Topic: What to do with this church image?

[Greg Groess]

thnx



I did look a bit a your image, Hope you don't get mad. Though i think you have  a green cast (can be very good seen) next to that small protrait on the stove. Though that green cast is every where .

The walls are in fact Green. As is the chair and the glass of the table..note the white towel and the visual contrast between it and the chair. The white trim on the fireplace and the upper trim around the room are most obvious places to look for a neutral white.

It goes all over the image. See the table and the couch left  f.i. I think you did perhaps focus to much on the painting instead on the WB and WP / BP edit : just saw you also got a slight blue cast on the (dont know the English word) that thing around the stove.

The skin tone pin on the portrait is more of a landmark for finding correct color.  With all the cross casts in the image the portrait should look good.  It is a place the eye will expect correctness.  The fireplace measures black using the hue clock in Curvemeister.

Also a yellow / green cast on the walls next to the doors.  I am sure those walls are white as can be.

See note above re: the real color of the walls...Sorry but they are Green...

Also you seem to have a slight red cast on the sealing.

That is the curtains reflecting on the ceiling

I had problems there too. Though as try to fix WB the wall doe give us a white point. If one does have cast still there one needs to check more imo. I think value's are of a great help there. Also i feel that the devastating effect casts often is underestimated. Talking about a hammer  ;-) Cast sure is one. I have learned, first to get WB WP BP right, work from there.  Or if one does go for artistic affects np. Though for a starter like me it feels the way to go.

You are correct in getting the Color balance figured out but when your starting point is so far off as in the original it is difficult to correct this image without an "rough" correction first to get you close.  You need to choose better memory objects for determining the neutral points.  If the trim of the room is white and the walls have tone there must be color in the walls that belongs there. 

Btw on this image one needs to be careful or one blows the highlights out side. Btw to prevent it i did use for EV correction not LAB though CMYK K channel.

The exterior is not "that important to the overall image...It is about the room interior and that is where most of the correction is focused.  The Exterior could be brought back into the image with a mask and a multiply layer.  It was not the main point of my correction and visually I had no problem with leaving it too bright.  It provided a nice counter balance to the dark fireplace.

It aint that easy.

Agreed...but it is also not impossible...

i did try an another shot.



i think., i am getting closer. Thnx to you i did also check more on the photo above the stove. I think i got it ok now. (was not so for sure)  Lab is indeed a hammer, though it feels good to me, for sure on WB fixes., i think , i hope i just have to practice more.

Lab is certainly the easiest color space to think in.. It is considered a perceptual space and is very much in line with how we "see" color.  It is on the other hand severely limited in how it handles mixed lighting.  It would not be my first choice in color correcting an image this far out of balance.

Must i am after a relook not to happy also on what i have now. The floor is indeed to much desaturated. , it aint easy

Does your program allow you to use RGB color space?  if so try to make adjustments in that space first to get the image closer to the correct balance..then try LAB to make minor adjustments...

Keep at it...Keep asking questions.

Greg

[headlights]

Hey Greg,

You got me on a better track.

When using LAB (still feel it is great to fix WB issues, and my idea to move the center points according to marker values can be a great assist, if not a very good to perfect way to deal with the issue) One has to be sure to choose the correct points of correspondence. As you did point to me. Thnx!!! A big help is do look during the process very well to what the light is doing in the picture. Be aware of false assuptions (as i did first by thinking the walls should be white. You are right they are not! As you said the color differences are very subtile.  I have values though miss your color clock. To bad the clock made by Chris also does not work on linux. :-(

The  couch and the walls. To my current findings it has some slight greens and some slight yellows in it.

So i did do it over again working with three reference points.

1 Did start with cover on the stove
2 A point at hard point on the ceeling where also the red cast can occur.
3 A point a the grass in the picture on the wall above the stove

I think i am very close now. There one color in the image that does bother me now. And that is the brown in the floor.  If i look at that it tells me i am close but not really quite.

I tells me i went to hard on the reds, (prolly correcting the sealing) ?

See:




We did pick a hard image to get real good. Though those are the once one gets better by.

For me i rather start in LAB and go to CMYK for the final touch. When one needs to be real subtile Luminance blending CMY channels seems to me then just great. (or if one does like RGB in RGB, though i am no RGB dude) That man from mars methode that other guy (http://livestre.am/DSAO) did show, does blow my mind off. He sure is on track there imo.

Though some how his way does not work in the program i use

It is atm still very beta, though if you want to check it. a windows  version is here. http://code.google.com/p/delaboratory/downloads/list (export option does not work there, fixed in svn) As said i use linux svn version .

Post script:

I did check on the value's have now on the shadows on the portraits on the stover cover. (how is that one called in proper English ;-)

those should have imo a close to natural grey?

I know the guy who did make the shoot somehow, i could ask him about the wall colors as we only have educated guesses. You feel it dif, though i think i even still have a green cast. (color overlay that not should be there and covers other colors)


Btw on LAB beeing a hammer, I do not know or CM does have a curve fine control by f.i keyboard control. Aka for the very small moves on a curve a kinda micro adjustment?

I did try software that has Ctrl + mouse wheel for per pixel up or down and Shift + mouse wheel for per pixel left and right. It does not support layers and values. :-( So they have a great micro adjust though one still walks blind fouled.

Btw i have a European background. so it might i just look diff couse of that.

That is also why i like this methode, it goes by values and light. all is free from there.

[Greg Groess]

I might have to try this in LAB and see if I can get closer to your thought process...

Greg

[Greg Groess]

I know that for some of you this is going too far from the original posting but highlights has asked a good question and deserves a good answer....

To take the Lee Varis image he linked to through a similar process....

Shot 1 is the original Image with CM hue clocks on "critical" areas.  I put this inquotes because the choices are somewhat subjective but reflect the goals you should be having for your images...
I have placed hue clocks on the faces for skin tones and on the watch band and head band as target black areas.  I do not know if the head scarf is white but for now I would guess that it is not.  I also mark some highlight areas in the background so I can get a scope of how severe the cast is out there.  Notice that the skin is off colored and the shadows are cold <too Blue> 

Note:  This picture is about the people.  The background will be adjusted as much as possible to make it look better but I do not know what color the glass behind them is and if it is causing the color shift to be more severe than I would think it is.

Shot 2 is the image with the adjustment curves. 
I applied no masking to this image I simply used RGB to adjust the colors based on the brightness range which is RGB's strongest feature.  I reduced the blue overall in the image and made the people <the main subject> look better.  I did not try to remove all of the color from the background I simply muted it a bit. 

Shot 3 is the compare image the correction is on the left and the original is on the right.

Some thoughts...

The technique used by Lee Varis and highlights is the opposite of the color expansion process we are beginning to teach people.  Using LAB to kill off the color cast is a very dangerous way to go down a road that is better driven with RGB.  The loss of yellows and blues is especially visible since most of us are very yellow blue sensitive.  There is a loss of color fidelity that is most disturbing to the eye and poorly restored at the end.  While we might have to resort to these kinds of correction in a rare picture; they are not good mainstream techniques for general color correction.  I fear that the loss of color fidelity is written off in the goal of getting rid of any color cast.

In his final image which I'll attach in the next post he has killed off the color in the background to emphasize he people and illustrate the technique.

Greg

[Greg Groess]

Final Lee Varis Image from his website...

[headlights]

what is your final image?

i suppose the left one above the final of Lee Varis?

If i see a pop up of the people, not addressing the cast issue

Lee Varis just did kill, all colors there in the back ground. Quick but very dirty way imo. He did offer it as an easy way for newbs. I would not do it that way.

The issue here is how to deal with cast due to camera WB error. in a complex light situation. What Chris did on his one was on that image i dirty though very effective way to get it done.

The bandoline image. i did remove the cast and got, i think, pretty much was there when shot.  Aka removed the cast. I did not work to get the image good looking as it is not my image. 


btw here the original image:



Imo as Dan Margulis would say a bad original. The photographer did almost al wrong he / she could do. Wrong WB and wrong DOF

Lee did first use PS to get the WB at the people instead at the background as the camera did. Aka he did go just to a cooler WB temp creating a blue cast, which he simple did desaturate.

I really think this can be excellent  handled in LAB. To repaint to new intrepretation in RGB is not where i would feel fine.

Here what i do suggest to get a proper start:



If you want you could indeed go further in CMYK. to work with the faces. RGB curving is the last i would do

with a simple overlay one could pop up f.i

getting in only two moves this.


though i feel the WB issue is solved rather well

and work in CMYK on fine tuning.
What is your approach and result? Curving in RGB from the start is last i would do. But that is perhaps because i am more a LAB (basics) and CMYK (faces) HSV / HSL  (saturation) dude. Again what that guy on the movie did with man of mars, does blow my mind.

Luminace curving , i love it. So subtle and powerful for final touch.  I really love the idea, the subtility and the power of it as he did show on his movie. awsome

 Chris did post on his website also some astonishing  results with Man from Mars on WB issues.

[headlights]

 I did went back to the beginning of this topic the church of chris.

i did apply my idea's of LAB WB adressing combined with the idea's of luminocity blending that guy did show in his movie

giving me this.



btw what you did on that livingroom is pretty awesome.

you did way better then i did.

[headlights]

Good luck guy's

[Greg Groess]

The church is looking much better.  Thanks about the living room.
I do hope you come back and visit us we never try to burn a bridge or shut a door.

Greg

[headlights]

Greg,

Thnx , good luck with your show nxt week.