🤩 Are you looking for versatility?? FOUND IT!! 🤩
I hope that «Plus Ultra» bring you something to make you happy.
If, you like the ONE FOR ALL series, please, leave a 👍🏻 and post your masterpiece. Available on YODAYO for generations (click here).
Do you like music?? Lo-Fi and more?? Do you want to help me??
Subs to my YouTube Channel: Chaos Lo-Fi - 〘 Music for Relax, Study, Reading, Inspire, Gaming 〙
The concept behind this model is to allow YOU to have the widest range of features with a single model. I'm happy to present the XL model of the ONE FOR ALL series!!!
My intention is to offer a ONE FOR ALL solution, so that you can save time and disk space. No, you didn't read it wrong, a single model, for ALL your creations, this is ONE FOR ALL «Plus Ultra»!!!
☕ I LOVE COFEE!! Want to buy me a cup?? ☕
This model is specialized in:
EVERYTHING!! 😱🤩😃
Thanks to the versatility of Stable Diffusion XL technology, it is now possible to create a model capable of generating great results in 2D, 2.5D, 3D, Anime, Furry and Realistic.
BUT THERE'S A LIMIT!
Your prompt!!
The XL model works differently from the SD 1.5 we're so used to, so we're all learning.
Creating amazing art in XL requires amazing prompts, so be patient with the learning curve as the technology is still in its infancy and we have a lot to learn and evolve.
Recommendations:
Recommended VAE: SharpSpectrumVAEXL - v1 | Stable Diffusion VAE | Civitai
Generation Recommendations/Tips:
Sampling method (Final version shall work with almost all samplings):
DPM++ 3M SDE Exponential;
DPM++ 2M SDE Karras.
Euler
Sampling steps:
30 (The XL works differently than the SD 1.5. So the recommendation is to stay between 20 and 30, but I confess I work with 50 too).
Width/Height:
Here's the big difference. If the SD 1.5 works with 512 x 512 base, the XL works with 1024 x 1024. I personally use the 1024 x 1440 resolution a lot.
CFG Scale:
I use 8 as default.
PAY ATTENTION TO THE NEGATIVE PROMPT!!
As I said before, XL works differently from SD 1.5.
So start your generation without the negative prompt. That's right, start WITHOUT NEGATIVE PROMPTS. Negative prompts are very powerful in XL, so only use them when you're not getting the result you want. Don't fill up on negative prompts as many people mistakenly do in SD 1.5.
Anything beyond that depends on your taste and the results you want. 🥰
You can use my generations for reference (click here).
Thank You!! 💓💓💓
This model is only possible because there are people with incredible skills and effort. My intention at no time is to belittle any other model, on the contrary, ONE FOR ALL is an ode to these incredible people, without whom this model would not exist. Thank you very much!
If you happen to like the ONE FOR ALL series, please leave a good review and post your creations, I'll love to get to know the amazing universe you'll create! 🤩🤩🤩
If you want to thank me, like every Brazilian I love coffee, another way to encourage or thank me is to follow me on Instagram. 🥰🥰🥰
If you have any difficulties or questions, or just want to have a chat, you can contact me on Instagram or here on Civitai.
Last statements:
Please use this model for fun or work, but respect the legal regulations in force in your country and international treaties, especially regarding impersonation and pedophilia. I declare myself exempt from any violations resulting from the use of any model from the ONE FOR ALL series or its base models, being the creator of the content solely responsible.
This checkpoint (ONE FOR ALL «Plus Ultra») is not available for use outside Civitai on external generators. If you find this template running on any generation site or available for download outside Civitai, please contact me.
If you are responsible for an online generation site and would like to use any model from the ONE FOR ALL series, please contact me.
Finally, all images generated using any ONE FOR ALL model belong to their creators (user), as well as the use-fruit and responsibilities for the content created. This means, but is not limited to, the right to sell the images generated, for example.
The use of this checkpoint is conditional on acceptance of these terms.
(SP, BR, 20/01/2024)
Description
About Final Version
It took me two months to get something that I liked and that made me feel that there had been some development in relation to version 3. Sorry, it may not be the best version, however, I'm so tired and today was a worst day. If I come to the conclusion, this version isn't good at all, I'll delete it. Thank you for understanding.
Recommended VAE: SharpSpectrumVAEXL - v1 | Stable Diffusion VAE | Civitai
This led me to two conclusions:
1 - Version 3 is, without false modesty, VERY GOOD;
2 - As with the SD1 versions, I believe that this version of SDXL has reached the limit of what I can develop.
These two factors, together with the popularisation of the Pony model and the recent explosion of Flux, have led me to the sad conclusion that the time has come for the final version of PLUS ULTRA to be presented.
Just like the previous version, the «Plus Ultra» model is an SDXL 1.0 model with DPO. However, given the popularity of Pony models, I took the step of adapting the checkpoint to understand Pony tags and reproduce similar results. However, please note that this is not a Pony model in essence, so some prompts may not generate the intended result. Also, compatibility with LoRAs Pony tends to be perfect, so feel free to use LoRAs for Pony.
Final version, it took 2 months of training and daily adjustments, no kidding, I lost count of how many versions I tested and how many images and how many times I had to clean my SSD to have space until I got the final result.
AND YES!!! THIS TIME WE HAVE 🥒🍆🥕🪵🌡️!! LOL
One complaint with version 3 was that the NSFW wasn't pornographic. This time, I hope that I have corrected it.
However, I warn you, for more realistic results, I recommend Skip Clip 1 and CFG 4. Adjusting Anime and Realistic to a more intense degree is very complex, so this was the best solution I could find.
About DPO (Direct Preference Optimization).
DPO fine-tune on datasets scored as highly visually appealing by an aesthetics classifier, to bias the model to visually appealing generations, while maintaining distributional guarantees and improving generic text-image alignment in addition to visual appeal. The training is via gradient descent, automatically validate checkpoints with the 500 unique prompts of the Pick-a-Pic validation set: measuring median PickScore reward of generated images. DPO scheme implicitly learns the differences in between images by taking an expectation over the inner term of the prompt.
So, DPO pay more attention on your prompt and try to generate a better image from it.
I'm very grateful to the authors of the Checkpoints and LoRAs used in the ONE FOR ALL series and I recommend everyone to visit them and appreciate their work (if your name isn't here, it's because any of your assets were been used):
#
A
B
C
D
E
F
G
H
I
J
K
L
M
N
O
P
R
S
T
V
W
X
Y
Z
And to all those who in some way collaborated or encouraged this work, I affectionately offer this simple but sincere Special Thanks:
Special Thanks to those creators:
Special Thanks to my amazing followers:
AND YOU WHO IS AMAZING TO READ SO FAR HERE!! WOW!! 😱🤩💓
YOU ARE REALLY AMAZING!! THANK YOU VERY MUCH!! 🥰🥰🥰
FAQ
Comments (11)
This model is incredible. I have always loved all of your models, since the 1.5 series you put out, but this is just excellent.
As a side note, I know people are getting hype about Flux and Pony, but at the same time, these things all have strengths and weaknesses. There are some things that 1.5 seems like it will probably still do better, and there are definitely huge advantages with XL over Pony.
In any case, this is a superlative model, as always from you. I have always treasured your work, and it's likely I always will. You are the true ally of artists and creators.
I am enjoying the model a great deal, but I seem to have hit a snag. For some reason, it's just not giving me flaccid penises anymore. It keeps wanting to give me erections everywhere, or GIANT dicks, and I can't even stop it by putting "erection" in the negative prompt. Any suggestions?
LoL
Sry, it was really unexpected, hmmm, I think I didn't put any flacid penis on the trainning, that's why, so we could try to trigger it from a merged model. I'll take a look at the models prompts and make some test and I'll give an answer asap.
Thank you.
Sry, I think it just can't. 😅
I never thought about the possibility of relaxed images, which makes perfect sense, but it's something I've overlooked. Even using LoRA, I ended up generating guys with two penises, one relaxed and one erect, it was bizarre to say the least. I'll try a few more things, but I don't think it'll come to anything, I'm sorry. 😅
The weight of trained images are too heavy, and even if the recipe models can do that, the trained model get priority over the others, so it generate the big huge penis or duplicate. So, I think it's die hard or nothing. 🤣🤣🤣
Btw, do you have any model that you use for that?? If the owner of model allow to merge, I can make a small fix and release a fixed version.
@ChaosExperience Hello! I'm very grateful for your help, sorry it took so long to answer. I don't have any XL models that I use for a reliable source of flaccid images, unfortunately. There are some really nice 1.5 models like the Tastefully Nude series that incorporate a robust functionality for less giant ones, but I haven't started to work really in-depth with XL until relatively recently. I'll do some looking and see what I can find!
I still super love this model though, it does so many things so well and is really super attentive to prompts! It's invaluable beyond description, I would just love to be able to do everything with it lol!
@hushicho I'm uploading something else, please wait.
@ChaosExperience Ah, you're amazing! I'm so sorry to hear about the difficult times you've experienced recently. It seems like this year has been so monstrous for most people I know. I hope for you to be able to feel better quickly, and I'm so sorry for your loss. I will look forward, very much, to trying this new model you've posted. Thank you so much for sharing it with all of us, your work is always so enjoyable.
@ChaosExperience I tried the new one, but it's much harder to work with and doesn't seem to pay nearly as much attention to the prompts. Unfortunately, it's uneven at best with regards to anatomy. But thank you anyway for the effort, it is appreciated.
@hushicho Oh, lol... Well, it's weird, but yeah, it depends a lot of the prompt and the samplers too. Anyway, if the Final version fits better, then, that's it. At least one version will work.. 😄😄😄
@ChaosExperience It is an improvement on some other results though, so it's definitely improved in some things! Honestly I think it could be a perfect all-in-one solution if you kept developing it, but that's of course your call. Thank you again for the attempt!
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.