Sampling method：DPM++SDE / DPM++ SDE Karras (Recommend)
AddNet Lora Weight : 0.5 / 0.6
Steps: 40, CFG scale: 8,CodeFormer, Size: 432x768,, Denoising strength: 0.2, Clip skip: 2, ENSD: 31337, Hires upscale: 2, Hires upscaler: SwinIR_4x,
(full body:1.2),ultra-detailed, ultra-detailed, finely detail, 8K, HDR, highres, absurdres:1.2, Kodak portra 400, film grain, (realistic:1.5),ultra-detailed,best quality, masterpiece, solo, (extremely detailed eyes and face),baby face, an extremely delicate and beautiful, beautiful detailed eyes, 8k wallpaper, incredibly absurdres, ((simple background)), solo, brown hair, long hair, (brown eyes), smirk,expressions, beautiful detailed eyes,(full body:1.2), nice hands, perfect hands, <lora:GoodHands-beta2:2.5>
Negative prompt: nsfw, paintings, sketches, (out of frame),lowres, bad anatomy, bad hands, text, error, missing fingers, extra digit, fewer digits, cropped, (worst quality:2), (low quality:2), (normal quality:2), artifacts, signature, watermark, username, blurry, missing arms, long neck, humpbacked, bad feet, nsfw, malformed limbs, poorly drawn hands, mutilated, fused fingers, bad face, bad anatomy, (long body), missing arms, missing limb,logo,cherry blossom, plum blossom, (extra fingers, deformed hands, polydactyl:1.5)
Description
训练数据：
📈 126 images with 10 repeats, equaling 1260 steps.
📉 Divide 1260 steps by 4 batch size to get 315.0 steps per epoch.
🔮 There will be 30 epochs, for around 9450 total training steps
Lora 权重 0.4 ~0.5
FAQ
Comments (1)
作者的人物模型好真实，不知道有没有国内联系方式想请教合作下~
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.