This series is focused on women and the female body, in every kind of setting or style. It can do fine art, social media style, surrealism, cinematic, and pornographic very well. The main goal is to have as much diversity as possible in an SDXL checkpoint. Diversity in ethnicity, body type, faces, posing, personality, and locations.
If you like it post your creations!
Donations to my Whop account will exclusively go towards hardware upgrades. Civitai Buzz will be used for site generation and promotions.
Upgrading my local hardware is my on-going goal so I can train locally.
Local training is imperative today. With Ai data centers there will be a vast drain on resources in the areas they operate. Since model & lora training will no doubt be hosted in the new centers, I want to be able to run locally only. The moment my Ai hobby causes actual harm, on any scale, is when I start collecting stamps instead.
General Prompt Guide
Weights in my prompts are primarily for testing the model's capabilities and trying to uncover limitations during test generation. However, I've created the tutorial below to help understand how model weights can be used more effectively.
Positive Attention
Weights between 0.5 & 2.0 are usually sufficient.
I sometimes use weights between 2.0 & 3.0 when I want to force attention to a specific ethnicity for testing.
Higher weights are also helpful when you want to magnify size or specific details. The difference between: large breasts, (large breasts:1.2), & (large breasts:1.8) is obvious.
Recognized weights are usually -5.0 to 5.0 with Dynamic Prompts.
Attention Equation
The parentheses ( ) on their own adds attention weight of 1.1. Unless you specify a weight.
(hyper-realistic) = 1.1
((hyper-realistic)) = 1.21
((hyper-realistic:1.3)) = 1.43
Brackets [ ] are negative attention weight of 0.91, not negative weight, it's just attention that starts under the 1.0 default of all words without weights. Stacking them decreases that number.
[deep shadows] = 0.91
[[deep shadows]] = 0.81
[[[deep shadows]]] = 0.72
Negative Attention
Weights over 3.5 are not usually effective or are a detriment during tokenization.
Negative weights do not have much effect in prompts.
Using too many high weights in the same prompt can cause distortions & deformities.
Using high weights will give specific keywords more attention, but that is often at the expense of other keywords.
Bonus
Prompt with high weights:
(HDR:2.2), cinematic, (natural sunlight:1.23), (hyper-realistic:1.45), (full body shot:0.6), masterpiece, incredible details, [feeling of solitude], mysterious, (deep shadows:0.72), Beautiful (30yo Malaysian woman:2.52), hazel hair, (mid_length grungy hairstyle :1.43), (natural large breasts:1.6), (wearing an unbuttoned stylish silk shirt), (brown pearl g-string, leather lace sandals:1.41), (sexy pose, reclining slightly on elbow:1.68) on a stunning designer wicker couch, at a rustic beach house rooftop with driftwood furniture, seashell decor, a view of the ocean, perfect for casual summer parties, sensual atmosphere, beautiful, guilty pleasure, shot on Canon EOS 5D, RAW.
Prompt without high weights:
(HDR:0.86), cinematic, (natural sunlight:0.8), (hyper-realistic), (full body shot:0.6), masterpiece, incredible details, feeling of solitude, mysterious, (deep shadows:0.72), Beautiful (30yo Malaysian woman:1.32), hazel hair, (mid_length grungy hairstyle), (natural large breasts), (wearing an unbuttoned stylish silk shirt), (brown pearl g-string, leather lace sandals), (sexy pose, reclining slightly on elbow:1.12) on a stunning designer wicker couch, at a rustic beach house rooftop with driftwood furniture, seashell decor, a view of the ocean, perfect for casual summer parties, sensual atmosphere, beautiful, guilty pleasure, shot on Canon EOS 5D, RAW.
To force clip & vae in FP32
Forge:
./webui.sh --clip-in-fp32 --vae-in-fp32 --no-half-vae(or add the commands to your webui-user.bat (Windows) or webui-user.sh (Linux) in
COMMANDLINE_ARGS= --clip-in-fp32 --vae-in-fp32 --no-half-vaeFor all commands (many are obsolete now since the developers have made them default)
Comfy:
python main.py --fp32-text-enc --fp32-vaeI don't use Comfy, so this is what I've seen on the web. Here is a link to all commands:
https://gist.github.com/ravetank/cedcd1ea89738b2e79b9f01e45a1a03d#file-comfyuihelp-txt
Description
VAE is baked in for both versionsThese are the settings I use for this version.
faceSmash
Eular a SGM Uniform
30 steps
CFG: 5-6
Hi-res Fix
4x Foolhardy Remacri ExtraSmoother
8-15 steps
CFG: 5
Denoise: .4 to .5
V8 fp16
Internal encoders have been replaced with Simulacrum clip-L & clip-G and converted to fp16
https://civitai.com/models/1044804/clip-l-and-clip-g-full-fp32-zer0int-and-simulacrum
Forge:
./webui.sh --clip-in-fp32 --vae-in-fp32 --no-half-vae(or add the commands to your webui-user.bat (Windows) or webui-user.sh (Linux) in
COMMANDLINE_ARGS= --clip-in-fp32 --vae-in-fp32 --no-half-vaeComfy:
python main.py --fp32-text-enc --fp32-vaeI don't use Comfy, so this is what I've seen on the web. Here is a link to all commands:
https://gist.github.com/ravetank/cedcd1ea89738b2e79b9f01e45a1a03d#file-comfyuihelp-txt
FAQ
Comments (7)
(Full Copy) Prompts do not reproduce images
I have my Forge set to produce seeds with CPU, not the default GPU. However, I've never seen the need to reproduce images exactly. So I don't keep a record of seeds used before Adetailer changes them (when I need to use it). Although, if the prompt itself gives wildly different results then that is something I need to investigate.
@ironically_irascible
That do not explain that the images do not even closely resemble the examples. Reproducing prompts is important to validate that ones setup works correctly. If you can't reproduce the prompt and the image quality is 1/10th of the example, then there is no way to figure out whats causing it as there are endless amounts of parameters.
But if you don't assure the prompt reproduces (doesnt need be pixelperfect but 98% to the example with same level of quality), you can't start making stuff with it as the quality is all over toe board then
@TScitai What version are you using? The uploaded models are the same versions I used to make the images. I use them for at least a week after uploading. So if I download the version you're talking about and get different images entirely, then I need to contact Civitai. I can also replace the model with the original.
Also, CPU seeds will produce very different images than GPU across different computers. From the Forge Settings:
"Random number generator source. (changes seeds drastically; use CPU to produce the same picture across different videocard vendors; use NV to produce same picture as on NVidia videocards)"
@ironically_irascible In general the generating UI should put all parameters into the image itself, inclusing CPU/GPU/NVidia GPU seeding, VAE etc. Im using different UIs, mostly Automatic1111 due to the number of working extensions for it, ComfyUI for some special workflows and sdnext.
Its sadly quite a common occurance on checkpoints that prompts do not reproduce the image, which is really a pain in the ass as it often results in greatly different image quality
@TScitai I agree with the last sentence, for sure. I ignore seeds for the most part. But the quality is required. Which is why I tested both versions of v8 and concluded that fp32 model wasn't what I thought I was uploading. I can't figure out what changed from the version I tested and the version uploaded.
I very much appreciate you pointing the discrepancy out to me. I never would have discovered it myself since it didn't occur to me that the model could deviate from the tested version. CPU vs GPU does change seeds "drastically", but it certainly can't explain why the preview images were so different. It was probably an error on my part, rather than something changing after uploading. So I am definitely going to be more attentive than I was before.
V2 has been the best realistic XL model on all of Civ for a while now, and I'd say V8 is better, but only slightly.
It's "sensitive", so you're going to want to keep the CFG a little lower, probably at 4 or less. It also tries to put pussy's on everythying. EVERYTHING. Like I've had pussy's show up on a dude's thigh...
Speaking of dudes, it can struggle with male anatomy in more complex scenes. It doesn't do gay scenes very consistently, will try to make everyone a futa or some abomination human centipede.
It doesn't do distance well (no models do really) but close ups of faces are amazingly well rendered.
It also skews young so, be careful...
Use terms like "adult woman" not "girl" or you'll look like a creep.
Looks like we don't have an active mirror for this file right now.
CivArchive is a community-maintained index — we catalog mirrors that volunteers upload to HuggingFace, torrents, and other public hosts. Looks like no one has uploaded a copy of this file yet.
Some files do get recovered over time through contributions. If you're looking for this one, feel free to ask in Discord, or help preserve it if you have a copy.