・テンプレートのオリジナル・ワークフローに対して「① モデルの変更, ② シード値を固定(同じ画像を再現できるように)」を変更する
Text to Image ワークフロー
Text to Image ワークフロー(プロンプト・エンハンサー付き) SubGraph
Prompt: A high-resolution, surreal digital illustration showing a human hand holding a martini glass. The image is overlaid with whimsical, expressive ink-style doodles, including a cartoon figure inside the glass, a drawn citrus wedge on the rim, and various abstract sketches and faces surrounding the glass against a clean, white background. The style seamlessly blends a realistic, lit photograph with loose, hand-drawn marker artistry, creating a playful and artistic juxtaposition.
This workflow generates images with Krea-2 Turbo, a fast distilled text-to-image model. The workflow is organized into a few parts: each handles a different step so you can generate quickly without digging through every node.
Text to Image (Krea-2 Turbo) Subgraph
Double-click this node to open the full pipeline. Inside, the canvas is grouped into several areas: Models: Loads the diffusion model, text encoder, VAE, and optional style LoRA (LoraLoaderModelOnly). Prompt: Enter your image description in Text String (User Prompt). Prompt Enhancement: Expands your prompt before generation. Enabled by default (prompt_enhance). You can turn it off to use your prompt as-is, or replace the built-in module with OpenAI or Gemini API nodes. Feature Switch: Quick toggles on the left side of the subgraph: ・prompt_enhance: enable or disable prompt expansion ・enable_lora?: enable or disable the style LoRA
Image Generation: Handles latent sizing, sampling (8 steps), and VAE decode to produce the final image.
When using a LoRA, set enable_lora? to true, select the matching file in LoraLoaderModelOnly, and adjust lora_strength as needed.
LoRA & Trigger Word (CustomCombo)
On the main canvas, the CustomCombo node is where you choose which style LoRA you are using (e.g. krea2_coolblue, krea2_darkbrush). The workflow uses your selection to automatically append the correct trigger word to your prompt. ・Download LoRA files from Comfy-Org/Krea-2/loras and place them in ComfyUI/models/loras/ ・Select the LoRA name that matches the file loaded inside the subgraph ・The trigger word is applied for you: no need to type it manually ・For trigger words and recommended strength values, see the LoRA Trigger Words and Settings note beside this workflow
You are an expert prompt engineer for text-to-image models. Your task is to expand the user's prompt into a highly effective image-generation prompt.
Think step by step about the request before writing the answer: - What is the subject and mood? - What visual styles, mediums, and lighting options would fit? Consider two or three alternatives and pick the one that best serves the caption. - What composition, framing, and grounded details will help the text-to-image model?
Then output a single expanded prompt paragraph.
Follow these rules strictly: 1. **Faithfulness First:** Preserve all original subjects, actions, colors, and spatial relationships. Do not add new objects, props, characters, or animals unless the user clearly implies them. 2. **Practical T2I Structure:** Write a prompt that a text-to-image model can parse cleanly. Group subjects with their own attributes and actions. Use grounded phrasing for poses, interactions, and spatial layout. 3. **Style Planning Stays Internal:** Use your internal reasoning to choose style, medium, framing, and lighting. Do not emit planning tags or wrappers in the visible answer body. 4. **Text Rendering:** If the user requests visible text, quotes, labels, or typography, specify the exact text clearly and wrap requested words in quotes. 5. **Avoid Over-Specification:** Do not invent highly specific clothing, colors, materials, or scene details unless the input supports them. 6. **Structure:** Write one cohesive paragraph after the thinking block. No bullets, JSON, or markdown. 7. **Respect Existing Detail:** If the user's prompt is already detailed, lightly polish and finalize rather than heavily expanding — preserve their phrasing and direction. 8. **Respect the Human Form:** Treat depictions of people with dignity. Assume clothing covers genitals and intimate anatomy. 9. **Preserve User Medium:** When the user explicitly requests a medium (e.g. "photo of", "photograph of", "illustration of", "painting of", "sketch of", "3D render of"), honor it. Do not pivot to a different medium to avoid difficulty — match the user's stated intent.
Prompt: A high-resolution, surreal digital illustration showing a human hand holding a hurricane glass. The image is overlaid with whimsical, expressive ink-style doodles, including a cartoon figure inside the glass, a drawn citrus wedge on the rim, and various abstract sketches and faces surrounding the glass against a clean, white background. The style seamlessly blends a realistic, lit photograph with loose, hand-drawn marker artistry, creating a playful and artistic juxtaposition.
Prompt: A high-resolution, surreal digital illustration showing a human hand holding a Pilsner glass. The image is overlaid with whimsical, expressive ink-style doodles, including a cartoon figure inside the glass, a drawn citrus wedge on the rim, and various abstract sketches and faces surrounding the glass against a clean, white background. The style seamlessly blends a realistic, lit photograph with loose, hand-drawn marker artistry, creating a playful and artistic juxtaposition.
A beautiful Japanese woman in her 20s, standing gracefully in a softly lit, minimalist indoor setting with natural light filtering through sheer curtains, wearing a simple, elegant white blouse and dark trousers, her hair styled in a loose, flowing bob with a few strands framing her face, eyes gently closed as if savoring the moment, hands resting lightly at her sides, posture relaxed yet poised, surrounded by subtle ambient details like a small potted plant on a wooden side table and a faint glow from a desk lamp behind her, rendered in a photorealistic style with soft focus background and shallow depth of field to emphasize her serene expression and delicate features, capturing a quiet, tranquil mood that feels both intimate and timeless.
A beautiful Japanese woman in her 20s, captured in a bust-up angle, standing confidently with soft natural lighting that highlights her delicate facial features and smooth skin texture; her hair gently frames her face, styled in a loose, elegant manner, and she wears a simple, light-colored top that complements her slender frame; the background is softly blurred with warm, diffused tones to emphasize her as the central focus, creating an intimate yet polished portrait style — rendered as a high-resolution photograph with shallow depth of field, capturing the subtle play of light on her shoulders and collarbone, evoking a serene and graceful mood.
A Japanese woman sitting alone at a cozy café in the evening, sipping coffee from a ceramic mug with steam rising gently, warm ambient lighting casting soft golden hues across her face and the wooden table, the background blurred with subtle bokeh of dimly lit shelves and hanging lamps, her dark hair loosely falling over her shoulders, dressed in a comfortable sweater, the scene captured as a photograph of intimate quietude, shallow depth of field emphasizing her relaxed posture and the gentle curve of her hand around the mug, no other people visible, the atmosphere calm and contemplative, the coffee shop’s interior adorned with muted earth tones and soft textures, the moment frozen in serene stillness.
A tall, elegant businesswoman with a poised and confident posture, standing upright in a professional setting, dressed in a sleek, high-necked tailored suit with structured shoulders and a minimalist silhouette, her hair styled in a sophisticated updo, facing slightly toward the viewer with one hand resting lightly on her hip and the other holding a thin leather-bound notebook or tablet, rendered in a photorealistic style with soft studio lighting that highlights the texture of her fabric and the subtle sheen of her polished shoes, set against a blurred modern office background with glass partitions and ambient desk lights, capturing an air of quiet authority and refined professionalism.
A mysterious 27-year-old professional model, standing in a dimly lit studio with soft directional lighting that highlights her sculpted features and elegant posture, wearing a sleek, minimalist black bodysuit with subtle texture, her long dark hair cascading over one shoulder, eyes half-lidded with an enigmatic gaze, hands gently resting at her sides, the background blurred to emphasize her as the sole focal point, rendered in high-resolution cinematic photography style with shallow depth of field, capturing both her poised stillness and the quiet intensity of her expression.
Medium shot, side profile of a young East Asian woman with dark hair tied back in a ponytail and subtle bangs. She is sitting at a wooden counter next to a large window inside a bright cafe, resting her chin on her hand and looking outside with a gentle smile. On the table in front of her is a clear plastic cup filled with iced coffee and a black straw. The lighting is soft and natural, coming from the window, illuminating her face and creating a warm, airy atmosphere. The background is softly blurred, showing minimal cafe interior with wooden furniture and neutral-toned walls. Clean composition, shallow depth of field, high-resolution photographic style.
A medium close-up shot of a young Japanese woman with a short, black bob haircut and bangs, smiling warmly at the camera over her shoulder. She is wearing a blue denim jacket. Her hands are held near her chest, loosely holding a pair of wooden chopsticks. She is standing outdoors on a narrow street next to an open-air food stall or ramen shop. The background features glowing traditional Japanese paper lanterns with black Japanese characters written on them, including the word "ラーメン". Steam gently rises from the food stall counter. The background is softly blurred with a shallow depth of field, showing hints of passersby and city lights at dusk. Natural, soft lighting illuminates her face, highlighting her skin texture and clear expression. Shot with a 35mm lens, realistic textures, detailed denim fabric.
Wait — the user’s input is minimal, and no medium or style has been specified. That means I must assume a default medium (likely “illustration” or “digital art”) and choose a visual style that best suits the whimsical, surreal subject: a white yeti with horns reading a book titled “Ostris + Krea2 Style Reference.” The mood should be curious, slightly absurd, and artistic — not menacing or realistic.
Step 1: Subject and mood — The yeti is the central figure, white, with horns, engaged in an intellectual act (reading). The mood is playful yet scholarly, with a touch of fantasy. The book title suggests it’s about art styles, so the illustration should feel like a stylized reference sheet or concept art.
Step 2: Visual styles — I’ll pick “digital illustration” as the medium since it’s flexible, supports stylized characters, and allows for clean lines and vibrant colors. For style, I’ll go with “cartoonish but detailed,” to balance whimsy with clarity — not too childish, not too photorealistic. Alternative styles like “anime” or “concept art” could work, but “cartoonish but detailed” better serves the tone of a “style reference” book.
Step 3: Composition — Centered shot, the yeti sitting upright, book open on its lap or held in front. Its face should be expressive — perhaps slightly bemused or focused — to emphasize the act of reading. Background should be simple and uncluttered to keep focus on the character and book. Lighting should be soft and even, avoiding harsh shadows to maintain the illustrative quality.
Step 4: Grounded details — The book’s title must be clearly visible, in bold sans-serif font, centered on the cover. The yeti’s horns should be curved and elegant, not jagged. Its fur should be fluffy and white, with subtle shading to imply texture. No other objects or characters — no animals, no humans — unless implied by the context (which it isn’t).
Step 5: Text rendering — The title “Ostris + Krea2 Style Reference” must appear exactly as written, wrapped in quotes if necessary for clarity, but since the user didn’t request typography, I’ll just write it plainly. The font should be modern and legible, possibly with a slight gradient or metallic sheen to match the “reference” theme.
Step 6: Avoid over-specification — I won’t invent clothing, colors beyond white and
&clipboard{[Trigger Word], editorial fashion photo, soft window light, a young japanease woman in an oversized beige trench coat standing in a minimalist concrete room, 85mm lens, shallow depth of field, muted earth tones, film grain}:
&clipboard{[Trigger Word], ghibli-style anime illustration, warm afternoon light, a small seaside town with terracotta rooftops, fishing boats in the harbor, soft clouds, gentle waves, painterly textures, wide establishing shot}: