Brought to you by Cold BeyondKeywords:JetAugmentedHuman, mask, leather outfit, a pair of sunglasses, futuristic suit, lacy collar, off the shoulders jacket, sleeves pushed up, red pants, belt, white bootsSome Keywords that seem like useless: black gun, black purple gun, ringsFor Generating Images:Do my technique, use 512,512 or 512,768 or 768,512 and set batch size to 4.If you like the results, copy the seed number and paste it to seed.And open HiResFix and ADetailer for Face and Hand. And then generate again. If necessary use Impainting to fix issues, inside Img2Img this time.Then lastly use Tile Resemple in ControlNet and upscale the image.You can check this tutorial for Upscaling with Tile Resample:<div data-youtube-video><iframe width="640" height="480" allowfullscreen="true" autoplay="false" disablekbcontrols="false" enableiframeapi="false" endtime="0" ivloadpolicy="0" loop="false" modestbranding="false" origin playlist src="https://www.youtube.com/embed/yv4J4orS-SY?si=gAgF-4fLf37vWh1K" start="0"></iframe></div>This is my first ever LORA. I bought the Augmented Human Pack (Vex Skin) for Jet in Modern Warfare 3. I wanted to make her AI arts. Seeing no LORAs of her, I decided to make my own. Excuse me if the LORA is not good enough. But there were no photos of her with this skin. So I had to take images inside the game. If I had better quality images, it woulda been much better. But still it does a decent job :)The 4 tutorials that were helpful to me a lot for learning LORA Training:<div data-youtube-video><iframe width="640" height="480" allowfullscreen="true" autoplay="false" disablekbcontrols="false" enableiframeapi="false" endtime="0" ivloadpolicy="0" loop="false" modestbranding="false" origin playlist src="https://www.youtube.com/embed/9MT1n97ITaE" start="0"></iframe></div><div data-youtube-video><iframe width="640" height="480" allowfullscreen="true" autoplay="false" disablekbcontrols="false" enableiframeapi="false" endtime="0" ivloadpolicy="0" loop="false" modestbranding="false" origin playlist src="https://www.youtube.com/embed/70H03cv57-o" start="0"></iframe></div><div data-youtube-video><iframe width="640" height="480" allowfullscreen="true" autoplay="false" disablekbcontrols="false" enableiframeapi="false" endtime="0" ivloadpolicy="0" loop="false" modestbranding="false" origin playlist src="https://www.youtube.com/embed/PadSEwA3xiU" start="0"></iframe></div><div data-youtube-video><iframe width="640" height="480" allowfullscreen="true" autoplay="false" disablekbcontrols="false" enableiframeapi="false" endtime="0" ivloadpolicy="0" loop="false" modestbranding="false" origin playlist src="https://www.youtube.com/embed/xXNr9mrdV7s?si=0e_4iwd8V0NcsB5l" start="0"></iframe></div>A Few Notes:<ul><li>The model was trained using Dreamshaper 8, </li></ul>I set v2 / v_parameterization / SDXL off, because it was the only way. It seems the Dreamshaper 8 was based on SD V 1.5<ul><li>The model is fp16.</li><li>I only made safetensors version.</li><li>Train Batch Size 2, Epoch 20 (I choosed 15th Epoch between the 20 of them. You can see their differences between them in the image I posted)</li><li>While training, I set the size to 1024,1024 while the people are setting it to 768,768 when it's originally 512,512. Now setting it 768,768 is better than the default 512. But I'm not so sure about 1024. </li></ul>But this is for the training, so it shouldn't affect your image generation.Nevertheless I'll make a newer version with 768,768.IF YOU WANNA SUPPORT ME, PLEASE SUBSCRIBE TO MY YOUTUBE CHANNEL ^^<a target="_blank" rel="ugc" href="https://www.youtube.com/channel/UCQnUY0G3HJooPT1V3DgvYLw">https://www.youtube.com/channel/UCQnUY0G3HJooPT1V3DgvYLw</a>

Jet Augmented Human Skin - Call of Duty

Brought to you by Cold BeyondKeywords:JetAugmentedHumanv2, mask, leather outfit, a pair of sunglasses, lacy collar, off the shoulders jacket, sleeves pushed up, red pants, belt, rings on fingers, white bootsFor guns: holding a black purple gun, holding a black gun, handguns in holsters<ul><li>I used some fancy prompts for fun (outside of training) You can check them out from the prompts of my posted images with this LORA.</li></ul><ul><li>I personally didn't use gun prompts for generating images (outside of training) because Stable Diffusion is terrible at handling hands, holding guns. But there are good LORA models for guns or swords. Maybe you should combine 🤔</li></ul><ul><li>For some reason, this LORA tends to make the background darker, even though I deleted background of each image. Lower the rate for more freedom, such as:0.7 or 0.6 instead of 1</li></ul>Version 2 is here! I tried to improve this one as much as I can. But since this skin/character doesn't have images, or arts/concept arts; I tried my best with the images I took in-game.Training Details:Based on RevAnimated model in this version.Max Resolution is 768,768Clip Skip 2 this time (To make it look a little more anime/soft)Dimensions 32 as I rememberAlpha 8, for more freedom.

Files with SHA256: 1628ffbb9f99d538d1e3f6cb85f54fb3656874b3e32dcd379d516870deca56b1