AI Models

The models behind the pipeline

Three complementary approaches to image colorization, working together.

AI Workflow

The processing pipeline

One grayscale image enters. Three specialized AI models restore it in parallel. You compare and download the best result.

01Input

AI SCAN

Original Grayscale Image

ResolutionAuto-detect
FormatJPG / PNG
Processing Ready

02Engine

AI Colorization Engine

Multiple neural models analyze, restore and enhance image colors

03Parallel Models

Zhang Model

Research · CNN

Scientific color prediction model focused on realistic historical image restoration.

Fast Processing
Natural Colors
Research Based

DeOldify

Restoration · GAN

Deep learning model specialized in restoring old photographs with realistic colors.

Historical Photos
Strong Restoration
Classic AI Approach

FLUX AI Enhancement

Generative · Diffusion

Advanced generative model that improves details, realism and visual quality.

Detail Enhancement
Modern Generation
Realistic Output

04Comparison Studio

Comparison Studio

Side-by-side inspection with a real-time before / after slider.

Original

Zhang

DeOldify

FLUX

05Final Results

Your AI Restored Images

Download individual model outputs or export them all in a single ZIP.

Balanced

Zhang Result

Historical

DeOldify Result

Ultra Detail

FLUX Result

Zhang Colorization

Fast · Balanced

GitHub

Strengths

Limitations

Best use case

Technology

Deep CNN · CIE Lab

DeOldify

Historical

GitHub

Strengths

Limitations

Best use case

Technology

GAN · Restoration

FLUX Enhancement

Generative

GitHub

Strengths

Limitations

Best use case

Technology

Diffusion · Generative

Side-by-side comparison

Model	Technology	Main Strength	Color Realism	Detail Enhancement	Best Use Case
Zhang	Deep CNN · Lab	Fast, balanced color	Medium	Medium	General purpose
DeOldify	GAN · Restoration	Historical warmth	High	Medium	Archival portraits
FLUX	Diffusion · Generative	Vivid, detailed	Very High	High	Hero & final output