	Restoration AI Gallery		A gallery of restored images by various AI applications.

Restored Images

AI programs have brought speedy, miraculous restortation capabiliies to genealogists and historians. A variety of programs have been used here to illustrate their capabilities.

Damaged photographs and documents can be restored to their original or near-original condition. Tasks that took many hours before can now be accomplished much quicker.

Types of damage that can be fixed

Scratches, tears, folds, blemishes.
Missing pieces of the photograph or obscured portions of the image.
Restore colors.
Photographs that have faded or become dark.
Torn photographs stitched back together.

See the Prompt descriptions for each sample and try one of these on your images. Different application may return a different 'look' of the same image and prompt.

See Other Galleries

Note: All restored images are the property of Gary W. Clark and AI treatments were conductd by him.

Examine Restored Images Closely

AI programs are not perfect, the can mistakes, and create dreaded 'hallucinations'. Here are some common image characteristics to examine closely. Then decide if it is 'good enough'.

Examine Faces - One of the most common problems (and complaints) is that the AI program substitutes some elses face. This can happen for a variety of reasons:
The submitted image was severely damaged.
The submitted image is low resolution.
Or the AI program is just lazy!
Count the Fingers – Some early restorations have trouble with fingers on a hand. There were to many, to few, or misaligned.
Check for Age – Sometimes a restored face may look younger than reality, though the process may reduce wrinkles & flaws.

Try Again

Try to use a higher resolution image.
Add conditions to the prompt (e.g. use original faces)
Use a different application

Original	Restored
Civil War Era Tintype Prompt: create missing portions of man and the background (Kaze.ai )	Civil War era Tintype (Click image for larger) A similiar restore was made in 2011 with manual Photoshop edits in 4-5 hours. This AI restore was accomplished in less than 30 Seconds
c. 1918 Photograph Prompt: Restore photo, rejoin two broken pieces. (Gemini 3.0)	c. 1920 4 x 5½" Print
WWI Era Photograph Prompt: remove all scratche, blemishes and dark areas, darken overall image slightly and increase contrast. (Gemini 3.0 & ChatGPT)	Gemini 3.0 ----------------------------------- ChatGPT This was in poor condition, background items were not distinguishable. Now we can see it is a tent and bed, and the soldier is recognizable.
c. 1900 Photograph of a railroad work gang. Prompt: rrepair large crack across the photo, reduce sepia tint, increase contrast, sharpen image. (ChatGPT 5) (Click for larger image)	Amazing transformation. (Click for larger)
c. 1860 Tintype – Clark GG-Grandmother Prompt: Restore tintype photo, lighten and increase contrast, remove noise, keep original face.	Gemini 3.0 --------------- -------------ChatGpT 5.0 This plate was in a wooden case, typical of the era. It was nearly unrecognized due to the darkening of the tintype.
1943 Photograph – Author's sister, Shirley. Prompt: remove scratches (Kaze & Gemini )	Kaze ------------- ----------Gemini 3.0 This was only a 2" x 2½" photo, but scanned at high resolutoin (600 DPI)
c. 2000 Photogtraph Prompt: Remove scratches, sharpen image, keep original face. (ChatGPT 5.0 & Gemini 3.0)	ChapGTP 5.0 -------------------------------------Gemini 3.0
1960 Photograph – Xmas at Johnsons, grandfather with money tree surrounded by his grandkids. Prompt: Restore colors to natural state, reduce red tint. (ChatGPT 5.0)	Perfect restoration. Many 1960s & 70 prints and slides deteriorated to red or green tints.

Restoration AI Gallery

Restored Images