<h2 id="heading-146">Imitation model. Modelled after retired mixed-race Japanese AV star Sofia Takigawa (滝川ソフィア).</h2>Anyone that is into mature type of JAV actresses will definitely know this one.For some context on the actress's background, she is technically half-portugese, half-japanese despite her japanese nationality. This former actress really needs no introduction, good expression, a genuine xxx machine during her prime debut. Figure composition wise, let the LoRA speak for itself. Mixed race LoRAs are fun to make and explore, since you get to explore how the different models weighs different facial dynamics differently!<h3 id="heading-108">~~~~~~~~~~~~~~~~~~~~~~~~~~ Recommended LoRA strength: 1.0 (0.7-1.1 are plausible range dependent on checkpoints; higher values lead to more refiend eyebrows and more "serious" expression; big 0ppai regardless)</h3><h3 id="heading-109"></h3><h2 id="heading-110">Main Prompt:</h2><h3 id="heading-147">sofiatakigawa_jav, breasts, eyelashes,</h3>variations: single hair buns, black hair, tied hairfacial variations: natural makeup, grin, teeth, parted lips**I do not find the need to use the facial prompts, as most checkpoints recreate her partial "duck face" thick lips perfectly.What you can however do, is to use "lip gloss" which will give her pink glossy lips like how she sometimes appear in action.~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~Checkpoints that work well for this LoRA:1) PerfectWorld, or other photoRealistic Hentai models (European model works well for this LoRA). This time, PerfectWorld does very well.2) Realisian. Tend to generate minor-asian bias, which is understandable. the "eyelash" prompt can address that.3) RealisticMix. Similar to Realisian actually.4) ClearRealismMix-Eur, or asian-based/eurasian checkpoints. Not perfect, but still better than pure-European checkpoints.~~~~~~~~~~~~~~~~~~~~~~~~~~~~Note for achieving good resemblance:Use Adetailer for baseline generation (at 512, 768 etc) only, try to control LoRA strength at 0.9-1.0, adjust your denoise accordingly. For this LoRA, recommended to use Adetailer at ~0.4 denoise strength for close-up, and 0.45 to 0.55 for full body.For this LoRA, you dont need to set the prompts for Adetailer separately. I prefer to use just the main prompt, and turn ADetailer on. for good resemblance, you only need to pay attention to 2 things: the shape of the eyebrows, and her thick lips (like real lips).Recommended approach: is to not use HiRes, but upscale it with img2img instead. Though in retrospect, HiRes isn't too bad for this LoRA.Use a denoise strength of 0.25 -0.3 (anything higher will overcorrect face during the process), and turn off Adetailer.~~~~~~~~~~~~~~~~~~~~~~~~~~~~Issues:So far no issues, apart from badHands, certain generation seeds can make her look much younger (once in a while), but I do not see the need to use "mature" prompts to offset that just yet; after-all we want to retain her smooth skin. Actually, there's one issue - in img2img upscaling, sometimes the shape of her eyes can be alittle off - I guess you can try running Adetailer again with very low denoise to try and fix that.Post-production edits/Reproducibility:Some preview images underwent inpainting manually on the clothing part because of the nature of some of the training data. Prompt responsibly.Happy prompting. And do share [SFW] creations~

JAV Sofia_Takigawa

v1. Release based on Epoch #10. Other epoches perform very similar. Trained under max reso 768x768. Experimental, but seems to work just fine. 86 images.

Files with SHA256: a94174e3ad9846441d5f9f484a1bbb9f6d9a52bf72c4fab4c02bca9e0f0e6179