Luminous Landscape Forum
Raw & Post Processing, Printing => Adobe Lightroom Q&A => Topic started by: Ellis Vener on August 07, 2011, 07:38:16 am
-
Just discovered something I had not noticed before in Lightroom's histogram: even without turning on the boxes around the clipping indicators ( keystroke : J) the clipping point triangles will light up to let you know there is clipping somewhere in the photo.
-
That's the entire point of them being there.
-
Yes kaelaria, but not everyone knows this. In fact I have been surprised by the number of people I know who have been using Lr for a couple of years who were unaware of it. Not everyone is as expert a user as you are.
-
No need for the snaky remarks, I'm just stating a fact.
-
The nice thing is that the color of the triangle shows which channel is clipping. When the color is white, then all channels are clipping.
-
Great point Francisco!
-
Great point Francisco!
I'm sure Kaelaria was about to mention that! ;)
-
Only issue is the clipping isn’t based on what you’ll end up after you render the data unlike ACR. Its Melissa RGB.
-
Only issue is the clipping isn’t based on what you’ll end up after you render the data unlike ACR. Its Melissa RGB.
Now everyone should say: ohhhhhhhhhhhhhhhhhhhhhhhh :-[
-
The clipping points will be the same for ProPhotoRGB
-
The clipping points will be the same for ProPhotoRGB
For the primaries yes, but its using a totally different gamma encoding so it isn’t going to be the same and, if you are not using ProPhoto, its going to be even farther off.
-
Gamma encoding will change the shape of the histogram but will not affect the extreme values. At the maximum signal level of 255 (assuming 8 bits) the value is the same, it is where the linear and gamma curve intersect, so there is no way to alter the clipping point by gamma encoding.
If you use any other color space, then I agree with you.
-
Gamma encoding will change the shape of the histogram but will not affect the extreme values. At the maximum signal level of 255 (assuming 8 bits) the value is the same, it is where the linear and gamma curve intersect, so there is no way to alter the clipping point by gamma encoding.
Well I haven’t tested it, maybe you have. I’d be surprised if an sRGB TRC which is what MelissaRGB is showing is the same in the darker areas as a 1.8 true gamma curve of ProPhoto RGB. I don’t think it will be much, point is, unlike ACR, we are not getting the proper (ideal) feedback.
-
The clipping point as indicated by the triangles is minimally affected by gamma encoding choices. This is because of rounding. Effectively, we apply a gamma encoding and then see what round up to 255 and what round down to 0 (i.e., 8-bit rounding). The choice of gamma value will affect values just at that threshold. Not worth worrying about, but just FYI in case you wonder why.
-
Eric, the one million euro question is: why Adobe decided not to adapt the histogram to the chosen output colour profile like ACR does?
-
Alas, the choice of output profile in LR is not as clear cut as in ACR. That is, in LR the output paths are Export, Print, and Web (and Publish, I suppose). Even if one were to focus on just one of these (e.g., Export) it is possible to have many export templates set up (with different choices of color space). Hence it's not clear which one to use. Using the "last Export template used" would be inconsistent and confusing, since it could vary over time, even within one LR session.
To be honest, I don't think checking for clipping via histograms (even if the user gets to choose the color space) is particularly useful. I understand it's the only way available currently, but we need a better way in the future.
-
Maybe a nice way could be softproofing with destination space histogram
-
I agree its not as clear cut as ACR but it wouldn’t be that difficult IMHO, to just allow the user to option click on the Histogram and call up a list of profiles as we see in print (filtered such that you can limit the list). It doesn’t have to sync with the other modules, only the Histogram and associated RGB percentage values (with perhaps, the ability to set the scale to 0-255). What happens in other modules isn’t important, no histogram or numbers to deal with. Let Web deal in sRGB.
Doing this would at least allow parity with ACR. In fact it would provide more options which might be an ACR issue. With the option for scale, at least a user could see the same values in LR as they get in Photoshop or would see in ACR. Then I’d want to see Photoshop provide the percentage options for its info palette. I actually prefer, and find many newer users learn dealing with the numbers as percentages easier than 0-255.
-
Hope this isn't getting off topic, but I see the highlight clipping indicator change when adjusting the brightness slider. I thought the brightness control was not supposed to affect the extremes. Or, is the clipping indicator not accurate in this regard. Maybe this has to do with the rounding?
-
My understanding of the clipping indicators is that they will indicate clipping in any channel. Probably by adjusting the Brightness slider, you're pusing one channel over the limit. Saturated primary colors seem to be most susceptible to this. This is a subject I'm very interested in - someone please correct me if I'm wrong about this!
-
Hope this isn't getting off topic, but I see the highlight clipping indicator change when adjusting the brightness slider. I thought the brightness control was not supposed to affect the extremes.
Brightness used only by itself will not clip...if you are seeing clipping when adjusting Brightness, it means some other adjustment is actually pushing the boundaries and Brightness is being added to the other adjustment.
-
Values close to the maximum (i.e 252 -254 in 8 bits) could become 255 by adjusting brightness and this will activate the indicators. This also could be due to rounding errors. Anyway, I have seen this behavior too.
Something to consider is that brightness does this in a smooth way, so there should not be an abrupt transition as in a regular blocked highlight.
-
I guess clipping warning works with a threshold. So although some highlight area could not be strictly clipped, for high Brightness values tonal compression could be so strong that wide areas as a whole reach the clipping warning threshold.
This is ACR's brightness control represented as curves:
(http://www.guillermoluijk.com/article/acrps/brillo_acr.gif)
I guess LR is similar (BTW if someone posts 2 images from LR, before/after, I will calculate the transfer curve. The only condition is that only the control under study is applied, any other setting must be set to 0, neutral,...:
-
I guess clipping warning works with a threshold. So although some highlight area could not be strictly clipped, for high Brightness values tonal compression could be so strong that wide areas as a whole reach the clipping warning threshold.
This is ACR's brightness control represented as curves:
(http://www.guillermoluijk.com/article/acrps/brillo_acr.gif)
I guess LR is similar (BTW if someone posts 2 images from LR, before/after, I will calculate the transfer curve. The only condition is that only the control under study is applied, any other setting must be set to 0, neutral,...:
Guillermo's graph provides a nice illustration of the answer to my question, assuming that Lightroom's brightness control performs similarly to ACR's.
-
Yes, LR's controls are exactly the same as ACR's, including Brightness.
As noted earlier, the clipping indicators will be slightly conservative because of the roundoff. So in 8-bit space, if you had something that was like 254.6, then technically it hasn't clipped yet, but for the purposes of the clip indicator it'll get rounded off to 255, and hence shown as clipping. Bottom line is that if you see the indicator, then you've either clipped data or you're very close to clipping data.
I am indeed aware of the request to have a better way to preview clipping in other spaces. I hope to be able to offer a solution in the future.
-
Eric, the one million euro question is: why Adobe decided not to adapt the histogram to the chosen output colour profile like ACR does?
Guillermo,
Reference to one of the early Lightroom podcasts (#6 as I recall), includes a discussion between Mark Hamburg (the architect of LR) and Thomas Knoll (the architect of Phososhop and ACR) where they discuss histograms, color spaces, info readout, etc. Mr. Knoll suggested that LR could allow the user to select the color space for output rendering as does ACR, but Mr. Hamburg wanted to keep things simple.
Once the decision was made, internal politics at Adobe would make criticsm of that decision by Adobe engineers rather awkward.
Regards,
Bill
-
... but Mr. Hamburg wanted to keep things simple.
Understandable,
considering that neither LR nor ACR offer specific tools to defeat saturation clipping from gamut conversion.
Without intending to overemphasize this aspect, pRGB to aRGB or sRGB is still among the few reasons where I occasionally prefer to resort to Photoshop.
Peter
--