


7 months ago
low-poly style Swedish , 30 years old, woman, blonde, long hair, Black mesh mini but not see-through, long- heeled shoes, slim, perfect neck and arms Veck, sTender, DOF, Aperture, insanely detailed and intricate, character, hypermaximalist, hyper realistic, super detailed , samplers DPM++ 2M SDE steps 35 , Cfg 3. 5-4. 5, 4x-UltraSharp upscale 1. 5 steps 35 denoising strength 0. 34 Style cinematic ,<lora:add-detail-xl:1> . low-poly game art, polygon mesh, jagged, blocky, wireframe edges, centered composition

8 months ago
Swedish , 30 years old, woman, blonde, long hair, Black mesh mini but not see-through, long- heeled shoes, slim, , perfect neck and arms Veck, sTender, DOF, Aperture, insanely detailed and intricate, character, hypermaximalist, hyper realistic, super detailed , samplers DPM++ 2M SDE steps 35 , Cfg 3. 5-4. 5, 4x-UltraSharp upscale 1. 5 steps 35 denoising strength 0. 34 Style cinematic

2 months ago
3 D illustration of a Slavic couple with their dog sitting on the steps, wearing black outfits and sunglasses, holding coffee cups in their hands, posing for Instagram photos. The man is dressed in all-black , including leather shoes, while his wife wears dark pants, a blazer jacket, a cap, white sneakers, glasses, and has long hair. They both have stylish expressions and pose together. Their small brown poodle sits between them. The image is a full-body shot from the front. --ar 30:37 --motion low --video 1

2 months ago
Prompt principal (Stable Diffusion) “Realistic sports photography, 40-year-old male padel player, brunette buzz-cut (almost bald), five-day stubble, striking a powerful overhead smash above his shoulders on a professional blue padel court, using a dark pro padel paddle with holes (no tennis racket), distant full-body shot, dynamic pose frozen mid-air, dark blurred background to accentuate motion, natural side lighting creating dramatic highlights and realistic shadows, detailed clothing textures (technical T-shirt, shorts), visible ball trajectory blur, crisp depth of field, 85 mm lens look, f/2.8, ISO 400, 1/2000 s –‐ green eyes sharply in focus — sharp, photorealistic, high detail, RAW style” Variaciones Ángulo bajo + contraluz “Low-angle view, golden-hour backlighting, warm rim-light outlining the athlete, flare highlights, cinematic sports photo, ⌀85 mm, f/2.0, 1/4000 s, ISO 200” Plano lateral con motion blur de fondo “Side-view pan shot, background motion blur streaks, cool color grading, shutter 1/125 s, follows player, sense of speed, sports reportage style” Iluminación artificial nocturna “Night match under stadium LED floodlights, high-contrast lighting, cool white highlights, darker shadows, slight rain mist particles backlit, dramatic ambience” Prompt negativo (sugerencia) “tennis racket, extra limbs, extra balls, duplicate players, watermark, logo, text, cartoon, lowres, out of focus, deformed hands, bad anatomy, poorly rendered face, overexposed, underexposed” Parámetros recomendados Parámetro Valor sugerido Notas Modelo base SD 1.5 / SDXL (fotográfico) O usa tu checkpoint fotográfico favorito (por ej., Realistic Vision). Sampler DPM++ 2M Karras (ó Euler a) Karras suaviza luces; Euler a da contraste. Steps 30-40 (1.5) / 25-30 (SDXL) Menos ruido residual en caras. CFG Scale 6.5 – 7.5 Suficiente fidelidad sin over-constrain. Resolución 1344 × 768 px (16∶9) Escena panorámica con margen lateral. Aspect ratio --ar 16:9 (controlnet/resizable) Mantiene sensación de pista amplia. Seed Ajusta; prueba 123456 para reproducibilidad Cambia para variar composición. High-res fix / Upscale Latent → 2×, R-ESRGAN 4xSharp (0.33 denoise) Mejora texturas de pala y ropa. ControlNet (pose) Activado (weight 0.8) Introduce esqueletos de smash para consistencia corporal. LoRA / Embedding (opcional) “Sports-Photog-LoRA” @ 0.6 strength Refuerza estética fotográfica real. Explicación técnica Composición Partimos de un plano lejano (long shot) con fondo oscurecido mediante bokeh y sub-exposición selectiva; esto dirige la atención al smash y al contraste azul de la pista. El ángulo de 85 mm simula un tele corto común en fotografía deportiva. Iluminación Natural side lighting crea modelado en músculos y marca la barba. Para la variante nocturna, los LEDs ↑ contraste: compensa con rango dinámico (ISO 1600, pasos extra). Color & detalle La paleta fría (azul pista + verde ojos) contrasta con piel cálida; el prompt fija el paddle oscuro para evitar confusión con raqueta. Negativo & CFG El negative prompt elimina artefactos frecuentes (doble pala, tenis, textos). CFG ≈ 7 mantiene obediencia sin rigidez; sube a 8-9 si se desvían detalles clave (calvicie, ojos verdes). Flujo SDXL Genera base a 512 × 896 px (ratio 9:16 invertido si prefieres vertical), hi-res fix 2×, denoise ~ 0.25 con R-ESRGAN. Opcional: ControlNet Pose o OpenPose para clavar anatomía del smash.

7 months ago
Create a beautiful image of a bride dressed in a traditional Kerala saree and a groom in a mundu & shirt, posing in a serene location such as home, amidst paddy fields, or on temple steps. The couple should be holding a palm leaf scroll or a chalkboard with "Save the Date – [30-04-2025]" written on it. Use natural lighting to enhance the authenticity and traditional feel of the photoshoot. Capture the intricate details of their attire and surroundings to convey a sense of warmth and tradition.

6 months ago
raditional Japanese pagoda standing tall in a foggy mountain valley, positioned slightly off-center to the right in a mid-range view. Dense mist surrounds the base and rolls through the distant layered hills. Entire scene rendered in monochrome charcoal tones with no color grading. Warm interior lighting with slight intensity variation across floors, one or two brighter windows, and subtle light spill for realism, while preserving the tranquil, cinematic mood. Subtle stone steps or a narrow rocky path leading toward the pagoda, partially revealed through the mist for realism and grounded presence. Soft pink petals drifting in the air — extremely minimal, serving only as atmospheric accents. Petals should have around 30–50% saturation, subtle but noticeable against the grayscale charcoal background, with mild size variation and slight foreground presence. Three to four birds flying with varied size and distance, soft motion blur, and partial mist integration. Bare branches subtly frame the scene on the sides. Full cherry blossom trees in very soft, subtle pink. No foliage should obscure the pagoda. Moody ink wash aesthetic, inspired by Ghost of Tsushima — cinematic yet understated. High detail in the pagoda structure, gentle shadows, calm, and mysterious tone. Minimalist, peaceful composition with deep atmosphere and silence.

4 months ago
Swedish , 30 years old,woman,blonde, long hair,Black mesh mini but not see-through, long- heeled shoes, slim, ,perfect neck and arms Veck,sTender, DOF,Aperture,insanely detailed and intricate, character,hypermaximalist,hyper realistic,super detailed , samplers DPM++ 2M SDE steps 35 , Cfg 3.5-4.5, 4x-UltraSharp upscale 1.5 steps 35 denoising strength 0.34 Style cinematic

16 days ago
# Keeps 589 bright, boosts jewelry shine, replaces background, and maps to Casinofi duotone. import cv2, numpy as np from google.colab import files from PIL import Image # Upload your image when prompted up = files.upload() fn = list(up.keys())[0] img_bgr = cv2.imdecode(np.frombuffer(up[fn], np.uint8), cv2.IMREAD_COLOR) # --- Palette (BGR) --- HEX = lambda h: (int(h[5:7],16), int(h[3:5],16), int(h[1:3],16)) SHADOW = np.array(HEX("#0F1011"), np.float32) MID = np.array(HEX("#8E7A55"), np.float32) HILITE = np.array(HEX("#E6D2A1"), np.float32) HILITE_PLUS = np.array(HEX("#EBDDB7"), np.float32) # extra-bright cream for 589 # --- Helper: gradient map (shadow -> mid -> highlight) --- def gradient_map(gray01): g = gray01[...,None] t1 = np.clip(g/0.5, 0, 1) t2 = np.clip((g-0.5)/0.5, 0, 1) low = SHADOW*(1-t1) + MID*t1 high = MID*(1-t2) + HILITE*t2 return np.where(g<=0.5, low, high) hsv = cv2.cvtColor(img_bgr, cv2.COLOR_BGR2HSV) # --- Masks --- # Background (yellow) range bg_mask = cv2.inRange(hsv, (15, 120, 120), (40, 255, 255)) # tune if needed # Shirt (blue) range – helpful for separate contrast if you want shirt_mask = cv2.inRange(hsv, (95, 80, 40), (130, 255, 255)) # Numbers “589” (white-ish areas on shirt) gray = cv2.cvtColor(img_bgr, cv2.COLOR_BGR2GRAY) num_mask = cv2.threshold(gray, 210, 255, cv2.THRESH_BINARY)[1] # bright white # Jewelry (gold/yellow highlights) jew_mask = cv2.inRange(hsv, (12, 60, 120), (30, 255, 255)) # gold tones # Clean masks a bit def clean(m, k=3): kernel = cv2.getStructuringElement(cv2.MORPH_ELLIPSE, (k,k)) m = cv2.morphologyEx(m, cv2.MORPH_OPEN, kernel, iterations=1) m = cv2.morphologyEx(m, cv2.MORPH_CLOSE, kernel, iterations=1) return m bg_mask = clean(bg_mask, 5) shirt_mask= clean(shirt_mask, 5) num_mask = clean(num_mask, 3) jew_mask = clean(jew_mask, 3) # --- Step 1: Replace background with deep charcoal --- out = img_bgr.copy() out[bg_mask>0] = SHADOW # --- Step 2: Convert subject to Casinofi duotone --- # Work on non-background regions subj = out.copy() subj_mask = (bg_mask==0).astype(np.uint8)*255 subj_gray = cv2.cvtColor(subj, cv2.COLOR_BGR2GRAY).astype(np.float32)/255.0 mapped = gradient_map(subj_gray).astype(np.uint8) mapped = cv2.bitwise_and(mapped, mapped, mask=subj_mask) bg_area = cv2.bitwise_and(out, out, mask=bg_mask) out = cv2.add(mapped, bg_area) # --- Step 3: Boost numbers “589” to brighter cream and keep edges crisp --- num_rgb = np.zeros_like(out, dtype=np.uint8) num_rgb[:] = HILITE_PLUS num_layer = cv2.bitwise_and(num_rgb, num_rgb, mask=num_mask) out = cv2.bitwise_and(out, out, mask=cv2.bitwise_not(num_mask)) out = cv2.add(out, num_layer) # Optional: thin dark stroke around numbers edges = cv2.Canny(num_mask, 50, 150) stroke = cv2.dilate(edges, np.ones((2,2), np.uint8), iterations=1) out[stroke>0] = (out[stroke>0]*0 + SHADOW*0.9).astype(np.uint8) # --- Step 4: Jewelry shine (Screen-like brighten in cream) --- # Create a cream layer and blend additively where jewelry mask is j_layer = np.zeros_like(out, dtype=np.float32) j_layer[:] = HILITE j_mask_f = (jew_mask.astype(np.float32)/255.0)[...,None] out_f = out.astype(np.float32) out = np.clip(out_f + j_layer*0.35*j_mask_f, 0, 255).astype(np.uint8) # --- Step 5: Gentle contrast pop on subject only --- subj_mask3 = cv2.merge([subj_mask, subj_mask, subj_mask]) subj_pix = np.where(subj_mask3>0) sub = out.astype(np.float32) sub[subj_pix] = np.clip((sub[subj_pix]-20)*1.08 + 20, 0, 255) out = sub.astype(np.uint8) # Save cv2.imwrite("output_casinofi.png", out) files.download("output_casinofi.png") print("Done. Download output_casinofi.png")