forward to creating more art and engaging with our fantastic creative community once again. Here's a new piece to mark my return—hope you enjoy it!

Just offering encouragement to hang in there. It seems you are on to a innovation within your art form so don't give up on it and thanks for your efforts.

Good luck!

The moment is electric, as the sensation of soft flesh pressing against soft flesh.
Their two tense bodies press together, locked in a breathless standstill.

Each girl feels the other's heartbeat through soft, yielding tit flesh—both hesitant, both hungry for the fight to begin.

This is the last calm before the storm, blurred lines between fantasy and reality where fantasy teeters on the edge of reality, and the chaos of battle is just a heartbeat away.

These are absolutely incredible! Probably the best AI art that I have seen on here.

Incredible! I've been waiting for this day! Your results are truly next gen and you should be very proud of what you've managed to achieve.

I would love to hear details about your process. Are you still generating 3D models, but then augmenting the final image with AI? If so, what tools are you using?

Good to have you back!!! loved all your work

Quote from: Abacust on August 15, 2024, 11:04:17 PM
Incredible! I've been waiting for this day! Your results are truly next gen and you should be very proud of what you've managed to achieve.

I would love to hear details about your process. Are you still generating 3D models, but then augmenting the final image with AI? If so, what tools are you using?

Thank you so much, Abacust (and all other sweet supportive commenters)! Your comment(s) put a big smile on my face and truly made my day :-* I'm thrilled that you appreciate the work—I've been working hard to push the boundaries and hearing that it's resonating means the world to me.

As for the process, get ready for some geeky details! I start by using Daz3D to create a scene, just like in the "good old days." Then, I use AI (specifically, Stable Diffusion through the SD Forged UI) running locally on my computer to take that 3D render and enhance it with a touch of AI magic. The AI uses the render as an anchoring point, allowing me to blend realism and artistic flair in a way that feels just right.

If you're interested, I'd be happy to dive deeper into the workflow or even help you replicate these results if you've got a decent PC setup. I'm all about sharing knowledge within our amazing community—more artists means more creativity, and I'm totally against that scarcity mindset. Let's keep pushing each other to new heights!

Thanks again for your support, and feel free to reach out if you want to chat more about the process!

Her voice drips with challenge: "Are you just going to sit there?"
His gaze hardens, fists flexing as the words sink in.

On the edge of chaos, tension buzzes in the air, the line between calm and an explosive couple vs. couple brawl growing razor-thin.

He turns to her, a fire igniting in his eyes, and asks,

"Are we going to take that?"

One spark, one move, and it all kicks off.

So you have any trouble with nudity filters? Is there a cloud model that doesn't have filters that would allow you to continue using your workflow? How long does the AI portion of your workflow take for a single image? What hardware are you using?

I have... Questions


Edit: I just realized that NSFW art generation has jumped in the past few months, nice! Lots of good free options now, but we need to find a good image to image option so everyone can do this. Who's with me!

As one of the many people dabbling with AI I apologize and I would encourage the real artists to keep working at it.  AI allows people with me with no identifiable skill to bring our imagination to life, but there are a lot of constraints and gotcha's.  Extra limbs, weird faces, constraints on models, lots of processing, limits on clothing, etc.  I can create an image, but I really can't create a story.  I will say that as I am scrolling over DA, I kind of skip most of the obviously AI generated artwork.  That is probably just me, but it is the way I scroll.  Most of the AI stuff is pretty but passionless.

The struggle is nearly over, but the emotions are far from it.

A tight rear choke hold, her grip yielding as exhaustion hits. Beneath them, the soft sheets offer a cruel contrast to the harsh reality of their battle.

Tears stream down both their faces—pain, exhaustion and raw emotion blending into one.

Incredible.

We want to know more about your process if you're still interested in sharing.

Also, how long does it take to do this from start to finish?

Hey everyone!

First off, massive thanks for all the love—seriously, you guys make this all worth it! Like I promised (looking at you, Abacust, and the rest of you sweethearts :-*), I'm here to share how you can dive into AI just like I have.

Step 1: Download and Install Stable Diffusion Forged
Before we get to the fun part, you're gonna need Stable Diffusion Forged (SD). Here's the magic link:

• https://github.com/lllyasviel/stable-diffusion-webui-forge (https://github.com/lllyasviel/stable-diffusion-webui-forge)

Now, here's the deal—just follow the instructions on that page. If you're digitally challenged or tend to accidentally nuke your PC every time you install something, just hop on YouTube. There are plenty of friendly tutorial videos (seriously, I swear 80% of the population on YouTube is Indian guys showing you how to install everything—legends).

Step 2: Get Familiar with the Program (Yes, You Need to Put in Some Work)

Now that you've got Stable Diffusion Forged up and running, it's time to get comfortable with the interface. I'm not here to handhold you through the entire learning process (we don't do lazy here), but I'll give you the big lines to start with:

• First focus on installation—get that bad boy running smoothly. Don't rush ahead before you've got it all set up and running correctly.
• Don't waste time searching for the "right" model (yet). Instead, start with something tried and true. Download Realistic Vision V5.1 from Civitai here.(https://civitai.com/models/4201/realistic-vision-v51 (https://civitai.com/models/4201/realistic-vision-v51)) It's a great general-purpose model that can handle most things without fuss.(read the instructions)
• Prompts: Another tip here—don't reinvent the wheel. Steal from the greats. Head over to Civitai and check out how the real nerds are doing it. Imitation is the greatest form of flattery, and it's also the first step in learning (ever heard of "imitatio" and "emulatio"? Yeah, all that).
• ControlNet: Once you've played around a bit and feel comfortable with the basics, it's time to level up with ControlNet—this is where you can get really precise with your images. But don't jump straight into this unless you've mastered the basic stuff first.

Post Scriptum: Don't Be Lazy – Learn Before Asking for Help
After messing around for a bit, then we can talk ControlNet (don't worry, I'll explain it below). And if you need a more detailed explanation of the basics, just let me know. However... please don't be that person who cries for help without at least watching 1-2 YouTube tutorials first. Seriously—give it a try on your own before crying wolf!

How to "Git Gud" with AI - Tip #1: ADetailer! ADetailer! ADetailer![/size]
(https://i.postimg.cc/vBq0r7mb/ADetailer.png)
Alright, now that you've got everything set up and running (hopefully without melting your GPU), it's time for me to share one of my favorite tips to level up your AI game.

First off, ADetailer is the GOAT. That bit of code is pure magic and I want to have its babies. It can take your "meh" results and turn them into absolute masterpieces. If you're not using it yet, stop what you're doing and check this out first.

What does ADetailer do?
Essentially, it scans your freaky AI-generated faces and hands and re-does them, making them look human again.

How to use ADetailer?
It's super easy to set up—here's a handy guide (https://civitai.com/articles/2034/after-detailer-adetailer-stable-diffusion-guide-by-thinkdiffusion) that walks you through it.

Here's how it affects your results:

Without ADetailer:
Behold the flaming hot pile of AI trash...
(https://i.postimg.cc/G2MwNsXm/image-2024-10-11-T175832-265.png)

With ADetailer:
Oh yes, that's it babyy
(https://i.postimg.cc/yNyzmG99/image-2024-10-11-T180256-204.png)

The Magic Recipe (aka very basic Prompt and Settings)

• Positive Prompt:
  A structured jumpsuit with espadrilles, poised, Escort, Average Height, Narrow Hips, Diamond-Shaped Face, Dark Skin, Brunette Hair, grey-blue Eyes, Straight Nose, Thin Lips, Long Hair, round breasts, Barbara Palvin, a secluded, sandy beach at sunset


• Negative Prompt:
  (deformed iris, deformed pupils, semi-realistic, CGI, 3d, render, sketch, cartoon, completely nude, drawing, anime), text, cropped, out of frame, worst quality, low quality, jpeg artifacts, ugly, duplicate, morbid, mutilated, extra fingers, mutated hands, poorly drawn hands, poorly drawn face, mutation, deformed, blurry, dehydrated, bad anatomy, bad proportions, extra limbs, cloned face, disfigured, gross proportions, malformed limbs, missing arms, missing legs, extra arms, extra legs, fused fingers, too many fingers, long neck


• Sampling method: Euler A
• Schedule type: Automatic
• Width: 513
• Height: 768
• Sampling steps: 10
• CFG scale: 7
• ADetailer model: face_yolov8n.pt

I've left the ADetailer prompt and negative prompt empty, but if you want to dive deeper into that, let me know!

Alright, that's it for this tip—now go forth and experiment!

Let me know how it goes, and feel free to drop me your experiments (or, you know, ask if you get stuck somewhere).

How to "Git Gud" with AI - Tip #2: CONTROLNET, the Man, the Myth, the Legend!

Alright folks, now that we've got ADetailer working its magic, it's time to introduce you to the next game-changer: CONTROLNET.

This absolute marvel doesn't just suggest how your AI should behave—it demands it. CONTROLNET essentially forces constraints on your generation, slaps that bitch in the face, calls her a slut, and throws on the restraints. Yeah, it's that serious.

How to Install ControlNet (if you somehow messed it up)
ControlNet should come standard with Forge UI, but if you've managed to fuck the installation up, don't panic. Here's the guide (https://stable-diffusion-art.com/controlnet/#Installing_Stable_Diffusion_ControlNet) to unfuck yourself and do a clean install. Skip to the installation part and get back on track.

Now, let's see this magic in action:

Prompt without ControlNet:
Cute but trash
(https://i.postimg.cc/JhZnBmRQ/image-2024-10-11-T193127-114.png)


Finding a random image on the www that is in line with the pose I want to portray
(https://i.postimg.cc/908FjQBc/beautiful-girl-fighter-standing-pose-boxer-sauna-91582323.webp)

After using ControlNet OpenPose:
(https://i.postimg.cc/nznHjDRH/pose-4.png)

Combining the Two:
Amazing!
(https://i.postimg.cc/85bztHJJ/image-2024-10-11-T193303-061.png)

When you add Upscaling, ADetailer, and a little extra spicy sauce:
(https://i.postimg.cc/hGBn6mq9/image-2024-10-11-T193006-929.png)

The Magic Recipe (Settings):
(Keep in mind these are the same for every result!)

• Positive Prompt RAW candid image of (age 21 swedish girl:1.3), ( model, big blonde hair, pale skin:1.3), (small breasts, thicc ass:1.2), (cinematic establishing shot:1.3), (Professional photography techniques for authenticity:1.2), (background is brick wall, overcast day:1.3), (determined expression, fighting pose, clenched fists:1.2), (wearing tight jeans, plain baggy tshirt:1.2), (highly detailed skin, blue eyes:1.3),
  
• Negative prompt (deformed iris, deformed pupils, semi-realistic, cgi, 3d, render, sketch, cartoon, drawing, anime, mutated hands and fingers:1.4), (deformed, distorted, disfigured:1.3), poorly drawn, bad anatomy, wrong anatomy, extra limb, missing limb, floating limbs, disconnected limbs, mutation, mutated, ugly, disgusting, amputation, smiling, happy, (dark eyes), (gloves, boxing wraps:1.2), logo, blue background, blue clothes, (top:1.2), watermark, (large breasts, nipples), leaning, night, indoors,
• Sampling method: Euler A
• Width: 513
• Height: 768
• Sampling steps:30
• CFG scale: 7
• ControlNet model: OpenPose
• ADetailer model: face_yolov8n.pt

Now, seriously—go try this out. ControlNet gives you complete control over your output, and trust me, you'll wonder how you ever lived without it. Some things are simply better when chained up  :-*

WOW! this is incredible, thanks for such a detailed response, it was worth the wait!

I had no idea it was so involved but the results are insane. Thanks for showing the before and after of each step too, that was really eye opening.

Now I wonder if there are any online services already that combine these steps into one service... Maybe I could make one if not.

Do you have any suggestions when it comes to attempting to do very NSFW content, or do these models not care about that? Also, do you have any other thoughts on the limitations of this technique?