2022-03-21_09-20-18.30_DSC6948.NEF.arp (13.1 KB)
2 Likes
Oops! In my previous version I forgot to apply highlight recovery in Filmulator. Similar process to last time, but blends of -1, 0, and +1.
2 Likes
Actually, in darktable, all of the colour channels are somewhat clipped in the projected light area. fortunately, the strong differences in the colours of the glass means that there is enough information at each part of the image of the window for highlight reconstruction to work its magic. WIthout the highlight construction, darktable gives me large swathes of the faux pink as a result of clipping.
The extreme of the projected brightness versus the dark interior shadows is what makes this such a challenge. I “cheated” somewhat by using the tone equalizer to a fairly large degree (at least, I consider it an unusually large degree).
1 Like
To my knowledge darktable doesn’t have the ability to show the histogram of a RAW image, which makes it hard to determine if one or more channels are clipped or not.
If I look at the “normal” histogram in RawTherapee I do see a mini spike in the red as well. This does however depends on the input colour profile that is chosen. If I choose Camera standard, the default for this camera in RT, I see this mini spike, but if I choose the DCP profile that matches both the camera and the used picture control (i.e.: Nikon D3300 Neutral) I do not see this spike, only the blue and green one.
About the clipped areas and looking at your edit: Highlight reconstruction only does a rather average job here, especially the blown spot on the bottom of the pillar, but this noticeable discolouration and artefacting can also be seen in other areas. This is not darktable specific and limited to your edit, though, the other edits all have an obvious hard time creating a pleasant colour transition, tackling artefacts and/or disguising the blown areas.
“Technically,” you’re right. But that’s clear. ok.
It’s just how I didn’t twist this highlight reconstruction, and still it turned out to be a trehsold. It’'s possible to make masks, of course. But I don’t think they were somehow made proprietary during development.
I know what you mean. But the color highlights in the lights in dt are stronger than the lights in Resolve. Such highlights are only whitewashed with a bolder stroke. Tried different variations of working with highlight reconstruction. Most likely, you need to try more classic tools.
Okay - I’m interested in that. But it’s the technical part that interests every user, and this is more of a technical point than an aesthetic one.
I guess that’s the difference between closed and open source/.
The user, if working with opensource, is obliged to understand the technical aspects, whereas he could work more productively just on the moment of what he knows.
But this is already a philosophy. I understand that this is not anyone’s …
Yes, it’s off-topic. Closed.
Anyway i prefer white than extreme recoveries that are so ugly and unnatural… 
I love churches; light, colors, compositional elements. Although, the Sagrada Família in Barcelona drove me nuts… 
In rawproc, librtprocess highlight reconstruction, no tone curve other than the TRC in the output color profile:
Been to Brisbane twice, second trip was two weeks working at Boeing House, quite sad I didn’t get to this gem…
2 Likes
Are you referring to my off-topic? 
Not at all.
I mean that most of the time no extreme recovery leads to a pleasant result.
Either leave burnt parts white or expose them properly
1 Like
Ah, well, in theory, yes))
2 Likes
In darktable, AFAIK, the histogram shows the current state of the edit.
However the toggle to show RAW clipping identifies pretty much all the window projection on the column as clipped. I’d have to read the docs again, but this usually means 1 or more of the colour channels. The fact that darktable was able to reconstruct quite successfully suggests that in each instance (each chroma patch of the projection), it was only 1 channel that was clipped, or, at least, proves that at least 1 of the channels was not clipped.
You cant get the histogram but the raw overexposure is now the actual sensor data…
I see that some tonemapping is applied in Davinci. I would have to google it but you can see it says luminance mapping…that would certainly alter your starting point wrt where DT starts…
Yes, that’s the case.
I know that there’s the show overexposed button, but I did not know it was fixed already. Read the bug-report (PR?) a few weeks back about it not showing the correct info. Rather nice to be able to see actual correct data shown now 
Just checked, it was a PR by Jandren. This one: #11391
HI everyone. Thank you so much* for the many renditions - I am quite overwhelmed by how many of you shared your vision of how this one should (or could) look.
@dqpcoxeas, my initial impression of yours was “too dark,” but it is now really growing on me precisely because it draws the eye immediately to the projected image; the image, which looks beautifully reconstructed, has however quite a purple cast.
@ggbutcher, thanks for your slant on things 
@Thomas_Do - I like the warm tones, which are true to the stone when observed from outside in natural light. Inside, mostly with artificial illumination, the tones were not so warm. However as an aesthetically pleasing take on the building, its a winner.
@XavAL, thank you for the neon highlighter colours!
@david - thank you for the exposure stacking … a very nice rendition of detail in the vaulting.
@age - a very nice rendition of the projected image on the column.
@hatsnp - the lines, tones and textures certainly make for a captivating monochrome treatment; thank you.
@Claes - thank you, your choice of crop and the black border very nicely lead the eye to the projected image. Moreover, it is only now that I can identify fully the white dove above the head of Jesus -the window must be a depiction of the Baptism of Jesus. Again, the human foot at the bottom is clearly discernable. So nice job on the reconstruction.
@marter, the difficulty is choosing what is most important in the image … and in that respect, I am now feeling less pleased with the capture - I was probably trying to have my cake and eat it also. You certainly achieved the aim you stated: colourful windows and visible stone structure.
Have I missed anyone (apart from discussion about different software capabilities)? I hope not. Thank you, again, for your sharing your creative thoughts.
4 Likes
Had a lot of fun with this one, you can really play with the colors and go in pretty much any direction. My first thought was to go for an “etheral” high-key look.
RT5.8 dev.
2022-03-21_09-20-18.30_DSC6948-1.jpg.out.pp3 (23.4 KB)
3 Likes