1 person found this helpful
Yammer P wrote:
So, what I'm trying to ask is: is there a combination of exposure settings to avoid?
Nope...not really. If the settings give you what you want visually, in general, that's what you want to use. And yes, the order YOU do something doesn't make any difference to the actual processing pipeline which has its own processing order.
The preview image is re-rendered using the entire combination of settings every time you move a control.
If you move one control (e.g. Exposure) and that leads to a block of pixels being "whited-out", then you move another control (e.g., Recovery) that undoes that condition, then you haven't "overexposed" anything.
Move the controls around freely until you like what you see in the preview and histogram, then press Open or Save. It's really as simple as that.
1 person found this helpful
Anyway, this is not entirely true. There are some issues using exposure slider in gradient/brush tool, so if you change exposure for blown highlights (whether increase or decrease), you can't compensate with other slider (say global recover or exposure sliders). I illustrated this in a post about half year ago, Eric responded that this was a know issue. So, exposure slider in this tool should be used carefully or avoided for blown highlights (although it would be quite useful - to correct only blown parts of the picture, without affecting hue/saturation/contrast on other parts, like with recover slider), unless this problem is already addressed
That's very interesting, Vit, and exactly the sort of thing I was concerned about. Do you suppose you can find the thread again?
My concern is a technical one, rather than an operational one. I know how ACR works in practice, it's the mechanisms "under the hood" which concern me - probably without reason, but informed reassurance would be welcome.
Let me give a hypothetical example...
A cloud is exposed at -0.3EV of full exposure. Land is underexposed by 2 stops.
Global exposure is increased by 2 stops, overexposing the sky.
A 2-stop exposure gradient is applied at the horizon.
An adjustment brush is used to darken the cloud by 0.3EV.
Now, we all know that this results in the land being boosted by two stops, with the sky being boosted and darkened by equal amounts (theoretically cancelling each other out). So the cloud should be down to -0.6EV, with the effect of the adjustment brush.
Bearing in mind that this is a hypothetical scenario, and there are lots of other ways of producing similar results, using different combinations of positive and negative values of compensation, is there a chance that the internal calculations of ACR can "max out" areas of image before reducing them again? Or is the overall compensation calculated BEFORE application, or is the headroom so high as it is never reached?
It would be nice to know, in case certain combinations are problematic.
It's on the end of this thread
It seems that I've forgotten the details in the meantime - using exposure slider with brush tool can have strange effect on the highlights of the whole picture (not only brushed parts) . Anyway, this is actually visible on blown saturated colors (like colors around the Sun on my example), so you'll face this problem on relatively small number of pictures
My example above was rubbish, but hopefully people are understanding what I'm getting at. I think you described it better in the other thread. If the raw data isn't clipped, and the compound effect of multiple exposure/recovery adjustments should't theoretically clip the image, then the question is: are there any combinations of settings which may clip the data during the internal conversion process?
For those old enough to remember early pocket calculators, it's a bit like adding numbers over the maximum, causing an out of range error, even thought the next step was a subtraction.
Part of ACR processing workflow is documented in dng sdk, so I can tell only for that part, which includes conversion from sensor color space to output color space (using profile) and usage of sliders exposure and blacks. My answer would be: it is possible that color in output color space (say sRGB) appears not to be clipped, meaning R,G and B values are in range 1..254, but some of channels got clipped during conversion. This could be the case even if none of channels in sensor color space was clipped. There is a number of scenarios for this to happen, so, like Jeff said - If the settings give you what you want visually, in general, that's what you want to use
I stand corrected, I suppose.
Seems to me to be yet another good reason not to try to paint in ACR.
Anyway, this can happen only when colors have relatively high saturation and/or high brightness. I just wanted to say that it can happen, but in real life it's rarely a problem