I primarily use DT to convert portraits to B&W. Toward that my pixel pipe and module order is:
Denoise->Lens Correction->Exposure->Tone Equalizer #1 (see comment below)->Crop->Color Contrast (to prep color image for conversion)->Color Balance RGB #1 (see comment below)->Color Calibration (to convert to B&W)->Color Balance RGB #2 (see comment below)->Local Contrast->Tone Equalizer #2 (see comment below).
Tone Eq #1 is for adjusting the mask, correct for clipping, and also ensure histogram is spread as wide as possible, allowing for more points to control the tone. Tone Eq #2 is to dodge/burn B&W image (converted in Color Calibration and adjusted thereafter) before exporting.
Color Bal RGB #1 is to prep the color image before conversion. I do this by increasing chroma and vibrance and increasing mid-tones. For low-key images I increase shadows, decrease highlights; for high key, I decrease shadows, increase highlights. Color Bal RGB #2 is to adjust the image after conversion. I only adjust the sliders under Perceptual Brilliance, mostly shadows and mid-tones. As to why I do it this way with two instances of Color Bal RGB: I’m emulating Boris and his YouTube videos on converting portraits to B&W.
I believe I’m doing this wrong. By the time I get to Tone Eq #2 the histogram is far from where it was when I finished with Tone Eq #2. It shows clipping and/or is more compressed than where it was when I was done with Tone Eq #1. Clearly the adjustments I’m making in Color Bal RGB #1 and #2 are messing up those I made in Tone Eq #1.
Which brings me to my question: would it be better to leave Tone Eq #1 out of the pixel pipe? i.e., the only instance of Tone Eq would be at the very end, to dodge/burn the converted image.
Do you have a raw + xmp you could share? That would make it easier to see what you mean and propose solutions.
One thing I will note, is that you should be very careful with perceptual brilliance. If you go past 20% or so, the math tends to break down and can cause problems.
Just to clarify as it was a bit confusing are you talking about the actual histogram or the one displayed for the mask in the tone eq…they are not the same…the display in the tone eq is just for the mask not the image…
Yes, I understand, they aren’t the same. Bruce Williams in his video on the Tone Eq does an excellent job of clarifying how they are different.
Adjusting the Mask sliders (e.g., Compression) impacts the mask histogram and also the image’s histogram. After having adjusted it to span as many points across the mask histogram I then make adjustments in (non-mask part of) Tone Eq. I then work my way up the pixel pipe. In my second instance of Tone Eq the histogram (and mask histogram) are different than where I left them in the first instance. Hence my Q…
I guess I’ll make more sense with the raw+xmp and screenshots. I’ll get to them the next several days. And, oh yes, I don’t adjust the slider for perceptual brilliance any. I noticed the impact it has. I just tweak the mid-tones, shadows, and highlights under perceptual brilliance emulating what Boris did in his video tutorials.
Yes for sure you would expect them to be different… what you see is based on the current input into the module…your first instance gets different input than the second…and if you go changing modules that come before these it can change things as well…This is why for example you will see Boris move instances… he wants that effect to take place with a certain input and not where it comes in the pipeline as that may undo the efforts of a few modules already set as desired but coming behind that… your mask will be based on the input data provided to the tone eq wherever it is in the pipeline whereas the final histogram is the entire pipeline result…
I thought so: after all the modules preceding the second instance of Tone Eq (and the final module in my pixel pipe) feed into it and hence the disparity between what I saw in it vs the first instance…
I guess I can do what I need with Tone Eq in just one instance, the final one.
What circumstances if any would warrant use of two instances of Tone Eq, one early followed by conversion to gray scale and then another to dodge/burn…?
Its not an issue right…every single module is in a linear chain. SInce DT is not nodal you can’t break that so each module feeds the next. That is why you can take some care and move things when needed but the default locations are usually quite reasonable.
If you set up tone with the tone eq and then make lots of changes to the lighting in the CB module well you will need to control those or correct them later with a second instance… you just have to experiment to see what workflow is the most efficient and gives you the results you need…
If you simply duplicate the tone eq one right after the other you will see the histogram of the mask is very different… for subsequent instances you just need to re-define it so you can target the adjustment where you need it…
One issue is trying to do too much with one instance you will usually break the image or you can. I will often use a skewed histogram and separate instances for highlights and shadows… Just focusing on one adjustment allows you to also perhaps introduce a mask to the module to help out and to not try and push and pull the tone so that one change compromises another… You can also do some blending in color channels for certain corrections as well…so there can be numerous reasons to use multiple instances…