Prompt Formats

uvoo generates prompts for four AI image generation models. Each uses a different syntax and parameter system. Below is the exact format for each, with annotations explaining how brand signals map to prompt elements.

Midjourney (v6.1)

Midjourney prompts are free-form text followed by parameter flags. uvoo structures them as: subject → lighting → mood → style cues → palette → composition → parameters.

editorial product shot, [subject description],
soft diffused natural light, warm shadow fill,
[mood tags], [texture cues],
palette: [hex1] [hex2] [hex3],
centered composition, generous negative space,
--ar 4:3 --style raw --v 6.1 --s 200

Key parameters:

• --ar — aspect ratio, derived from your most common crop ratio
• --style raw — reduces Midjourney's default stylization for more photographic output
• --s (stylize) — set between 100–300 based on how "designed" your brand feels vs. documentary
• --no — negative prompt, appended with common off-brand elements detected in your outlier assets

DALL-E 3

DALL-E 3 uses natural language. It has strong prompt adherence and excellent text rendering. uvoo uses a structured sentence format that OpenAI's system prompt layer handles well.

A [style] photograph of [subject].
[Lighting description]: [specific lighting details].
Color palette restricted to [hex1], [hex2], [hex3].
[Composition instruction]. [Mood adjectives].
[Texture/surface notes]. Photorealistic, commercial photography quality.

Notes:

• DALL-E 3 doesn't support negative prompts — instead, uvoo adds explicit positive constraints ("smooth surface only", "no visible text")
• Color hex values are included but DALL-E interprets them loosely — CIELAB descriptors are added alongside for perceptual guidance
• Best for: product shots with text overlays, campaign graphics, marketing materials

Stable Diffusion (SDXL / Flux)

Stable Diffusion uses weighted prompts with comma-separated tokens. Emphasis is controlled with (term:weight) syntax. uvoo outputs prompts optimized for SDXL and Flux.1.

(editorial photography:1.3), [subject],
(soft natural light:1.2), (warm tones:1.2),
[mood tags], (negative space:1.1),
color palette [hex1] [hex2],
(professional product photography:1.2),
8k, high detail
Negative: harsh shadows, oversaturation, busy background,
(off-brand colors:1.5), stock photo feel

Notes:

• Weights above 1.3 can cause artifacts — uvoo caps at 1.4 for primary style cues
• Negative prompt is critical for SD; uvoo populates it from your brand's detected off-brand patterns
• Best for: custom fine-tuning workflows, local generation pipelines

Ideogram v2

Ideogram excels at photorealism and is best for product photography. It uses clean natural language with a style modifier.

[Style descriptor] product photo: [subject].
[Lighting]. Palette: [hex1] [hex2] [hex3].
[Composition]. [Mood]. Photo-realistic.

Notes:

• Use PHOTO or REALISTIC style modifiers via the API for commercial photography
• Ideogram has strong color accuracy — CIELAB L*a*b* values are included in the API request's color guidance field when available
• Best for: product photography where color accuracy matters most

Prompt fields reference