Guidance: 4
Steps: 20
Sampler: ddim
Scheduler: ddim_uniform
Overview
Alternate Lowleg Reality invites you to explore a world where fashion is bold yet tasteful. This model specializes in generating images of women adorned in stylish, skimpy outfits that are daring but respectful. Built upon the Flux Dev model (version 1), it enhances facial features and introduces unique fashion elements, offering a fresh aesthetic for your creative projects.
Features
Enhanced Facial Details: Improved facial representations for more expressive and captivating characters.
Unique Fashion Styles: Generates daring yet classy outfits without nudity, maintaining a respectful portrayal.
Alternative Reality Aesthetic: Transports users to a different world with a specific and immersive style.
Intended Use Cases
Creative Art Projects: Perfect for artists exploring alternative fashion and styles.
Concept Design: Ideal for developing unique characters in games, stories, or visual media.
Visual Storytelling: Enhances narratives set in alternative realities with distinctive fashion elements.
Training Data
Trained on a custom dataset created through a complex workflow, the model utilizes self-generated images from various tools and models. While still in progress, the training data focuses on diversity and uniqueness to continually improve the model's output.
Technical Specifications
Base Model: Flux Dev Model Version 1
Compatibility: Compatible with standard Stable Diffusion setups; no additional dependencies required.
Hardware Requirements: No special hardware beyond standard Stable Diffusion requirements.
Licensing and Usage Terms
All licensing conditions of the Flux Dev model apply to Alternate Lowleg Reality. Please refer to the Flux Dev model's licensing terms for detailed information. The model adheres strictly to these terms, and they cannot be altered.
Version Information
Current Version: 0.1 Pre-Pre-Alpha
Development Status: Early release. Future updates will focus on refining training data and enhancing model performance.
Keywords and Tags
Alternate Reality, Stylish Fashion, Skimpy Outfits, Non-Nude Model, Flux Dev, Creative Art, Unique Aesthetic, Bold Design, Alternative Style
Description
First model with some bugs that needs to be ironed out
FAQ
Comments (5)
a thinned down LoRA version of this would be great for max compatibility, I don't think many of us can handle a 22GB checkpoint
Yes good point. I planned to make a lora of this by splitting the difference of this checkpoint and the base flux dev checkpoint.
I would say the 22gb is great to have, but sounds like a lot of ppl want just the unet/gguf quants/nf4(which I would imagine will go away because gguf is more flexible...but it's slower..maybe it won't go away??) and a smaller fp8 AIO. That's a lot to put out when you make a model....I feel like GGUF Q8 would become the regular seeing that it's very close to fp16 and smaller and less gpu memory but a bit slower... and then you get all the other quants for ppl with various gpu vram....crazy time to be a ai gen nerd! lol
@pychobj2001741 When dealing with GGUF, I tend to gravitate towards Q4 or maybe Q6, because it looks like the quality decrease after Q4 is pretty minimal. Plus, Q4 can be run on 12GB GPUs, so it’s more compatible with lower VRAM setups. I think there’s a clear line between quality and compatibility around there.
You know what would also be really beneficial for CivitAI? If they did a VRAM amount survey, so we could get a better sense of which Q number would be the most practical to use.
@beignets different audience ofc, but on Steam the majority of users are between 8GB and 12GB of VRAM. https://store.steampowered.com/hwsurvey/Steam-Hardware-Software-Survey-Welcome-to-Steam
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.