Introduction

When users recolour old photos, the face is usually the most important part of the image. Clothing, walls, skies, and backgrounds can tolerate small color mistakes, but a face that looks too orange, too smooth, too modern, or too artificial will immediately make the whole image feel fake.

A good AI recolour image prompt should not only say “colorize this photo.” It should tell the AI photo colorizer to preserve the original face, identity, skin texture, shadows, age details, and vintage atmosphere. This is especially important when you recolour black and white photos of family members, grandparents, portraits, wedding photos, school photos, or historical people.

If the prompt is too simple, the result may have plastic skin, wrong eye color, over-bright lips, modern makeup, or a face that no longer looks like the original person. If the prompt is clear, old photo colorization can look more natural, respectful, and believable.

Basic Face Recolour Prompt Structure

A strong face-focused prompt should include six parts:

Identity: keep the original face and facial structure unchanged
Skin tone: use natural, realistic, era-appropriate skin color
Texture: preserve pores, wrinkles, freckles, film grain, and age details
Facial features: keep eyes, lips, eyebrows, nose, and expression accurate
Lighting: follow the original light direction and shadows
Style: avoid modern beauty retouching, heavy makeup, or oversaturated colors

Example:

Recolour this black and white portrait into a natural realistic color photo. Keep the original face, identity, facial structure, expression, pose, and age details unchanged. Use soft, believable skin tones with natural shadows. Preserve wrinkles, pores, film grain, eye details, hair texture, and vintage photo atmosphere. Do not smooth the skin too much, do not add modern makeup, and do not change the person’s facial features.

Natural Skin Tone Prompt

For old photo colorization, skin tone should look soft and human, not orange, red, waxy, or overly polished. The prompt should guide the AI recolour image tool to respect the original lighting and the person’s age.

Example:

Recolour this old black and white face photo with natural skin tones. Keep the person’s original facial identity, age, expression, eye shape, nose, lips, and jawline unchanged. Use soft realistic skin color with slight warmth, natural cheek shadows, visible skin texture, and preserved film grain. Avoid orange skin, plastic smoothing, heavy blush, glossy makeup, or modern beauty filters.

Use this type of prompt for family portraits, passport-style portraits, yearbook photos, and close-up black and white photos where the face is the main subject.

Eye, Lip, and Hair Prompt

Eyes, lips, and hair often become too bright or too stylized during AI photo colorizer workflows. To keep the face natural, describe these details carefully but do not over-control them.

Example:

Recolour the face naturally while preserving the original eyes, lips, eyebrows, and hair texture. Keep the eyes realistic and not overly bright. Add subtle natural lip color, not lipstick unless it appears historically appropriate. Keep the hair color believable based on the grayscale photo, with soft highlights and natural shadows. Do not change the face shape, expression, or original character of the person.

This works well when you recolour old photos of one person and want the final result to feel like a real vintage color portrait.

Wrinkles and Age Detail Prompt

Many AI tools try to “beautify” faces automatically. For old family photos, that can be a problem. Wrinkles, smile lines, under-eye detail, and skin texture are part of the person’s identity.

Example:

Recolour this old portrait while preserving all natural age details. Keep wrinkles, smile lines, under-eye texture, skin pores, facial shadows, and original expression visible. Use gentle realistic skin color and avoid making the face look younger. Do not remove age marks, do not overly smooth the skin, and do not change the person’s identity.

This is useful for grandparent photos, historical portraits, memorial images, and archive-style old photo colorization.

Group Photo Face Prompt

Group photos are harder because each face may have different lighting, size, and detail. The prompt should ask for consistent but not identical skin tones.

Example:

Recolour this old black and white group photo with natural, realistic face colors. Keep every person’s identity, facial structure, expression, pose, and age unchanged. Use believable skin tones for each face based on the original lighting and shadows. Preserve film grain, soft focus, and vintage texture. Do not make all faces the same color, do not over-sharpen small faces, and do not add modern makeup or artificial smoothing.

This helps when users recolour black and white photos of families, weddings, classrooms, teams, or historical groups.

Bad Prompt vs Better Prompt

Bad prompt:

Colorize this face photo and make it look beautiful.

Better prompt:

Recolour this black and white portrait into a natural old photo colorization. Keep the original face, identity, expression, hairstyle, skin texture, and age details unchanged. Use realistic skin tones, subtle lip color, natural eye detail, soft vintage lighting, and preserved film grain. Do not beautify the face, do not smooth the skin too much, and do not make the colors look modern or oversaturated.

The better prompt works because it tells the AI what to protect, what to color, and what to avoid.

Follow-Up Prompts for Fixing Unnatural Faces

If the first AI recolour image result looks unnatural, use follow-up prompts instead of starting over.

If the skin is too orange:

Make the skin tone softer and more natural. Reduce orange and red saturation while keeping the original face, shadows, and vintage texture unchanged.

If the face looks too smooth:

Restore more natural skin texture, pores, wrinkles, and film grain. Do not make the face look like a modern beauty portrait.

If the eyes look fake:

Make the eyes more realistic and less bright. Keep the original eye shape, gaze direction, and facial expression unchanged.

If the person looks too young:

Preserve the original age, wrinkles, facial lines, and natural skin detail. Do not make the person look younger.

If the result looks too modern:

Make the color grading more vintage and historically believable. Reduce modern makeup, glossy skin, bright lips, and overly saturated colors.

Face Recolour Checklist

Before downloading the final image, check these points:

The face still looks like the original person.
The skin tone is soft, realistic, and not orange.
Wrinkles, pores, and age details are still visible.
Eyes and lips are natural, not too bright or stylized.
Hair color matches the vintage mood of the photo.
The lighting follows the original black and white photo.
The image still has film grain and old photo texture.
The result does not look like a modern beauty filter.

Conclusion

Natural face colorization depends on restraint. The best AI photo colorizer result is not the most colorful version, but the version that keeps the person recognizable, human, and emotionally believable.

When you recolour old photos, your prompt should protect the original face first. Mention identity, skin tone, facial features, age details, lighting, and vintage texture. Then add clear negative instructions such as no plastic skin, no modern makeup, no face reshaping, and no oversaturated colors.

For family photos, portraits, and historical images, this approach makes old photo colorization feel more respectful and realistic. A good prompt helps AI bring color back to the memory without changing the person inside it.

Reference

Zhang, R., Isola, P. and Efros, A.A. (2016) ‘Colorful Image Colorization’, arXiv. Available at: https://arxiv.org/abs/1603.08511 (Accessed: 3 July 2026).

Zhang, R., Zhu, J.-Y., Isola, P., Geng, X., Lin, A.S., Yu, T. and Efros, A.A. (2017) ‘Real-Time User-Guided Image Colorization with Learned Deep Priors’, arXiv. Available at: https://arxiv.org/abs/1705.02999 (Accessed: 3 July 2026).

Xu, R., Tu, Z., Du, Y., Dong, X., Li, J., Meng, Z., Ma, J., Bovik, A. and Yu, H. (2022) ‘Pik-Fix: Restoring and Colorizing Old Photos’, arXiv. Available at: https://arxiv.org/abs/2205.01902 (Accessed: 3 July 2026).

Pang, Y., Jin, X., Fu, J. and Chen, Z. (2025) ‘Structure-preserving Feature Alignment for Old Photo Colorization’, arXiv. Available at: https://arxiv.org/abs/2508.12570 (Accessed: 3 July 2026).

Li, B., Wang, Z., Li, F., Xu, J., Guo, J., Pei, R., Li, X. and Chen, Z. (2026) ‘ColorFLUX: A Structure-Color Decoupling Framework for Old Photo Colorization’, arXiv. Available at: https://arxiv.org/abs/2603.28162 (Accessed: 3 July 2026).