Scribble at 2026-11-07 06:25:05 Last modified: unmodified
chiaroscuro, professional photograph, ultra detailed, scenary, (cave), stunning upscale environment, ultra realistic, sharp focus, depth of field, HDR, 32K, 16K, photorealistic rendering captures the intricate folds of the fabric and its elegance with precision, natural lighting, movie promotional poster, vibrant details, photograph taken by Marc Adamus, with Leica SL2, APO-Summicron-SL 75mm F/2 ASPH, Fujifilm Acros 100 II <lora:High quality:0.9> high quality
Negative: bad anatomy, bad hands, three hands, three legs, bad arms, missing legs, missing arms, poorly drawn face, bad face, fused face, cloned face, worst face, missing fingers, extra fingers, ugly fingers, long fingers, horn, extra eyes, huge eyes, amputation, disconnected limbs, bad proportions, blurry, cropped, deformed, dehydrated, disfigured, duplicate, error, extra arms, extra legs, extra limbs, fused fingers, gross proportions, jpeg artifacts, long neck, low quality, lowres, malformed limbs, morbid, mutated hands, mutation, mutilated, poorly drawn hands, signature, text, ugly, username, watermark, worst quality, bad anatomy, bad hands, three hands, three legs, bad arms, missing legs, missing arms, poorly drawn face, bad face, fused face, cloned face, worst face, missing fingers, extra fingers, ugly fingers, long fingers, horn, extra eyes, huge eyes, amputation, disconnected limbs, bad proportions, blurry, cropped, deformed, dehydrated, disfigured, duplicate, error, extra arms, extra legs, extra limbs, fused fingers, gross proportions, jpeg artifacts, long neck, low quality, lowres, malformed limbs, morbid, mutated hands, mutation, mutilated, poorly drawn hands, signature, text, ugly, username, watermark, worst quality
Steps: 6, Sampler: Euler a, Schedule type: Bong Tangent, CFG scale: 1, Seed: 1526857842, Size: 1088x1536, Model: qwen_image_2.1_turbo_bf16, Model hash: 188471e7c1, Module 1: qwen_image_2.1_vae_bf16, Module 2: qwen3vl_8b_int8_convrot, RNG: CPU, Lora hashes: "High quality: c98dd2d25ba8", Emphasis: None, Version: neo-2.30
Qwen-Image 2.1 で出してみた１枚だが、chiaroscuro（陰影法）をうまく扱えていると思う。プロンプトのキーワードは自然文とキーワードの組み合わせで整理されていないのだが、それでも破綻が少ないように調整してくれている。これも、テキスト・エンコーダに他のオープン・ウェイトでも使われている Qwen が採用されているからだろうし、DiT (diffusion transformer) が同じく Qwen のモデルであることと相性が良いのかもしれない。ともあれ、出てくる画像の出来栄えが Krea 2 や MOSS-Image よりも構図という大局だけでなく細部においても上手く整えられていて良好だ。