For business inquires, commercial licensing, consultation and large scale image captioning for datasets contact me under [email protected], for custom LoRAs visit my ko-fi page.
Checkpoint Type: SDXL/Cartoon/General Use/Evolving/Project
Follow me on Twitter: @YamerOfficial Discord: yamer_ai
If you have enjoyed this checkpoint:
Perfect Design UNLIMITED VOID, a family of checkpoints that aims to being an improvement over the base SDXL 1.0 model, this can be considered as a side project of mine; This is a general purpose model that will try to give you some extraordinary results, both in quality and style, unfortunately the V1 and V2 versions are not NSFW friendly, V3 is better but can be improved, version 4 can handle NSFW prompts better, v5 and v6 both don't aim at nsfw but more at a sfw aspect and uniqueness to their style.
--> REDESTRIBUTION OF THIS MODEL TO PLATFORMS OUTSIDE CIVITAI AND SELLING IMAGES WITHOUT MY WRITTEN CONSENT IS FORBIDDEN! <---
You can contact me on Discord: (yamer_ai), Twitter: @YamerOfficial, email: [email protected] or directly using the Civitai chat system for my approval.
This version signifies an advancement beyond the base SDXL 1.0 model, notable for its considerable style variability. V3 addresses some errors associated with anime finger creation, while V4 demonstrates enhanced proficiency in handling not only hands but also various other generation errors. Although a general trend of 2.5 images emerges during the generation process, it is important to acknowledge that this trajectory can be influenced by the specified styles in prompts or the inclusion of artist names, V5 is great when it comes to more vibrant images while v6 is the best with everything absurd and unique.
For all users of this model, your enjoyment and constructive feedback are of utmost importance. Your engagement adds to the collaborative spirit that fuels this project, as you explore the model, your feedback will play a crucial role in its evolution.
The essence of the "Perfect Design" checkpoints lies in its dedication to excellence. Created through meticulous training and merging, I will try to update this model the sooner I can to keep up with other model that aim at absolute dominance in this field of AI art.
As you journey forth with it, I invite you to fully embrace its potential. Unleash your creativity, experiment, and witness the results. Your interactions will shape its trajectory and the course of AI artistry itself, piece by piece everyone can contribute to the advancement of stable diffusion checkpoints.
P͓̽e͓̽r͓̽s͓̽o͓̽n͓̽a͓̽l͓̽ ͓̽R͓̽a͓̽t͓̽i͓̽n͓̽g͓̽:͓̽
Quality: 9/10
Stability (Variety): 7.5/10
Flexibility: 7/10
CFG resistance: 20/30
Style: 8/10
Style Variety: 8/10
Recommended settings:
Image Quality: 1024x1024, 9:16, 1:1, 4:3
Steps: 10-100
CFG: 7-20
Fields where this model is better than regular SDXL1.0:
No embedding is needed
The style is much more variable
No refiner is needed
Less AI generated look to the image. (The plastic look is reduced, especially with specific styles)
No face fix needed (There is no need, faces already look beautiful)
Better saturation, overall quality, texture and tonality
Less blurry images or corrupted images
No adetailer required (Just use hires fix)
1) I was convinced that I had achieved a perfect design, grotesque and pristine.
2) All for nothing at all.
3) Imagination is more important than knowledge.
-Mason Lindorth -Miracle Musical -Albert EinsteinVersion 4 Update Notes:
This model has more vibrant colors and is capable of generating abstract concepts, it is truly an improvement over V3 in quality and uniqueness!
This version is more advanced, colors are more vibrant and composition have better uniqueness to them and NSFW looks overall better compared to version 3.
Version 5 Update Notes:
This version was uploaded during the holiday season and it's a special gift for the community, something colorful to play with that gives great vibes, it's great when it comes to absurd colors.
Version 6 Update Notes:
This version is the best for everything absurd, colors, saturation and quality are as always great but they explore even more weird fields, some double head image generation issues in v5 have been fixed.
Description
FAQ
Comments (6)
You said a lot of words in the description, yet somehow didn't say anything about what the point of this checkpoint is.
Is this checkpoint meant to be an improvement on the base model?
Is this checkpoint meant to replace the refiner?
Is this checkpoint a general purpose checkpoint that just has a different flavor?
Is this checkpoint meant to specialize the model so it's better for 2.5D?
Is this checkpoint meant to be better for anime?
Is this checkpoint meant to be better for nsfw?
etc. etc.
1) Yeah pretty much an improvement or I would not upload it and call it "PERFECT DESIGN", the point is literally in the name, the reason why I chose that name is because the model will try to give you good results and try to be the best model available.
2) Refiner? You decide not me, I only make model that I think are good, you can compare them to other models and give me your feedback on that, I just try to give the community the best I can.
3) If you knew anything about this model you would not even ask this question, the answer is the first line of the description;
4) Read the first line of the description to know the capability and there is a rating section to better describe how well the model handles different styles, aka 7.5/10 (very good).
5) For the third time, and now I will give you the answer, "everything" means everything, by using a bit of logic in the description you can even read that this model is both trained and merged, and expecting a merged model to only give you 1 style or very few change in the outcome demonstrates how little you know about models.
6) General use model, it will give you good results in any style and paired with a LoRA will give you even better results in the style you need.
Thank you for sharing your opinion when you clearly haven't even used the model and judjed the description only, if you think there are areas of the model that don't give you the result you expect or any other type of area the model can improve let me know and if you have already used it, remember to share a fair description of the model with images.
@Yamer you're being very defensive here, I'm trying to get an understanding of a model before I download it and invest time in experimenting. There are too many resources available to try everything and they take up a ton of space. Literally the only takeaway from your description is that this model is your attempt at 'perfect'. The spam of questions in my comment was meant as an illustration of how you had no useful information in your description, because I couldn't answer any of them unless I literally download the model and try for myself.. which is not the point of having a description on a model for download!
I appreciate you taking the time to answer though! Maybe you should consider less fluffy BS in your description and more about the intent of the model. Your arbitrary rating system is not helpful without a reference. I appreciate the effort you put into getting out some different models though and I'll give this a try.
@CallMeMaybe cool mate, I will update the description as soon as possible :3
@CallMeMaybe updated
@Yamer Whoa, thank you so much, your description is super helpful now!
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.