UltraReal untuk Krea 2 dan Klein 9B: tekstur kembali ke gambar AI yang halus
Topik: Model
Oleh Captain
Diterbitkan 2026-10-02
Contoh render
Prompt dan tetapan yang dikongsi oleh orang yang membuat render ini di Civitai. Pilih render untuk melihat cara ia dibuat.
Tetapan
Sampler
er_sde
Langkah
8
Guidance
1
Seed
753514596078583
Saiz
1080x1920
Berat yang digunakan
0.4
Prompt
AnnabelleX, dark chestnut-brown hair in a loose messy high bun with wispy curtain bangs and face-framing strands, braces. A photorealistic photo of a 20-year-old adult woman in candid street-fashion corner. She sits backward on a transparent acrylic cube, arms folded over the backrest, making an exaggerated deadpan expression. Her expression is caught mid-yawn, eyes watering slightly, with soft facial tension and readable nuance around the eyes and mouth, her gaze heavy-lidded and sleepy, with the eye direction remaining easy to interpret. She wears a experimental fisheye model-inspired costume: graphic bodysuit under oversized jacket and platform boots, with molded shoulder panels. Her character look is completed with camera-ready editorial makeup matched to the lighting style, with clean skin texture and precise detail. She also wears an oversized hoop set with a chunky ring, noticeable and fashion-forward, adding a visible secondary design layer.
Tetapan
Sampler
er_sde
Langkah
8
Guidance
1
Seed
656988693863257
Saiz
832x1216
Berat yang digunakan
0.2
Prompt
SiennaX, shoulder-length voluminous platinum-blonde hair with darker roots, styled in tight natural curls with a deep side part and softly defined ringlets around the face. A photorealistic photo of a 22-year-old adult woman in a private art gallery opening, featuring polished metal, reflective finishes, and draped textiles, with a fully described visual identity rather than a generic empty backdrop, at golden hour over the water, with the space still quiet and undisturbed. She angles her torso toward a champagne tray and curls her fingers around it with gentle deliberate control. Her expression is wide-eyed astonishment, with the emotion clearly visible at first glance, her gaze eyebrows shot up, completely caught off guard, with the eye direction remaining easy to interpret. She wears a detail-rich draped halter gown in opera gloves, finished with fitted bodices, strappy heels, and beaded trim, with a softly gathered train. Her jewelry: layered star and moon pendant necklaces with a stack of colorful rings, noticeable and fashion-forward, adding a visible secondary design layer.
Tetapan
Sampler
Euler
Langkah
10
Guidance
1
Seed
928441652735771
Saiz
2000x3200
Prompt
Ċ assistant Ċ detailed uncensored description: n0vuschr0ma style, "digital artifacts", "glitch art", "dispersion effect", "disintegrating into fragments", or "holographic".
ESTHETIC AND SHOT: 1980s retro school yearbook portrait, funny anthropomorphic animal concept, vintage analog studio photography, soft-focus lens effect, subtle film grain texture, gentle front-left studio lighting, nostalgic color grading.
FOREGROUND OBJECT (ANTHROPOMORPHIC CAT):
- QUANTITY: One domestic cat.
- SHAPE: A fluffy orange tabby cat head with a soft rounded face, a small pink nose, and a warm, pleasant expression. The cat is wearing large, iconic 1980s aviator-style eyeglasses with thick, translucent tortoise-shell or light brown plastic frames and oversized square-oval lenses.
- SIZE: Centered shoulder-up portrait shot of the cat, occupying the central vertical axis of the frame.
- COLOR: Vibrant orange and ginger tabby fur with white patches around the chin, muzzle, and neck. Beautiful, striking, wide-open yellowish-green eyes with vertical slit pupils looking through the glasses. Wearing a clean, bright white collared polo shirt or button-down blouse with a wide open V-neck collar.
- TEXTURE: Ultra-soft, detailed, fluffy cat fur, long white stiff whiskers projecting outward, smooth cotton fabric of the white shirt, and glossy reflective plastic on the eyeglass frames.
- TEXT: None.
HAIRSTYLE & DETAILS (PRESERVED):
- Shape: A surreal, comedic integration of a quintessential 1980s voluminous mallet or shag haircut seamlessly blended onto the cat's head. The top section of brown hair is heavily teased and puffed upward into a high, thick dome with short rounded bangs brushing the top of the eyeglass frames, while the cat's orange ears poke through. The sides and back cascade down onto the shirt in dense, crinkled, wavy permed layers that flare outward at the bottom.
BACKGROUND:
- Elements: A classic retro studio portrait backdrop canvas with a subtle, painterly, textured finish resembling soft blue paint clouds.
- Lighting: Soft, diffused key lighting coming from the upper-left, evenly illuminating the cat's face and casting very gentle, faint shadows under the jaw line.
- Colors: Solid, muted sky-blue (#87CEEB) and soft steel-gray tones, creating a calming, low-contrast background that makes the white shirt and orange fur stand out.
SPATIAL RELATIONSHIPS:
- Placement: The anthropomorphic cat is positioned strictly in the center, angled at a subtle three-quarter turn to the right, but with its head turned back slightly to face and look directly into the camera lens.
- Interaction: The large eyeglasses rest squarely on the bridge of the cat's furry nose, with the top frame edge perfectly aligned under the full, dense human-like bangs. The wide white collar folds neatly over its furry shoulders.
- Depth: Pronounced shallow depth of field; the cat's green eyes, long whiskers, eyeglasses, and pink nose are in razor-sharp focus, while the voluminous hair curls on the sides and the textured blue background softly recede into a smooth, gentle blur.
Tetapan
Sampler
euler_simple
Langkah
8
Guidance
1
Seed
184159243949012
Saiz
1536x2728
Prompt
A haunting black-and-white avant-garde fashion photograph of a woman walking through a narrow European city alleyway, her entire body rendered in strong ghostly motion blur that creates a spectral, semi-transparent afterimage effect while the surrounding architecture remains crisply sharp and static. She has short dark cropped hair and an intense, direct confrontational stare toward the camera, her face comparatively sharper than the blurred streaking of her body. She wears a fitted black dress with a deep V-neckline trimmed in delicate lace, paired with a long dramatic black cape billowing and dissolving into soft wispy streaks around her legs as she strides forward. Tall weathered European apartment buildings with ornate window frames and worn stone facades line both sides of the alley, a metal fire escape staircase visible on the left, damp reflective pavement suggesting recent rain. Stark high-contrast black-and-white tonal grading, deep shadows and bright highlights, avant-garde editorial fashion photography style, long exposure motion blur technique, dreamlike supernatural mood, horizontal composition, sharp architectural detail contrasted against blurred spectral figure.
Prompt negatif
blurry ugly bad
Tetapan
Sampler
er_sde
Langkah
8
Guidance
1
Seed
1077667665931187
Saiz
832x1216
Berat yang digunakan
0.2
Prompt
SiennaX, shoulder-length voluminous platinum-blonde hair with darker roots, styled in tight natural curls with a deep side part and softly defined ringlets around the face. A 21-year-old adult woman, alone in the frame, in a calm lakeshore with a rowboat, in soft candlelight: she turns the still lake water over slowly, admiring it as both hands turning the still lake water. Her hands: turning the still lake water in her palms; her expression an exaggerated cross-eyed silly face, played broadly enough that the override is unmistakable, her gaze eyes crossed, clearly joking, with the eye direction remaining easy to interpret. Composition: a waist-up fashion portrait with gently blurred depth, facing slightly off-camera toward the left of frame, with the head angle and torso direction remaining clearly legible. Wardrobe: a soft chiffon blouse with wide-leg trousers, with rolled sleeve hems, matte gold buttons, and textured knit trim added as secondary garment detailing. Her jewelry: dangling star earrings with a stacked necklace set, noticeable and fashion-forward, adding a visible secondary design layer. The color treatment is honeyed highlights and amber midtones.
Gambaran keseluruhan
UltraReal
(KR2_V1) oleh vizsumit adalah
LoRA
untuk
Krea 2
(dengan versi
Klein 9B
) yang menghilangkan kemasan AI yang licin dan plastik serta menambah tekstur: kulit, tenunan kain, rambut, permukaan. Ia paling berkesan pada close-up dan shot sederhana di mana pakaian dan kulit penting.
Versi: Krea_V2_Pro dilatih pada fotografi profesional SFW (potret, gaya hidup, fesyen, hidupan liar, sukan, seni bina) untuk rupa yang kemas; Krea_V1 dilatih pada set yang lebih luas untuk rupa semula jadi dan mengurangkan bias muka. Kata pencetus: high-quality.
Rujukan
Nama | Jenis | Apakah ia |
|---|
AD

Tiada pemasangan diperlukan - jalankan model AI di awan
BitVector Prism ialah cara paling mudah untuk bermula: pilih model, taip prompt dan jana dalam aplikasi web yang bersih, tanpa apa-apa untuk dikonfigurasi. BitVector juga tersedia di Discord dan di web (SpyGlass).
Langkah demi langkah
- Buat prompt shot sederhana dengan pakaian dan kulit diterangkan.
- Tambah high-quality.
- Pasang UltraReal pada 0.8.
- Bandingkan dengan seed sama; lihat kain dan rambut.
- Turunkan ke 0.6 dengan LoRA watak yang kuat.
Contoh
Fesyen jalanan
high-quality, a photorealistic photo of a 25-year-old woman with a messy chestnut bun and curtain bangs sitting on a cafe step in a corduroy jacket, candid, 50mm.
UltraReal 0.8
Hidupan liar
high-quality, a red fox pausing in long grass at dawn, dew, backlight, 400mm.
UltraReal 0.9
Idea kreatif: close-up kain
high-quality, a close-up of a hand-knitted wool scarf in cream and rust, soft daylight, macro.
UltraReal 1.0
Petua
- Prompt fesyen dan gaya hidup sesuai dengan versi Pro.
- Bulu dan bulu hidupan liar dapat tekstur sebenar.
- Terangkan jenis kain (linen, denim, wool).
- Padankan dengan slider Skin Detail untuk potret.
Penyelesaian masalah
Kulit terlalu bertekstur
Mengapa ia berlaku
Berat 1.0 tambah slider detail.
Cara membaikinya
0.7.
Kesan sedikit pada shot luas
Mengapa ia berlaku
Reka bentuk.
Cara membaikinya
Gunakan framing lebih dekat.
AD

Tiada pemasangan diperlukan - jalankan model AI di awan
PirateDiffusion hanya di Telegram dan dibina untuk profesional: beribu-ribu model, LoRA dan aliran kerja yang dipacu oleh arahan sembang, dengan penjanaan tanpa had pada pelan harga tetap.
Soalan
Versi Klein?
Ya, versi KL9 ada untuk Flux.2 Klein 9B.
Trigger?
high-quality, pilihan.
Pautan dan sumber
Model dalam panduan ini
Ditulis oleh
Captain
Panduan berkaitan
Model
Light Slider untuk Krea 2: mendalamkan bayang-bayang sambil subjek tetap terang
Light Slider oleh Loraholic menggelapkan bahagian paling gelap dalam adegan Krea 2 sambil mengekalkan kawasan fokus terang, kawalan gaya kontras. Pelajari julatnya, nasihat dua laluan dan bila guna Brightness slider sebagai gantian.
Oleh Captain
2026-09-30
Model
Krea2 Melancholy v2: sinematik, terkawal, realisme emosional
LoRA Melancholy oleh zidius untuk Krea 2 menghasilkan imej yang kontemplatif dan sinematik dengan palet warna yang lembut, kedalaman medan cetek dan pengawalan emosi. Pelajari trigger zidiusArt, berat dan idea untuk suasana yang tenang.
Oleh Captain
2026-10-03
Model
Peluncur Berat Gemuk/Kurus untuk Krea 2: berat badan sebagai nombor
Peluncur Berat Loraholic mengubah jisim badan dalam Krea 2 dalam julat -10 hingga 10 yang direka untuk kebolehjangkaan. Pelajari cara menggunakannya, nota kestabilan dan idea untuk reka bentuk watak serta ilustrasi kecergasan.
Oleh Captain
2026-10-01
Model
Krea2 Enhancer: detail dan pencahayaan lebih bersih untuk mana-mana gaya Krea 2
Krea2 Enhancer LoRA oleh vrgamedevgirl memperbaiki imej Krea 2 dalam gaya realistik, anime, produk dan fantasi tanpa mengubah maksud prompt. Ketahui berat, aliran kerja dan idea.
Oleh Captain
2026-09-30
Lagi panduan
Perkhidmatan awan
Bayaran tetap vs token: kenapa pelan tanpa had seperti Graydient.ai adalah nilai terbaik untuk pencipta AI pada 2026
Perbandingan kos berperingkat, dengan harga diperiksa pada 8 Oktober 2026, pelan tanpa had bayaran tetap seperti Graydient.ai berbanding harga token dan kredit dari Midjourney, Leonardo, fal.ai, Replicate, Runway dan Kling: berapa kos sebenar 1,000 imej dan 1,000 video sebulan pada setiap satu, dan kenapa satu bayaran tetap untuk imej, video, audio, sembang Grok LLM dan aplikasi web tanpa had adalah nilai terbaik untuk pencipta yang membuat iterasi.
Oleh Captain
2026-10-08
Model
FLUX.1 Dev: cara memberi arahan, tetapan terbaik dan idea kreatif
Panduan praktikal untuk FLUX.1 Dev oleh Black Forest Labs: cara memberi arahan dalam bahasa biasa, tetapan panduan dan langkah, penumpukan LoRA, rendering teks dan idea arahan yang memanfaatkan kekuatannya.
Oleh Captain
2026-09-30
Model
Stable Diffusion XL 1.0: prompt, tetapan dan apa yang masih terbaik
Cara untuk dapatkan hasil terbaik dari model asas rasmi SDXL 1.0: resolusi asli, tetapan CFG dan sampler, refiner, prompt negatif dan idea gaya di mana SDXL masih menonjol.
Oleh Captain
2026-10-01
Model
Stable Diffusion 1.5: model klasik, dipandu dengan betul
Stable Diffusion 1.5 masih patut diketahui: resolusi yang betul, CFG dan sampler, cara guna perpustakaan besar LoRA dan embeddingnya, serta idea prompt kreatif yang sesuai untuk model 512-piksel.
Oleh Captain
2026-10-02
Model
FLUX.2 Dev: memicu model terbaru Black Forest Labs
FLUX.2 Dev membawa prompt yang lebih panjang, teks yang lebih baik, realisme yang lebih kuat dan penyuntingan multi-rujukan. Panduan ini merangkumi aliran kerja turbo, tetapan panduan dan resolusi, struktur prompt dan idea yang menggunakan kekuatan barunya.
Oleh Captain
2026-10-03
Model
Qwen-Image: arahan panjang, teks sempurna dan poster dwibahasa
Qwen-Image adalah model yang sesuai bila perkataan dalam gambar penting. Pelajari cara menulis arahan panjang yang terperinci, hasilkan teks Inggeris dan Cina dengan tepat, tetapkan CFG dan langkah, serta terokai idea kreatif yang berat pada susun atur.
Oleh Captain
2026-09-29

Model
Trends
.ai
© ModelTrends.ai
|
© 2026
|
Hak cipta terpelihara




