Can I restore bad color in a full-size photo using its embedded thumbnail?

Asked 6/30/2019

3 views

2 answers

0

I have some older digital photos where the full-size JPEGs look overexposed or have strange color, but the embedded thumbnails extracted from those same files look normal. Is there a practical way to use the thumbnail as a color reference to improve the larger image in Photoshop or similar software, even though the thumbnail is much smaller?

Originally by Photography Stack Exchange contributor. Source · Licensed CC BY-SA 4.0

Photography Stack Exchange contributor

7y ago

2 Answers

3

I can't tell for Photoshop (but I assume the method would be similar). In Gimp I would do:

  • open the full scale image
  • add the thumbnail as an additional layer
  • scale up the thumbnail so that it exactly matches the full-scale image
  • set the thumbnail layer to "Color" blend mode

Rationale: the detail we perceive in the image is more due to luminosity than to color. With the method above, you create an image that takes its luminosity (and therefore its detail) from the full-scale image, and its colors from the thumbnail.

Originally by user75947. Source · Licensed CC BY-SA 4.0

user75947

7y ago

0

AI Answer

Generated from our catalog & community — verify before relying on it.

Yes—if the thumbnail’s color looks correct, you can use it to transfer color to the full-size image.

A practical method is:

  1. Open the full-size image.
  2. Place the extracted thumbnail above it as a new layer.
  3. Scale the thumbnail up so it aligns with the full-size photo.
  4. Set the thumbnail layer’s blend mode to Color.

This works because most perceived detail comes from luminance rather than color. The full-size image keeps its original luminance/detail, while the thumbnail supplies the color information.

The result may not be perfect, since the thumbnail is low resolution and may not match exactly, but it can be a straightforward way to improve overall color when the full-size file has a bad color rendering applied.

UniqueBot

AI

7y ago

Your Answer