Micro Contrast Lr vs Ps

[smoke665]

My workflow has been to complete most edits in Lr, then shift to Ps if needed for heavy work. However, of late I'm noticing a difference in the micro contrast tonal ranges between images that stay in Lr and those swapped over to Ps and back. When I do that it seems like the transitions are much more abrupt. Am I imagining things, or is there a difference?

 

[ebyelyakov]

No, not imagining. Here are some clues for you:

In Lr you have much more information, assuming you work in RAW. Depending on your settings and gear used, there are quite a few variables. Say my RAWs all 16 bit and open as 16 bit in PS. At the same time, I could open them as 8-bit which will reduce the luminance levels from 32768 to just 256. I did not mistake the number when said 32768 in connection to "16-bit in Photoshop", since in reality, Photoshop uses 15-bit reserving that remaining bit for math -- instead of binary divide/multiply operations they use shift which proves to be heaps more efficient CPU-wise. Like 50x times less cycles are required.

Some operations (frequency separation comes to mind) when executed incorrectly (say subtract instead add in Apply Image, speaking of 16-bit images) will result in losing one level of brightness.. Any consecutive operation will keep adding to it.

On an average quality monitor it is unlikely the difference will be visible. On a Pro monitor -- yes, even one level is noticeable.

 

[smoke665]

RAWs all 16 bit and open as 16 bit in PS.

Using ProPhoto in Lr and Ps, transferring files at 16 bit. The biggest issue I have is that ever so slight difference in transition especially from shadows to highlights.

 

[ebyelyakov]

RAWs all 16 bit and open as 16 bit in PS.

Using ProPhoto in Lr and Ps, transferring files at 16 bit. The biggest issue I have is that ever so slight difference in transition especially from shadows to highlights.

You see, in Lr, RAW -- I suppose they are true 16 bit (provided the camera gives a true 16 bit image). Once they come back from PS - they are in reality 15 bit. I reckon that's the culprit... Since I don't have Lr I could not experiment. In C1 I could not find any visible difference between the original RAW and a PSD exported to PS... I even added an empty curves layer to make sure PS re-saved it... BUT, an interesting observation -- when I attach colour read-outs to the image -- with RAWs they stay in the same physical spot -- so it would be easy to get unbiased LAB for the areas to see if there was a difference indeed... Unfortunately, the readouts didn't stay in place on PSD... Which is weird...

 

[smoke665]

@ebyelyakov I believe I read about this issue, at one time but I can't find it now. There's a ton of stuff about how to add micro contrast, but none that I'm finding on why I even need to boost it. I've read that a camera captures floating point file formats that save the additional information captured by going above 1 and below 0. Apparently Lr is able to utilize that additional information somehow but when you export to a 16 bit file for Ps, it drops the floating point??? The other issue might be color banding. Ps seems to be more sensitive to pushing the values within a gradient to much.

 

[ebyelyakov]

@ebyelyakov I believe I read about this issue, at one time but I can't find it now. There's a ton of stuff about how to add micro contrast, but none that I'm finding on why I even need to boost it. I've read that a camera captures floating point file formats that save the additional information captured by going above 1 and below 0. Apparently Lr is able to utilize that additional information somehow but when you export to a 16 bit file for Ps, it drops the floating point??? The other issue might be color banding. Ps seems to be more sensitive to pushing the values within a gradient to much.

With binary there is no "above 1" nor "below 0".

The camera sensor records luminosity values for each pixel using 16 bits for each - so the range varies from 0 to 65536. The RAW processing software, uses de-mosaic and whatever other algorithms (I'm at the edge of my knowledge in that area) to construct three separate channels - R, G, B from that. This is why your RAWs, exported to PS are almost triple the size (minus the "saving" PS offers, by using only 15-bits per pixel). My knowledge of binary operations does not expand into the floating point maths but some quick read on Wikipedia might help in understanding that. There was a thread 5 years or so ago on Adobe Forums where a developer explained the 15-bit stuff.
Gradients on PS, especially when using a good monitor, kill me! Zooming to 100% view will show the real situation -- others are internal previews, won't be surprised they are in 8-bit only.

The RGB is rather simple. The average of all values will give you the resulting luminosity of the pixel, the highest value will give you chroma, the middle will give you shift, and the lowest will determine the saturation. Starting point if the very centre of the colour wheel - the neutral. My avatar might help to visualise just that. So 15 or 16 bits - the change is rather negligible but quite visible... There is a claim human eye wont' need more that 200 levels to perceive smootheness which is probably true,,, Unfortunately PS fails to produce a smooth representation (as in lack of banding or posterisation) with 8-bit... Sometimes 15 isn't enough as well.