Join The Tinkerer on Whop to download this model and many more — all for one monthly payment. Plus, get early releases, private tools and a lot more. 👉 Join on Whop
CyberRealistic Pony blends all the charm of Pony Diffusion with the striking realism of CyberRealistic. The vibe? You get everything from adorable to bold (sometimes both at once) with crazy-detailed textures, moody cinematic lighting, and a hint of AI flair.
🧠 How to Use It
Sampling method: DPM++ SDE Karras / DPM++ 2M Karras / Euler a
Sampling steps: 30+ Steps
Resolution: 896x1152 / 832x1216
CFG: 5
Clip Skip: 2✅ Recommended Prompts
score_9, score_8_up, score_7_up, (SUBJECT), Negative Prompt examples:
score_6, score_5, score_4, (worst quality:1.2), (low quality:1.2), (normal quality:1.2), lowres, bad anatomy, bad hands, signature, watermarks, ugly, imperfect eyes, skewed eyes, unnatural face, unnatural body, error, extra limb, missing limbsit's also possible to remove the normal pony tags:
(SUBJECT), Negative Prompt examples:
(worst quality:1.2), (low quality:1.2), (normal quality:1.2), lowres, bad anatomy, bad hands, signature, watermarks, ugly, imperfect eyes, skewed eyes, unnatural face, unnatural body, error, extra limb, missing limbs🛠 ADetailer Settings
Adetailer model: face_yolov9c.pt
If you only want the main face being refined set 'Mask only the top k largest' to 1.☕ Support
Enjoying the ride? [Buy me a coffee] – but only if the Pony delivered something you actually like.
⚠️ Disclaimer
This model might generate sensitive content. Whatever you make with it is on you. Don’t do anything weird and try to blame the horse.
🔗 Links
Backup location: huggingface
💡 Need Better Prompts?
This custom ChatGPT was made to whip up top-tier prompts just for this model:
🔗 [Try it now on ChatGPT]
Description
Lot of small improvements. Slightly better use of some realistic character LoRA. Better lighting and reflections.
Include tags (score_9, score_8_up, or score_7_up) in the positive prompt for better results.
FAQ
Comments (50)
Excited to try this out! my realistic Pony models are gonna make sweet love to your model and create some beautiful little merge babies
Aaaaaaaaaaaaaaand here comes 6.7 hype train.
6.6 has better lightning. What worse: anatomy, pose adherence.
6.5 gives 10/10 buttjob, 6.6 fails, it gives pure penetration, besides buttjob. Thats just an example.
Doesnt understand from behind, always front view pose.
Eyes are clearer.
I expected a V7, but a fixed 6.7 is good too.
Having some way different output compared to nuanced changes in previous updates. Always loved the work, but right now it is very off for me
Can you explain? Curious what it's completely different etc.
I love your models, but why did you remove access to 6.5?? It was my favorite model to use
Cant you generate local?
I don't have control about what checkpoint Civitai choose to use. In a ideal world you can choose the version or use every checkpoint but unfortunately this is not the case
@Cyberdelia thank you so much for answering me, I'm not an expert here 😅 Okay now I get it. Congrats for the great models, really
@morireoluwakindrick418 I'm not an expert, I'm trying to understand how this all works. What do you mean generate local?
@v999 download the model, a program which can run it, thats it. There is ton of guide in article section. Search for beginner ones. You wont need buzz anymore. And you can generate so much, what do you want.
@morireoluwakindrick418 Thanks for the tip, but I guess I'd need a powerful computer to do that
@v999 what gpu do you have?
@morireoluwakindrick418 I have a Mac laptop, close to explode...
@v999 buy a cheap pc, with at least an rtx 3060 12gb. Its not expensive, and you can run everything, even flux.
With 2gb, you can run SD1.5, with 4gb SDXL and Pony. But these are bare minimum, speed will slow, of course.
Oh shit…. Now I have to try it with the prompts of my best images.….😁
Thanks CyberD!!
Oh, how awful! I feel for you ;)
I love your checkpoints! Just bought you a Ko-Fi!! Cheers! Keep up the good work!
Not necessary but appreciated :)
Thnx btw!
Hmmm... surprisingly, but I can't repeat the same quality as I made on version 6. 5.
Nice work 6.6 , not fully tested so can't say what the difference is yet but will in time
Try V7.
@Cyberdelia Damn i just posted more from my latest set and didn't see V7 , i might use the same maybe change it a little and try it on V7
@hotbabestyles sorry!
@Cyberdelia not a problem my friend , only issue is you get disrespectful people deside to repost those last images i posted on V6.6 and post them in V7 , thats why i posted them in V7 as well , hate when people do that , the paint is basically still wet ...
6.6 is producing anime like results for me. using the negatives as well..
it works aster so performance is better.
realistic accuracy is seems reduced, I'm sure you'll fix it, you is the G.O.A.T.
@civitai6999 local or civgen?
Yeah, I can see why you get these Anime results. Working on new version since this release.
@danelbrodix926 Yes curious about that also. Realistic is not worse then 6.5
@Cyberdelia Hi, finally you are here
Can you fix pose problems?
V7 when?
How you train your models, to remain full blood, but realistic?
@danelbrodix926 I've always been here, but sometimes I just don't feel like responding. Got pills for that ;)
@Cyberdelia Ok!
@danelbrodix926 Can I fix the problems? Absolutely. I have another version ready to upload. I’ll probably get a lot of negative comments for uploading so soon, but that’s just how it goes.
@Cyberdelia VVVVVVVVV7777777777?
@Cyberdelia Im tired of XL hybrids, i want v7:)))
@danelbrodix926 I can also call it "V7 Maximum Pro - Double Gold premium version"?
@Cyberdelia Whatever you want, just release it, later you can bugfix:)
@Cyberdelia All your best models are side versions, bugfixed. 4.2, i dont think anybody can surpass. 6.5 until now. Sdxl i dont know.
local, i tested it with very difficult context and v6.6 understands the task much better, only anime like is a minus, Cyberdelia is aware already and will make civitai great again
@civitai6999 I dont see any anime tending. I use 2m Karras, 4.5 cfg though.
@Cyberdelia v 6.6 works fine for me on A1111, i'll see v7 and see if i find any key differences in quality on realistic model.
@Cyberdelia refer to this to see if you spot any anime tendencies in v 6.6, i see none https://civitai.com/images/46195763
@pokezilla39966 Haha, you can make another one.)))))
me too, but with this generic negative im getting good realistic results:
https://civitai.com/models/411532
@Cyberdelia yeah, oof, i'll stick to v6.6 until you sort out issues with v7 next year, https://civitai.com/images/46297809 (for context, i changed nothing between ai generation, all used Euler A with Simple Schedule Type on A1111). NO WAIT, IT WORKS....only using your negative lora though, otherwise it's a over contrasted mess.
@pokezilla39966 you use the negatives as lora? I was putting it to embeddings
@civitai6999 i just downloaded the negative lora and put it in embedding file on local A1111, it works for me.
@civitai6999 v7 works for me if using xyz plot with v7 first, then say v6.6 then v6.5 then it doesn't create a over colored messes, if i use v6.5 then v6.6 then v7, v7 generates blurry messes, shown better in this https://civitai.com/images/46349102
Details
Files
cyberrealisticPony_v66.safetensors
Mirrors
CyberRealisticPony_V66.safetensors
cyberrealisticPony_v66.safetensors
cyberrealisticPony_v66.safetensors
cyberrealisticPony_v66.safetensors
cyberrealisticPony_v66.safetensors
cyberrealisticPony_v66.safetensors
cyberrealisticPony_v66.safetensors
CyberRealisticPony_V66.safetensors
cyberrealisticPony_v66.safetensors
CyberRealisticPony_V66.safetensors