Remove scratches, dust, and stains
Clean isolated marks and age damage while preserving real texture, clothing seams, hair, and background structure.
RESTORE OLD AND DAMAGED PHOTOS
AI Photo Restoration
Repair scratches, dust, stains, creases, faded tone, and damaged edges while keeping the people, place, composition, and period character of the original photograph recognizable.
- Repair visible damage
- Preserve facial identity
- Choose black and white or color
PHOTO RESTORATION / ARCHIVE REVIEWREADY
SCRATCHES
+FADING
+CREASES
+MISSING EDGES
RESTORATION, NOT GENERIC ENHANCEMENT
Repair the history in the photograph without rewriting it.
Photo restoration targets physical and age-related damage. It can clean contamination, rebuild small missing areas, recover tonal separation, and improve legibility before optional colorization or upscaling.
Repair creases and damaged edges
Reconstruct torn corners, fold lines, and small missing regions using the surviving photograph as the visual reference.
Recover faded detail and color
Improve contrast and tonal separation, then keep authentic monochrome or request restrained, period-aware color.
PRESERVATION RULES
Say what must remain unchanged before asking AI to repair damage.
The best restoration prompt separates documentary facts from defects. Lock identity and composition; name only the damage that should change.
PRESERVE THESE DETAILS
- Facial identity, age, and expression
- Original pose and body proportions
- Clothing, jewelry, uniforms, and era cues
- Architecture, objects, and composition
- Original black-and-white character when requested
Review: Inspect faces, hands, text, insignia, jewelry, architecture, and repeated patterns at full size. AI reconstruction is a visual interpretation, not factual recovery of missing information.
REFERENCE IMAGE METHOD
Use the best scan of the old photo as the reference—not a screenshot of it.
Upload the cleanest, highest-resolution scan or evenly lit photograph you have. Keep the full border when damage reaches the edges, and state exactly which features Image Edit must preserve.
- Scan flat when possible; avoid glass and glare
- Keep the full frame, borders, and torn corners visible
- Use even light and a camera parallel to the print
- Do not combine contradictory portraits of the same person
REFERENCE IMAGE / PORTRAIT CHECKREADY
HOW TO RESTORE AN OLD PHOTO
A controlled restoration in three passes.
Open the existing Filtrix Image Edit workflow, use the original photo as the reference, and make the repair brief explicit.
Upload the best source
Use a sharp scan or evenly lit photo with the entire print visible. Avoid reflections, aggressive compression, filters, and screenshots.
Describe damage and preservation
List scratches, stains, creases, fading, blur, or missing edges to repair, then state the identity and historical details that must not change.
Review, refine, and download
Zoom into faces and factual details. If needed, use another narrow Image Edit pass for one remaining problem instead of regenerating everything.
OLD PHOTO RESTORATION USE CASES
Match the repair brief to the kind of damage you can actually see.
Scratched family photographs
Remove dust, spots, scratches, and fold marks without smoothing away faces, fabric, or film texture.
Faded and low-contrast prints
Restore tonal separation and clarity while keeping highlights, shadows, and skin believable.
Torn edges and missing corners
Rebuild limited damage from nearby visual evidence while preserving the original crop and scene structure.
Black-and-white archive photos
Clean and repair the monochrome original first; colorize only as a separate, optional creative interpretation.
IMAGE EDIT PROMPT TEMPLATE
Give Image Edit a repair list and a preservation list.
Use the uploaded old photo as the reference image. Describe defects first, then lock people, objects, era details, and composition before requesting tonal or color changes.
Open Image EditCOPY-READY RESTORATION PROMPT
Restore this old family photo. Remove scratches, dust, stains, and fold marks; repair damaged edges and small missing areas. Preserve each person's identity, age, expression, facial proportions, pose, clothing, jewelry, original composition, and period-accurate details. Improve tonal separation and restrained sharpness. Keep it black and white. Do not modernize faces, invent accessories or text, over-smooth skin, or change the background.
PRACTICAL RESTORATION GUIDE
Prepare the source, direct the edit, and verify the history.
Better input and narrower instructions reduce invented details. Use this checklist before the first edit and again before accepting the result.
01SOURCE
Digitize the print correctly
Capture as much real information as possible before AI touches the image.
- Scan around 600 dpi when available
- Remove glass or photograph without reflections
- Keep edges and damaged areas inside the frame
02REFERENCE
Use the photo itself as the anchor
The original is the identity and composition reference. The prompt only describes the intended repair.
- Start from the least compressed source
- Use one coherent reference per restoration
- Name the faces and details that must stay fixed
03EDIT
Change one class of damage at a time
Broad rewrites increase drift. Narrow Image Edit passes are easier to inspect and reverse.
- Repair damage before optional colorization
- Separate face repair from background cleanup
- Upscale only after the restoration is approved
RECOMMENDED ORDER
A restoration sequence that limits drift.
Each pass should solve a visible problem while carrying forward the accepted details from the previous one.
- 01Crop and geometry
Straighten the print and keep the full border or tear evidence before reconstructing edges.
- 02Surface damage
Remove dust, spots, scratches, mold marks, and crease lines.
- 03Missing structure
Repair only the torn corners or small absent regions supported by nearby evidence.
- 04Faces and tone
Recover restrained clarity, contrast, and facial legibility without changing identity.
- 05Optional color and scale
Choose colorization as a separate interpretation, then upscale the approved restoration for delivery.
Archive integrity check
AI cannot know what was truly present in a missing face, medal, sign, building detail, or handwritten word. Keep the untouched scan, label colorized versions clearly, and treat reconstructed details as an interpretation.
AI PHOTO RESTORATION FAQ
Questions about repairing old and damaged photos with AI.
What is AI photo restoration?
AI photo restoration repairs visible age and physical damage in an existing photograph, such as scratches, dust, stains, creases, fading, blur, and limited missing areas, while trying to preserve the original people and scene.
Is photo restoration the same as image enhancement?
No. Image enhancement broadly improves clarity, noise, light, contrast, and color. Photo restoration specifically addresses damage and historical preservation. Enhancement or upscaling can be a later finishing step.
How should I photograph or scan an old photo?
A flat scan around 600 dpi is a strong starting point when available. Otherwise use even light, keep the camera parallel to the print, avoid glass reflections, and include the complete border and damaged edges.
How do I use a reference image for restoration?
Upload the best scan as the reference image. Tell Image Edit which defects to repair and explicitly preserve identity, age, expression, clothing, pose, composition, and period details.
Should I restore and colorize in one step?
Usually restore the monochrome photo first, approve the repaired structure, and colorize in a separate pass. This makes changes easier to compare and keeps the archival version distinct from the creative interpretation.
Can AI accurately recover completely missing details?
No. It can generate a plausible reconstruction from nearby evidence, but the result is an interpretation. Verify faces, writing, medals, jewelry, architecture, and historically important details before use.
Give an old photograph a careful second life.
Upload the best source to Filtrix Image Edit, state what must stay true, and repair only the damage the photograph actually has.
Restore an old photo