HomoFidelis

<h3 style="text-align: start;"><em><strong>Fidēlis = faithful, loyal, true, trustworthy, dependable</strong></em></h3><p style="text-align: start;"><br></p><p style="text-align: start;"><strong>Note:</strong> Preview images are generated without the use of LORAs.</p><p style="text-align: start;"><br></p><h3 style="text-align: start;"><strong>About</strong></h3><p style="text-align: start;">This model is aimed for photorealism but at the same time versatile and flexible with its generations. When I first made this model, I didn't really have any intention of releasing it but since I got so many request for it, I'm doing it anyway. The original base model I made was alright, it was realistic enough. But I wanted to add more versatility and diversity into it so I merge it with some of my favorite models that I've used before..</p><p style="text-align: start;"><br></p><h3 style="text-align: start;"><strong>My Settings</strong></h3><p style="text-align: start;">These are just some examples of the settings I usually use and it's not really a hard requirement so you're free to experiment and see which setting works best for you:</p><ul><li style="text-align: start;">Sampling Method: DPM++ 2M Karras or DPM++ SDE Karras</li><li style="text-align: start;">Sampling Steps: 25 - 50</li><li style="text-align: start;">Hires. fix: On</li><li style="text-align: start;">Upscaler: ESRGAN 4x+ or 4x_NMKD-Siax_200k or 8x_NMKD-Superscale_150000_G (if you have it)</li><li style="text-align: start;">Hires Steps: I usually put it at 0 so it'll be the same as Sampling Steps</li><li style="text-align: start;">Denoising Strength: 0.25 - 0.35</li><li style="text-align: start;">Upscale: 1.5 - 2</li><li style="text-align: start;">CFG Scale: 4.5 - 8</li></ul><p style="text-align: start;"><br></p><h3 style="text-align: start;"><strong>Extra Tips</strong></h3><ol><li style="text-align: start;">Try to keep the prompts short and concise. No need to fill up the prompts with unnecessary wording that does the same thing like high detail, ultra detail, absurdres, hires, photorealistic, hyperrealistic, etc. Keep it simple.</li><li style="text-align: start;">ADetailer is key if you want to get good looking faces. If you're generating a close up shot or portrait, most of the time you won't need it. But if its a full body or far shot, ADetailer is absolutely necessary unless you wanna get distorted or blurry looking faces.</li><li style="text-align: start;">Hires fix is highly recommended, especially for far and wide shots where a lot of details are getting lost. And if you're using LORAs, it is mandatory to use it because without it, the output might look a little blurry and overcooked.</li><li style="text-align: start;">Use StableSR extension in your webui if you have a low VRAM and can't use Hires Fix without getting errors. It's slower, but it allows you to upscale even higher than x2.</li><li style="text-align: start;">Use negative embeddings instead if writing paragraphs of words into the negative prompts. Avoid using too much at once. Here are some examples of the ones I usually use:BadDream &amp; UnrealisticDreamorCyberRealistic Negative</li></ol><p style="text-align: start;"><br></p><h3 style="text-align: start;"><strong>Credits</strong></h3><p style="text-align: start;">Models included in the merge are listed below. Credits to the original creators, go check them out:</p><ul><li style="text-align: start;">HomoFidelis_v3</li><li style="text-align: start;">Dreamshaper v8</li><li style="text-align: start;">Virile Reality v3</li><li style="text-align: start;">Juggernaut</li><li style="text-align: start;">Unstable Homoerotic Diffusion (UHD) v1.11</li><li style="text-align: start;">Photon</li></ul><p style="text-align: start;"><br></p><p style="text-align: start;"><br></p><h3 style="text-align: start;"><strong>v4.0</strong></h3><p style="text-align: start;"><strong>What's new:</strong></p><ul><li style="text-align: start;">Overall improvement</li><li style="text-align: start;">Rebuild using v3.0 as the base model</li><li style="text-align: start;">Slightly tweaked the weights of some loras used</li><li style="text-align: start;">adjusted the realism to allow for more creative results</li></ul><p style="text-align: start;"><br></p><p style="text-align: start;"><br></p><h3 style="text-align: start;"><strong>v3.0</strong></h3><p style="text-align: start;"><strong>What's new:</strong></p><ul><li style="text-align: start;">Overall improvement</li><li style="text-align: start;">Better penis generation</li><li style="text-align: start;">Better control over body type and size</li><li style="text-align: start;">Added versatility to generate other types of creative media</li><li style="text-align: start;">Slightly tweaked color grading</li><li style="text-align: start;">Slightly reduced bias towards NSFW generation</li></ul><p style="text-align: start;"><br></p><h3 style="text-align: start;"><strong>v2.0</strong></h3><p style="text-align: start;"><strong>What's new:</strong></p><ul><li style="text-align: start;">Slightly more photorealistic without sacrificing too much of its versatility</li><li style="text-align: start;">Easier to generate different ethnicity and race</li><li style="text-align: start;">More diverse body types (skinny and slender is still a little harder to generate without muscles, but I will work on that in the next update)</li><li style="text-align: start;">Penises are a lot better now, but prompting for specific penis types (erect, flaccid, etc) and size is still a little wonky. It works fine if your character is completely nude, but if clothes are involved, its basically random and often times it'll default to flaccid.</li><li style="text-align: start;">Slightly improved armpit hair and anus</li><li style="text-align: start;">Can generate a few basic faces of mainstream actors and celebrities without the need of LORAs</li></ul>

Files with SHA256: 5b0f7a6e765297a722f396cc8f29ce90d456b89bae01fef3305b7e23d040b740