3 Likes
night.scene.bar.pp3 (19.7 KB) RawTherapee 5.8 Development
Nice shot to work on with just the right amount of challenges. Everybody seems to be struggling with the church, which is to be expected I guess…
You, @gadolf , mentioned the yellowish street lights coming in. I would have left the scene more orangey if you hadn’t mentioned that. Not having been there I can only guess and sideways peek at your edit to get a feel about that part.
2 Likes
I tried to neutralize the blue color cast. And think it would have been much easier that this area was not overexposed. In any case, this was a nice exercise.
DSC_8666.nef.xmp (22,9 KB)
darktable 3.7.0~git702.851a4ac2b7-1
3 Likes
I have tried to make the inside neutral not disturbing the blue church
DSC_8666_01.nef.xmp (15.1 KB)
4 Likes
I’d like to improve the luminance clipping method in QRP. But according to Rawtherapee and darktable, the models they use a different.
Let’s assume the upper threshold is U, the lower threshold is L
| Software | Shadow | Highlight |
|---|---|---|
| Rawtherapee | R < L and G < L and B < L | R > U or G > U or B > U |
| darktable | same as Rawtherapee | R > U and G > U and B > U |
Which one is used mostly? Or maybe I shall implement both?
1 Like
This is great 
Reminds me a lot of Edward Hopper’s paintings.
Thanks for sharing all your different interpretations.
To start with, the church is illuminated by a very strong blue light, and @Rajkhand’s gets closer to it, although it lacks the purple accent present in mine. Next time, maybe I follow @paulmatth’s advice, or ask the priest to shutdown the lights for a while 
As for the interior, although the street lights (tungsten?) spread a strong yellowish tint, I think I failed it in my edit. Last night I got back to the scene and the interior is not as yellow, dark and flat as in my edit, so I would stay somewhere between @Rajkhand’s and @paulmatth’s, if we’re trying to go the realistic depict way.
In the end, when I thought I was mastering raw editing, I realize that I have to get back to Introductory Edit Class again 
. That’s fine, a sign that things evolve.
I’m not sure if “mastering raw editing” exists, as raw editing always involves interpretation and personal taste - and monitors! 
Be happy that things evolve for you, it’s always good to learn new things or to do existing things (your raw skills) better.
2 Likes
I was trying to change the annoying purple hue into something more yellow-ish in RT, but failed miserably!
DSC_8666_RT-1.jpg.out.pp3 (15.7 KB)
1 Like
Unless you want an artistic depiction, purple-blue is what that light looks like at night, more a less like my edit.
For reference, I took this minutes ago while walking the dog (cheap smartphone). They left one of the bulbs turned on, it might give a clue to what the hue might look like (look at the bulb borders)
2 Likes
Oh sorry, I thought it comes from the difference in colour temperature between inside the bar and church illumination 
Do they use blue tinted LED light?
No need for sorry, after all, playraw is a free style territory. 
As for the type of bulb, I don’t know nothing about types (tried to google but couldn’t determine for sure what type is this one). One thing is certain: the hue is strong purple-blue at night.
1 Like
Nice edits, @Zbyma72age , very clear, although you’ve eliminated the purple blue church hue. (which is probably the right decision aesthetics wise)
I can’t speak for Ingo or any RT code, in dt the “highlights reconstruction” is clearly a weaker point. The clip algorithms works ok in many cases (at least i think so) but it’s definitely not reconstructing. The “reconstruct color” is very prone to generating artefacts.
I have just started a fresh and “clean room” implementation for a hopefully much better color-propagating algorithm, let’s see. (I’ll give you a hint if there is something noteworthy)
About your question in an exact way, i don’t know about the history of the RT & dt algos and have no strong opinion on any advantage.
1 Like
Thanks, @hannoschwalm I should have been more specific, sorry.
The point is the new culling app @qdwang is building (QuickRawPicker, see this thread), where I found inconsistencies between QRP’s raw clipping indicator and dt’s. It’s not about highlight recovery, just the clipping overlay in both software.
EDIT: I think @qdwang is trying to change his code to match dt, but he noticed that dt determines clipping different from RawTherapee, so he asked which way I suggest him to go (but I’m not talking for him, just what I think is going on)
2 Likes