GPT Image 2 | Editorial Fashion Photography

メタ情報

Date: 2026-04-30
Tool: Codex (macOS app)
Model: GPT Image 2 via ChatGPT 5.5 (Medium)

ベースライン

プロンプト

[IMAGE TASK]
prompt = """
High-end editorial fashion photography, a professional model wearing haute couture, highly stylized, captured on a Hasselblad 500CM medium format camera, with a 120mm f/4 lens and f/8.0 for sharpness. Precise studio strobe lighting from a beauty dish defines facial structure and highlights every textural detail of the fabric. Photorealistic masterpiece, stunning realism, film grain texture, natural skin tones.
"""

result = client.images.generate(
    model="gpt-image-2",
    prompt=prompt,
    size="1024x1024",
    aspect_ratio="1:1",
)

出力

入力

プロンプト

/mtp <args>

[AGENT INSTRUCTION]
This request contains an MTP invocation.
Before generating or composing the image prompt, run the MTP compiler using the /mtp arguments.
Apply the compiler stdout as silent constraints to the image-generation task, specifically regarding composition, lighting, style, subject framing, and subjects.
Do not infer MTP settings from the image prompt.
Do not include this instruction, /mtp arguments, or MTP XML in the final image prompt.

[IMAGE TASK]
prompt = """
High-end editorial fashion photography, a professional model wearing haute couture, highly stylized, captured on a Hasselblad 500CM medium format camera, with a 120mm f/4 lens and f/8.0 for sharpness. Precise studio strobe lighting from a beauty dish defines facial structure and highlights every textural detail of the fabric. Photorealistic masterpiece, stunning realism, film grain texture, natural skin tones.
"""

result = client.images.generate(
    model="gpt-image-2",
    prompt=prompt,
    size="1024x1024",
    aspect_ratio="1:1",
)