🟡:Flux Models 🔵:SD XL Models 🔴:Expired Models
🟡 STOIQO Afrodite is a specialized model designed to generate high-quality adult photography content. It allows users to create realistic and explicit images with a strong emphasis on fine details and authentic textures. Perfect for those seeking precision in the adult photography space, it provides creators with the ultimate tool to produce high-end explicit visual content.
Versions and Recommended Settings:
MAIN SAMPLER: Euler+ Beta
GUIDANCE (In Forge:'Distilled CFG Scale'): 2+ | STEPS: 15+
(Recommended): GUIDANCE: 3.5 | STEPS: 25
We finally got to the release of the flux.1 Dev version of Afrodite. As for NewReality we are still in an experimental Alpha version, which still has many problems, but which I will be able to intervene thanks to your feedback.
The Main Download (Full 11GB) is the version containing only the Unet, so you can use it accompanied by the clips of your choice, depending on whether you prefer the fp8, the fp16 or want to experiment with new clip_l or t5.
In the file section of the sidebar you can instead find the Secondary Download (Pruned 20GB) of the AIO version with t5xxl_fp16, clip_l and VAE already included.
The Full/Pruned nomenclature is just to be able to keep the Unet as the main download. All the others change with the AIO as the main. The two models are composed as mentioned above.
Refiner is unnecessary and Clip and VAE are included
MAIN SAMPLER: dpmpp_3m_sde + exponential
CFG: 2+ | STEPS: 16+
(Recommended): CFG: 4 | STEPS: 25
Credits:
Afrodite is the result of countless hours of work, involving several hundred iterations through training, merging of models and LoRAs, fine-tuning specific blocks or captions, as well as performing additive and subtractive merges. Due to the complexity of this process, it is difficult to provide a precise, step-by-step account of every decision and experiment. However, each phase has played a significant role in shaping the final product, whether by direct influence or through valuable lessons learned along the way.
Given the intricacies and challenges involved, I want to extend my heartfelt gratitude to the creators of the models that have contributed to this journey, either directly or indirectly. Regardless of whether they were incorporated into the final iteration, their work provided inspiration, insight, and progress throughout the project.
To all the creators whose models supported this endeavor, I extend my deepest thanks. I encourage everyone to explore and support their work, as it deserves recognition for the incredible value it brings to the community.
EauDeNoire humblemikey socalguitarist ZyloO xlabs_ai aramintastudio Seeker70 Ai_Art_Vision PromptoAI alvdansen aip0pp SG_161222
WORK IN PROGRESS...
Description
FAQ
Comments (43)
Can you add recommended samplers, steps, and cfg, for WebUI and SDForge?
I was looking at the previous version, because I want to know that too for new models. There the creator used Sampler dpmpp_3m_sde_gpu,CFG scale 4 ,Steps 30... thats the comfy ones, but works well too for WebUI.You can use this as a starter and play around a bit
@Manuka Thanks, I already tried the Comfy settings (minus the nodes of course) in WebUI and SDForge. I'd also like to see recommended samplers, steps, and cfg, for WebUI and SDForge, but for all the versions. It would make it way easier to test between versions and see where one is excelling and on is falling behind.
Note: I have found that (on a Mac M1 Max) I have to pump up the sampler steps super high to create a sharper focus on subject, at least with DPM++ 3M Edxponential. Also, the simpler the prompts (pos and neg), the better the image turns out. This goes for all the version, at least for my set up. I also noticed that there are problems with hands being malformed across all versions. I assume that's just Stable Diffusion being bad at humans.
Would be good to actually include at least one of the prompts you used. --Leaving it out feels dishonest and unclear if you are actually using JUST the model to get the results
I can get similar results with just the model. No negative prompt, LoRA, or embedding.
Some quality modifiers I've found are: "a film grain photo of", "dark lighting", "dramatic shadows".
Like this: "a film grain photo of a beautiful Peruvian woman lying in the mud, dark lighting, dramatic shadows, garden"
I think the creator is underselling it if anything. It can produce much better photos than he showcased.
Hi @Megasherru , I updated all the generation data, sorry but I didn't notice, I don't know why it no longer uploads the workflows directly together with the images. Now you can consult them, I will investigate to be able to directly provide you with the workflow as well as the parameters.
Negatives are empty because they were not used to generate the example images, so as to provide you with even more real feedback on the model. There is also no use of LoRAs or other image alteration, just simple txt2img.
Thank you for letting me know about this absence, have a good time
Hi @snipsnapsnipsnap Thank you very much for the flattery, I am very happy that my work meets expectations.
Furthermore, a very spot-on comment, I'm trying to teach them a much more natural language for prompting and less and less need for negatives.
With some care you can generate higher quality images, I'm not underestimating the model, but in the examples I prefer to present images that can be recreated even by novices and with the limitations of web generation.
Thank you very much again for everything, any feedback is essential to allow me to improve my work.
Good at character, dressing, age.
But I notice it lacks the understanding of poses and NSFW things.
No matter what I do with this model I just get black images...
are you using a VAE perhaps?
the model is great, but is there a way of getting rid of the watermarks? I've tried with negative prompts, but it doesn't seem to be enough
Unfortunately not, for now we only need to work on the negatives. I'm trying to improve the image dataset and merge checkpoints to remove them in future releases.
Made to render the perfect female body. One of my fave AI models. Thanks and keep up the great work.
Maybe don't label it "people" as if it does all kinds of people well, if all it really does is women. Thanks.
I apologize for my non-inclusiveness in describing the model's abilities, but I consider the female gender as people, I don't know how you want to judge it. I'm working to improve the model's capabilities in generating different subjects, but it is much more difficult to create male pornography datasets, since it is less varied, and is supplanted by the amount of female pornography data present. managing thousands of data alone is not easy, hope to meet your requirements in future versions. thank you very much for the feedback. have a good time.
The Tensorart link for accessing the model given in the description does not work.
Here is an updated link: https://tensor.art/models/689929450617356570
Also, here is the link to NewRealityXL which is mentioned in the description: https://tensor.art/models/691832056705073836
Thank you very much for the report, I'm doing a rebrand of the various platforms and there may be confusion and errors until I complete it. Thanks again. stay tuned!
Love your work. Always happy to help :)
I can’t replicate these results, any ideas? Tried a1111 and comfyUI.
I just want to say that this model's performance for Asians is too poor, and I hope it can be improved.
Amazingly flexible while keeping high quality and great anatomie on even difficult poses. One of the best! I hope there will be an update some day that makes it even better ;)
I feel like I am getting better results with V2 - Anyone else? Or any tips to better adjust my workflow to new versions?
I tried to train a Lora with this model as base with koyha, but I'm getting errors. Someone knows why or how to fix it ?
Same
@smokezz merge it with the base model and then it works. Of course put only 1% base model in it. It's just there to put it in a finetunable state
Could you please make this model as lightening??
yes, both the lightning and hyper versions will arrive, I just need time, too many things to do! 🤯
What happened to this model that it's not available for download?
I don't know either, I'm investigating, before restoring it I want to confirm that it isn't due to violation of some new policy. Afrodite will be reborn, don't worry.
What Happend to this model? Why it is not available to download? We need this model Back please.
I don't know either, I'm investigating, before restoring it I want to confirm that it isn't due to violation of some new policy. Afrodite will be reborn, don't worry.
Why isn't available anymore?
I don't know either, I'm investigating, before restoring it I want to confirm that it isn't due to violation of some new policy. Afrodite will be reborn, don't worry.
Can you make a version of this for SD 1.5
hi @subhajit28101999224 every day something new comes out and it's hard to keep up, but yes soon I will also publish a lighter version of SD1.5, hopefully as soon as possible. with the advent of Flux now things get even more complicated. I would like to buy 'time' tokens. wait a little longer, stay tuned.
once upon a time there was a model...
months ago i downloaded a really good nsfw model named aphroacronxl_v10. it was a good model and i liked it a lot. one day i went to see if there were any updates to it and it was gone. my heart was broken. but i still had my copy...so i used it heavily. one day for shits and grins i searched the internet for my favorite model. what could have happened to it. after much searching, i finally found it hidden on another site. still the original version. much to my surprise, the 'author/poster' said it was a merge of afroditexl_20Bkdvae and acornIsBoningXL_v10. so i scurried back to my 'safe place' civitai. i then searched for these two models. YAY, again to my surprise these two models still exist. what's better is they both have been updated. so now i'm downloading them both. i'm a happy little camper. once downloaded i will merge them to try and recreate the model i was so fond of. then, all will be well with the universe.
and we all lived happily ever after...
the end
Loved this story! Update us PLEASE! XD
@midnight505050i have created aphroacronxl_v20 with the updated content from both models. to my delight the images/content are equivalent to the original v10 model. after some more testing i will post it for the community to enjoy. it's not a model for the faint of heart... :)
Love this model, are you planning to train a Flux version? I'm sure it would be very popular
It will come, but having FLUX a really raw NSFW dataset will be a bit of a longer job. 🤯😭
Why isn’t this available for the Civitai online generator? Looks interesting.
Unfortunately I am not the one who decides the availability of the online generation, I am organizing myself to try to provide you with an alternative but they are all becoming toxic services, so until a new solution I will be available for download for local use.
Hello, do you have a tutorial video on using this checkpoint or workflow for Comfy UI?
Details
Files
STOIQOAfroditeFLUXXL_XL31.safetensors
Mirrors
STOIQOAfroditexl_XL31.safetensors
STOIQOAfroditexl_XL31.safetensors
STOIQOAfroditexl_XL31.safetensors
STOIQOAfroditeFLUXXL_XL31.safetensors
afroditexlNudePeople_31.safetensors
STOIQOAfroditeFLUXXL_XL31.safetensors
STOIQOAfroditeFLUXXL_XL31.safetensors
afroditexlNudePeople_31.safetensors
STOIQOAfroditeFLUXXL_XL31.safetensors
SDXL1.0_AfroditeXL_31.safetensors
subpar9_sdxl.safetensors
STOIQOAfroditexl_XL31.safetensors
afroditexlNudePeople_31.safetensors
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.