// SECTION: PROMPT FRAMEWORK001
Super Prompt Formula V5

High-Fidelity Product Label Generation

Framework unificado para gerar imagens de produtos com labels de alta fidelidade em modelos de IA. Tipografia legivel, hierarquia estruturada e renderizacao fotorrealista.

Scene BlueprintCentral SubjectLabel StructureVisual StyleAnti-Hallucination
Learn now

// SECTION: CORE_PROBLEM002

O problema central com labels de produtos em IA

A maioria dos modelos de geracao de imagem trata o texto das embalagens como elemento decorativo, nao como um sistema de design estruturado. O resultado sao labels ilegsiveis e marcas irreconheciveis.

O Super Prompt Formula V5 resolve isso com 5 modulos estruturais: Scene Blueprint, Central Subject, Label Structure, Visual Style e Anti-Hallucination.

01

Texto ilegivel

Letras distorcidas e embaralhadas nos rotulos

02

Letras aleatorias

Caracteres sem sentido gerados pelo modelo

03

Logos distorcidos

Marcas e logotipos deformados ou irreconheciveis

04

Proporcoes erradas

Embalagens com dimensoes incorretas

05

Hierarquia inconsistente

Sem organizacao visual da informacao no rotulo


// SECTION: PIPELINE_FLOW003

Pipeline de Geracao

O fluxo completo: da imagem real do produto ate a geracao de alta fidelidade. O ULGE Prompt extrai a estrutura do rotulo, e a Super Prompt Formula V5 usa essa estrutura para gerar a imagem final.

INPUT01

product_image

Imagem real do produto com rotulo visivel

PROCESS02

Universal Label Grid Extractor

Analisa e extrai a estrutura do rotulo em formato padronizado

DATA03

label_layout_grid

Bloco estruturado com hierarquia, posicao e tipografia do rotulo

PROCESS04

Prompt Formula V5

Framework de 5 modulos estruturais para geracao de alta fidelidade

OUTPUT05

AI Product Generation

Imagem de produto com label de alta fidelidade, tipografia legivel e realismo fotografico

Input / Output
Process
Data

// SECTION: PROMPT_MODULES004

Arquitetura de Prompt V5

O framework e composto por 5 modulos estruturais. Cada modulo controla um aspecto critico da geracao: cena, sujeito, rotulo, estilo e anti-alucinacao. Clique para expandir e copiar.


// SECTION: ULGE_PROMPT005
Novo

Universal Label Grid Extractor

Prompt projetado para analisar qualquer imagem de produto com rotulo e retornar uma estrutura padronizada pronta para uso em geracao de imagens. Produz automaticamente o bloco label_layout_grid compativel com a Super Prompt Formula V5.

Maximiza
Fidelidade de rotulos
Hierarquia tipografica
Posicionamento espacial
Reutilizacao em prompts
01Leitura do rotuloLer o rotulo antes de gerar texto
02Estrutura visualPreservar estrutura visual do layout
03Hierarquia tipograficaOrganizar hierarquia tipografica completa
04Blocos reutilizaveisManter blocos reutilizaveis para geracao
ulge_prompt.txt
You are a packaging analysis system specialized in cosmetic and consumer product labels.

Your task is to analyze the product label in the attached image and extract the exact label layout grid with high fidelity.

Input image:
<product_image>

Objective:
Identify and extract the label structure, preserving typography hierarchy, spatial layout, and grouping of text elements.

Instructions:

1. Read all visible text on the label.
2. Group the text according to visual hierarchy.
3. Identify the spatial position of each text block.
4. Preserve line breaks exactly as seen.
5. Ignore background noise and reflections.
6. Identify decorative elements (lines, icons, separators).
7. Extract the label structure using a standardized layout grid.

Grid Model:

columns:
LEFT
CENTER
RIGHT

rows:
TOP
UPPER_MIDDLE
CENTER
LOWER_MIDDLE
BOTTOM

You must output the result in the following structured format.

Output format:

reference_image: <image_name>
product_name:
brand:
category:
container_type:

typography_hierarchy:
brand_primary:
brand_secondary:
product_line:
product_type:
description:
ingredients:
claims:
volume:

label_layout_grid:

left_section:

center_section:

right_section:

description_section:

bottom_left:

bottom_center:

bottom_right:

design_elements:

Rules for extraction:

• Preserve original spelling exactly.
• Preserve original capitalization.
• Preserve original line breaks.
• Do not translate text.
• Do not summarize text.
• Maintain label reading order.

The final result must produce a label_layout_grid block that can be directly reused in an AI image generation prompt.

O output do ULGE e ideal para pipelines de geracao de imagem. Use o bloco label_layout_grid como input direto para o campo [LABEL_LAYOUT_GRID] no Modulo 03 (Label Structure) do Master Template V5.


// SECTION: MASTER_TEMPLATE006

Master Template V5

O template completo combina os 5 modulos estruturais em um unico prompt otimizado. Substitua [IMAGEM_DE_REFERENCIA], [IMAGEM_DO_PRODUTO] e [LABEL_LAYOUT_GRID] com suas informacoes.

master_prompt_v5.txt
A photo-realistic commercial scene, meticulously designed using the 'Prompt Architecture V5' framework.
This prompt is a technical instruction for a generative AI model to blend an existing product with a reference composition,
achieving ultra-high-fidelity label rendering.
- The overall scene, including composition, lighting palette, background textures, camera angle, shadows, and reflections, 
must be a direct visual translation of the concepts found in the reference image provided as [IMAGEM_DE_REFERENCIA].jpg. 
- This is the visual blueprint for the atmosphere and spatial arrangement.
- The central subject of this composition is the exact container and bottle shape found in the product image provided as
[IMAGEM_DO_PRODUTO].jpg. This is the object being placed into the scene. Its packaging proportions, material properties, 
brand logo, and overall 
color palette must be preserved with perfect fidelity.
- The front label on the main bottle must be perfectly readable and textually precise. The layout, typographic hierarchy, and every single word 
and character must follow the structure below exactly, with zero distortion and perfect alignment (anti-aliased, vector-like sharpness):

LABEL STRUCTURE START
[LABEL_LAYOUT_GRID]
LABEL STRUCTURE END

- The visual style is an ultra-realistic, high-resolution commercial product photograph. 
- The style is that of professional editorial advertising. It features studio lighting that highlights 
product details, soft and precise shadows, and an extremely sharp focus across the entire label area. 
- The typography on the label must be optically clear and high resolution. 
- The final output must be completely free of misspelled text, distorted letters, or fake characters,
ensuring perfect packaging typography and a clean, premium label design.
Typography Reinforcement Tokens
sharp product label typographyeditorial product photographyhigh resolution packaging textclean cosmetic labelvector-like label claritypackaging typography fidelitymacro packaging detailno distorted textno misspelled textno fake lettersexact label replication