3 replies to this topic

[realflow100]

Why is it not possible to darken highlights or unclip them. but you CAN brighten shadows easily in photos and reveal hidden details (Maybe noisy. but shadows dont get clipped hard like highlights do in photos)

 

Why is that? 

 

why can you take a raw photo and brighten up the shadows so dramatically. but if highlights are clipped. theres very little to nothing you can do at all.

 

Why is noise always in the dark parts of images more than the light?

 

Is there any kind of fancy of AI deep learning image mainipulation software that can even moderately recover clipped highlights?

 

 

Is there any way to make a camera sensor act in reverse so noise would be in the bright parts of the image instead of the dark shadows.. and just read the image in reverse so that its normal way but noise happens in highlights instead of shadows? or is that just not physically possible?

[Nikon Shooter]

One may recover some shadow to a limit but softwares can not
generate information if no information was recorded. None can be
generated but noise…

One may recover some high lights to a limit but softwares can not

generate information if no information was recorded. None can be 
generated but luminance through a silvery… something.

[Ron]

An ancient film analogy. When shooting slide film if you overexposed.... oh, say the sky for example, the slide would often be completely clear where that over exposure occurred. All of the emulsion that makes up an analog (film photo) was washed away during processing. In other words, zero photographic data. You can't make something from nothing. There has to be some data there to work with. With digital it's pretty much the same. When you over expose a shot the overexposed area is often blank. Again, no data. Nothing for your software to work with. Photographic software is getting smarter all the time but, at this point, it still can't extrapolate an image, or part of an image, where no data was recorded.

 

The problem is the same for underexposed areas but to a much lessor degree which is why they can often be saved... with varying degrees of success. 

 

--Ron

 

Edit: I should have said that, with underexposure the reverse is actually the case. The sensor is over saturated with data. In extreme cases to the point where it outstrips the cameras built in dynamic range. The reason why underexposed images can often be saved is because there's usually still some data available. Often very noisy data, but data none the less. 

[Merco_61]

Noise happens everywhere, but it is much more visible when the signal level is low. 

Blown highlights mean that the sensor's analog output is saturated and can't record the full extent of the available light. As this happens before the conversion to digital, there is no more data to process. This supersaturation doesn't happen with the absence of light.