Recover high-contrast photos

[Sayesh Ti]

I have got an almost blown-out picture in a backlit sunset color photo. Which is the most promising method to enhance the (very) faint contrasts in the almost black front part of the picture? I enclose two original scans, one of which with a revision tentative (... r1). Thanks !

Angela XII TIFF 1200 Kontr -49 Unsharp Mask, Hgrund entf high r1.tif Angela III TIFF.tif

Edited January 15 by Sayesh Ti

[Ron P.]

@Sayesh Ti Welcome to the forums

From what I see in those images, if you're trying to recover the shadows or for that matter, the highlights, Affinity Photo is not capable of that. I think those images are beyond being recovered. The photographer should have followed the rules, Expose for the Highlights, let the shadows fall where they may.

If you are shooting very bright backlit images, you must make a decision. Do I want this to be a silhouette or not? If not, then you must try to light the shadows. Use a flash, or even a reflector bouncing light back into the dark areas. 

Blown out, means there's no data captured by the camera sensor, basically in the highlights.

Are you wanting the image to be a silhouette of the person? IF so you can isolate them using a selection, then invert it so the background is selected. OR just select the background. From there, work with the various adjustments, like Levels, Highlight/Shadows, or Curves to darken the background area.

 

[Sayesh Ti]

My gratitude, Ron P. Yep, that was what I also thought. Milking a stone is still magic. 😄

[Ron P.]

For what it's worth, I tried using Luminosity Range Mask with Curves, Levels, adjustments. Isolated the person and pushed the shadows, as well as reducing the highlights as far as possible. There's just no data there.

[Old Bruce]

18 minutes ago, Ron P. said:

... The photographer should have followed the rules, Expose for the Highlights, let the shadows fall where they may. ...

<cranky old man mode ON>

Back in the day... The mantra as (not actually) taught by Ansel Adams and Minor White was "Expose for the shadows and develop for the highlights." It served me quite well.

<cranky old man mode is always ON>

These new fangled electric cameras threw out most of the old rulebooks.

[Ron P.]

Ok, I've never read or heard that method.

If you expose for the shadows, that would cause highlights to be blown out, ie; not recoverable, IF the highlights were already at the brightest. So developing the highlights would be very difficult if not impossible. Most of the younger generation wouldn't have a clue on how to properly meter exposures, or how to use a meter, the one built in the camera or an external one.

I know what's done me well is following what the meter tells me, and if I can not increase exposure without blowing out the highlights, I use some way to throw light into the shadows.

Yes I know, and have seen many, read a lot about Ansel Adams.

[Old Bruce]

3 minutes ago, Ron P. said:

I know what's done me well is following what the meter tells me, and if I can not increase exposure without blowing out the highlights, I use some way to throw light into the shadows.

I wish more people would think to do this. As it is the shadows that we want to be brighter it doesn't take much to get them to a useful level.

[NotMyFault]

The only option is to scan the original (positive or negative or film slide?) again, and manually adjust the brightness of the scan lightning.

I was able to restore certain details, e.g. outblown highlights from negative film, by using a digital camera and long exposure.

it is generally not possible to restore outblown highlight from regular scans of printed images or positive film.

[Ron P.]

9 minutes ago, Old Bruce said:

I wish more people would think to do this. As it is the shadows that we want to be brighter it doesn't take much to get them to a useful level.

Oh, and that doesn't include being able to watch your shutter speed, or aperture. I've been caught several times, focusing my attention on what the exposure meter is reading, and not seeing that my shutter speed is too slow to get the exposure so I get a sharp image. A good way to practice that is to shoot where those change rapidly, like shooting wildlife, birds in flight.

 

[Ldina]

@Sayesh Ti I agree with all comments made above. Just thought I'd add one more idea (which doesn't always work, and is only worth it for important images).

You can try adding textures, colors, and detail from other images, textured backgrounds, using effects (such as procedural textures), or perhaps cloning portions of your image, and placing them in those blownout areas. This is best done on a separate pixel layer, where you can reduce opacity, try different layer blend modes, use masks to blend the effects, etc. You're actually creating a "composite image". If done carefully, you can sometimes "reconstruct" an area of an image that is normally blown out and unrecoverable (i.e., has no detail). Sometimes it works well, and other times poorly, depending on the nature and detail or the portion of the image you are trying to "resurrect". 

Before attempting something like the above, it's always best if you can start with an original that isn't hopelessly blown out. FWIW.

[thomaso]

2 hours ago, Ron P. said:

For what it's worth, I tried using Luminosity Range Mask with Curves, Levels, adjustments. Isolated the person and pushed the shadows, as well as reducing the highlights as far as possible. There's just no data there.

FWIW, too: Isn't it enough to reduce/increase the Exposure first to get a quick impression whether relevant data is generally available in an under-/overexposed image? (… and only proceed with adjustments, masks, etc. when more than just a few stripes in the pattern of the clothing appear in the person's silhouette)?

[R C-R]

4 hours ago, Old Bruce said:

Expose for the shadows and develop for the highlights.

For what little it is worth, doing a web search on phrase that turns up a lot of interesting (& sometimes conflicting) hits about what it means.

[Sayesh Ti]

My biiiig thanks ☺️ to everybody,

for supplying some more hints for the next steps. Before this I tried hard on that unique hardcopy from 1974, over 150 tentatives 🤣🤣, beginning with my scanner settings (Unsharp mask on/off, Deraster on/off, Background removal, varying Luminosity, Contrast and Gamma, TIFF format, and then post-processing on the PC, playing with RGB, gamma, and the like. Best results until now with 1200 or 600 dpi, Unsharp mask on or off, deraster off, Gamma at max., Noiseware Noise reduction in some cases, contrast reduced by ~50, reduced saturation and slight increase in Red and Blue. But now I doubt if the sensor of my then Nikon FE2 [* correction: Konica Auto S3] captured more than I already obtained. I believe the original was a Kodachrome film. Just for fun I attach one of my better results till now.

Next photos I'll take with two or three different settings, and flash is always an interesting option. But for this I'll need a time machine ... Thanks again folks for your recommendations. 🤗

Angela VIII TIFF 600 dpi, derast, Kontr red., r4.tif

Edited January 16 by Sayesh Ti

[NotMyFault]

Probably you have reached the limit set by your equipment.

In general:

• deactivate all „auto improvements“ like sharpening, denoise, de-Raster etc. they could improve some defects, but always do collateral damage to the original data which you would need in this case.
• You need 16 bit color channel depth. Your scanner seems to only output 8 bit. A workaround would be HDR bracketing of multiple scans with different exposure
• a backlit scene like this does not allow to recover details. It is method to intentionally remove details at capture time. The only way to keep such details would be to either use a camera sensor (or film) offering more dynamic range, or do exposure bracketing during initial capture. By principle you can only recover those details conserved at initial capture.
• Despite being a tiff file which allows lossless compression, the actual files show typical compression artifacts of 8x8 blocks. To recover details you must get a uncompressed image. Otherwise any method to recover details will only amplify those compression artifacts. Another cause of this effects are scanners that actually have a physical resolution of e.g. 300 DPI and upscale the image to 600 or 1200 dpi. 

Background: I have scanned 1000s of printouts, negative and positive film slides by scanners and digital cameras, and used scan services of commercial print services like Fujifilm since about 199x where you could include a picture CD when ordering printed copies of film slides. My actual gear (Canon R6 II, RF-100 Macro lens, led light dish) gives a million times better quality than all older methods. but 50% of the film slides started to physically disintegrate.

Some images from 1960 are still gorgeous and as detailed, colorful and sharp like an image taken today with the digital cameras. Unfortunately the quality of film and film cameras degraded in the 1970.