How would you save this Rose?

[Iliah]

Subscribe to see EXIF info for this image (if available)

Image © Kate RK

 

[MGlennn]

You are the master.... :redface: .....but here are a few quick & humble attempts :smile:

Subscribe to see EXIF info for this image (if available)

Subscribe to see EXIF info for this image (if available)

Subscribe to see EXIF info for this image (if available)

Subscribe to see EXIF info for this image (if available)

these are all actions...the first is popartist, then digidoodle(mottled), then digidoodle(watercolor), then Dave's simplifier....all available at ActionCentral..

Thanks for all your help in the past, by the way... :biggrin:

 

[MGlennn]

Here's another way.....you actually tried to teach me this one over a year ago on DPR...I was too dimwitted to catch on then, but have been working on it since.....channel blending.....

original...

Subscribe to see EXIF info for this image (if available)

mixing blue into red channel with apply image command etc... :smile:

Subscribe to see EXIF info for this image (if available)

 

[Iliah]

Dear Mike,

Interesting and creative versions IMHO, thank you. Let's try to take this a step further.

Please look at RGB channels and see which one contains maximum level of details. We will derive some rule from this a little later. Taking that "rich" channel, try to bring up those details with a curve on that channel. After that, convert the source image to Lab and try to extract maximum details using curve on Luminosity channel. In both cases we need to concentrate on what tonal regions we are going to affect with the curves. Can you please publish source channels, curves, and resulting b/w images?

 

[beaucamera]

Wow, Iliah, this is an interesting experiment. I can hardly wait to see what Mike posts.

Virginia
aka beaucamera

P.S. I'm following along with you two. After I did what you said, my image appears darker than the original. Is this what one is to expect? I might add that I reconverted to Adobe RBG 1998 after the processing in LAB.

 

[Iliah]

Dear Virginia,

you should not be concerned with the darkness of the whole imageat this stage. We are coking here single channel for future manipulations. We need maximum details in that channel to fight the lack of details in petals in Red channel.

 

[MGlennn]

Dear Mike,

Interesting and creative versions IMHO, thank you. Let's try to take this a step further.

Please look at RGB channels and see which one contains maximum level of details. We will derive some rule from this a little later. Taking that "rich" channel, try to bring up those details with a curve on that channel. After that, convert the source image to Lab and try to extract maximum details using curve on Luminosity channel. In both cases we need to concentrate on what tonal regions we are going to affect with the curves. Can you please publish source channels, curves, and resulting b/w images?

OK....here we go again....tomorrow when I am at the office....best regards...M
:wink:

 

[RFCGRAPHICS]

My version of photoshop winging it.

1) selected the background and used autofx mystical lighting plugins to add yellow tint and color.

2) croped out right side of the image to remove the red leaves

3) sharpend up rose, burned shadows

4) used autofx photographic edges plugin to add acid edge to image

Subscribe to see EXIF info for this image (if available)

Regards

RFC

 

[MGlennn]

Dear Mike,

Interesting and creative versions IMHO, thank you. Let's try to take this a step further.

Please look at RGB channels and see which one contains maximum level of details. We will derive some rule from this a little later. Taking that "rich" channel, try to bring up those details with a curve on that channel. After that, convert the source image to Lab and try to extract maximum details using curve on Luminosity channel. In both cases we need to concentrate on what tonal regions we are going to affect with the curves. Can you please publish source channels, curves, and resulting b/w images?

Hi Iliah...OK here we go ...this is what I did yesterday as you showed me last year...will add the curve steps later.... :smile:

Original.....

Subscribe to see EXIF info for this image (if available)

First, as always, one wants to duplicate the original layer to create a layer to work on. Then apply USM to image as below....

Subscribe to see EXIF info for this image (if available)

Next.... set blending on this layer to luminosity....to do this just choose the luminosity option in box below....

Subscribe to see EXIF info for this image (if available)

Next, we want to create a new empty layer, as below...

Subscribe to see EXIF info for this image (if available)

this shows the new empty layer in the layers dialoge box....

Subscribe to see EXIF info for this image (if available)

Next we want to merge the visible layers into the new empty layer..

Subscribe to see EXIF info for this image (if available)

Next...open the channels dialogue box by just clicking on the tab and you will see 4 channels, the RGB, R, G, & B.....as Iliah will likely explain later, the red channel is lacking in detail....the blue channel appears to have the most usable detail...so....we will try to merge that channel into the red one...we do this with the apply image command .....select the red channel (make it active) because it is the "target" and in the apply image dialoge box select the blue channel as the "source".....change blending to "overlay".....opacity to 30% (or whatever you want)....click ok....and you are done.....

here's the red channel...notice lack of detail

Subscribe to see EXIF info for this image (if available)

green channel...

Subscribe to see EXIF info for this image (if available)

and blue channel....more details

Subscribe to see EXIF info for this image (if available)

here's the apply image dialogue...(notice red channel is active)

Subscribe to see EXIF info for this image (if available)

and the finished product.....

Subscribe to see EXIF info for this image (if available)

original again for comparisons...

Subscribe to see EXIF info for this image (if available)

Will leave to Iliah to explain the whys etc.....will also go back later and try to do as he says with a "curve adjustment layer" on the blue channel to get even more detail.......as always, comments/critique are welcome....enjoy :biggrin:

 

[Iliah]

Dear Mike,

Thank you for your time and efforts.

My feeling about photography is very much the same as of the work of a detective: you figure the place you need to be, you figure the best time to be there, then you go and take some evidence, and finally you spend some time processing those evidences.

So, with this Rose we discovered that: Green and Red channels are nearly worthless. They contain very little detail. That's bad, and as per Mike's suggestion we are going to borrow some details from populated Blue channel to Red channel, using channel blending and building up variation in that flat Red channel. While we on this, let's continue our detective meditation for a bit. Why not to try blending to the Green channel, too, to increase luminosity variation? Second, why not to create a simple mask to protect other parts of the image from changes? - Red or Green source channels, as well as Select:Color should provide us with the ways to create that mask quite easily. Third, why not to try to copy the Blue channel and to work on it, building more contrast and emphasizing details before blending Blue channel into others?

Important question is why Blue channel is the most populated with details. The colour of blue is the least wanted colour in this Rose ("unwanted color", as Dan calls it). We can imagine it whatever but blue. The rule of thumb is that the channel containing unwanted colour is what adds variation to the overall colour. Blank that channel out, and you will see image to become totally flat. For the skin tone, it is often blue that adds variation, too.

Further, that will be interesting to see how the information is spread across the channels if we convert the image to CMYK.

 

[MGlennn]

Dear Mike,

Thank you for your time and efforts.

My feeling about photography is very much the same as of the work of a detective: you figure the place you need to be, you figure the best time to be there, then you go and take some evidence, and finally you spend some time processing those evidences.

So, with this Rose we discovered that: Green and Red channels are nearly worthless. They contain very little detail. That's bad, and as per Mike's suggestion we are going to borrow some details from populated Blue channel to Red channel, using channel blending and building up variation in that flat Red channel. While we on this, let's continue our detective meditation for a bit. Why not to try blending to the Green channel, too, to increase luminosity variation? Second, why not to create a simple mask to protect other parts of the image from changes? - Red or Green source channels, as well as Select:Color should provide us with the ways to create that mask quite easily. Third, why not to try to copy the Blue channel and to work on it, building more contrast and emphasizing details before blending Blue channel into others?

Important question is why Blue channel is the most populated with details. The colour of blue is the least wanted colour in this Rose ("unwanted color", as Dan calls it). We can imagine it whatever but blue. The rule of thumb is that the channel containing unwanted colour is what adds variation to the overall colour. Blank that channel out, and you will see image to become totally flat. For the skin tone, it is often blue that adds variation, too.

Further, that will be interesting to see how the information is spread across the channels if we convert the image to CMYK.

Super info....thanks for explaining all this. :smile:

I have tried several other variations now (curves on blue channel, cmyk color, curves on green channel, etc....)...but....I am not that good at critically examining an image for color/detail deficiencies....so....can only tell which I "like" the best...not which is most "correct".... guess I should just post the end result of several different variations and see which one everyone else likes best....????.....comments...?????

 

[Iliah]

IMHO nothing can be "correct" about this image (and many others, too). All that matters is how your audience like it after correction is done.

 

[MGlennn]

IMHO nothing can be "correct" about this image (and many others, too). All that matters is how your audience like it after correction is done.

I agree totally.....thanks again for your insights... :biggrin:

 

[Iliah]

Going RGB route, just building up details, and presuming the original file was in sRGB, going from:

Subscribe to see EXIF info for this image (if available)

to:

Subscribe to see EXIF info for this image (if available)

For to how it was done, please have a look at:
http://www.rawmagick.us/RoseRGB_Blend.psd

 

[MGlennn]

Wow....now that's nice.....will look at the demo.... :smile:

 

[MGlennn]

OK....looked at it....only have CS at home (CS2 at office)....so may be able to see more there.....I think I can see all you did, but it would help to have a step by step instruction in order of what to do first....looks cool :smile:

 

[cwilt]

Iliah,

I was going to play with this last night but family matters pulled me away for to long. My plan was to use the L channel in LAB for blending since it seemed to have less noise than the blue RGB channel. Would that not work?

Kind regards,
Charles

 

[Iliah]

Dear Charles, that will certainly work. I do not know if that is interesting for anybody else, but in the spirit of detective work I'm exploring possibilities here. First one was pure RGB work. I think it is useful to compare results with Lab and combined Lab/RGB workflows on the image. Also, we are working here on JPEG, source is resized, and saved less then in top quality - yet blue channel is "underexposed". L channel will borrow the details from all three channels R, G, and B; and it will probably be easier to bring up the details there without grossly amplifying noise.

 

[cwilt]

Dear Charles, that will certainly work. I do not know if that is interesting for anybody else, but in the spirit of detective work I'm exploring possibilities here. First one was pure RGB work. I think it is useful to compare results with Lab and combined Lab/RGB workflows on the image. Also, we are working here on JPEG, source is resized, and saved less then in top quality - yet blue channel is "underexposed". L channel will borrow the details from all three channels R, G, and B; and it will probably be easier to bring up the details there without grossly amplifying noise.

I will let you do the detective work then. Putting in a long day at work, :frown: and I probably will not get to it for several hours.

 

[Iliah]

Dear Charles, I will rest for now for a littlle