Colorization
What is Colorization?
Colorization is adding colour to black and white images or footage: either by hand or, increasingly, using AI that predicts what colours the original scene would have contained.
At a glance
- Also known as
- Colour restorationAI colourisationDigital colourisation
- Used for
- Restoring historical photographs and footageReinterpreting archival materialConverting greyscale AI output to colour
- Common tools
- DeOldifyPalette.fmAdobe photoshop neural filtersDaVinci resolve
Ready to create?
Direct scenes, design characters, and ship full films
All-in-one AI creative platform with simple, transparent pricing, no speed throttles, and an infinite Canvas for max creativity.
How it compares
Colorization adds colour where none previously existed, reconstructing or inferring colour information from greyscale data. Colour grading adjusts and stylises colour in footage that already contains colour information. Colorization is about adding data; grading is about refining what is already there.
Think of it like…
Colorization is like filling in a detailed line drawing that was originally published without colour: you are making informed decisions about what colours the scene contained based on context, knowledge, and reference, adding a dimension of information that was not previously captured.
Pro tip
When using AI colorization on historical footage, always cross-reference AI colour predictions against period-accurate reference materials for objects like uniforms, vehicles, and signage. AI models make educated guesses based on training data, and historically specific items may be assigned incorrect colours that require manual correction to ensure accuracy.
Types and variations
- Colorization approaches vary based on technique and application.
- Manual colorization involves human artists painting colour directly onto greyscale images using reference materials and historical research to guide colour choices.
- Semi-automatic colorization uses AI to provide a colour draft that human artists then review, correct, and refine.
- Fully automatic colorization relies entirely on AI prediction without human correction, suitable for rapid prototyping or lower-stakes applications.
- Style-based colorization can apply the colour palette of a reference image or artistic style to a greyscale source, rather than attempting photorealistic colour reconstruction.
- Selective colorization leaves most of an image greyscale while applying colour to a specific subject or element, using contrast between colour and monochrome for dramatic or artistic emphasis.
Ready to make your first scene in Morphic?
Try MorphicCommon use cases
- Colorization is widely used in documentary filmmaking and historical media production to make archival footage more emotionally accessible and immersive for contemporary audiences.
- News organisations and cultural institutions use it to restore and present historical photographs.
- Social media content creators use automatic colorization tools to process vintage photographs for engagement.
- In AI creative workflows, colorization is useful when working with greyscale-rendered or stylised imagery that requires colour for final delivery, or when exploring creative reinterpretations of classic visual material.
Ready to create?
Direct scenes, design characters, and ship full films
All-in-one AI creative platform with simple, transparent pricing, no speed throttles, and an infinite Canvas for max creativity.
FAQs
Colorization is the process of adding colour to images or footage that was originally created in black and white or greyscale. It can be performed manually by artists using historical references, semi-automatically with AI assistance, or fully automatically using neural network models.
AI colorization models are trained on large datasets of colour images. The model learns associations between image content ( sky, foliage, skin, architecture ) and typical colour values. When presented with a greyscale image, it analyses the content and predicts plausible colour values for each region based on those learned associations.
AI colorization produces plausible colour results based on statistical learning, not historical accuracy. For general subjects like sky and foliage, the results are usually convincing. For historically specific items like uniforms, vehicles, and period clothing, human review and correction using genuine historical reference materials is necessary to ensure accuracy.
Peter Jackson's They Shall Not Grow Old (2018) is widely regarded as one of the most technically impressive and emotionally impactful examples of film colorization, using a combination of AI assistance and extensive manual work to bring World War One archival footage to life in colour for the first time.
Yes. Tools like DeOldify and Palette.fm offer AI colorization for photographs, while some video editing applications include neural colorization features. For AI-generated greyscale imagery that needs colour for final delivery, these tools provide a quick starting point that can be refined manually in post-production.
Colorization of historical footage raises questions about authenticity and interpretation. Critics argue that adding colour to footage that was shot in black and white alters its historical character and may misrepresent history. Proponents argue that it increases accessibility and emotional connection. Best practice involves being transparent about the colorization process and presenting it alongside the original material where possible.
Selective colorization is a technique where only a specific subject or element within an otherwise greyscale image is presented in colour. The contrast between the monochrome environment and the coloured subject draws strong visual attention to that element and is commonly used for dramatic or stylistic emphasis in photography, advertising, and music videos.
