Assignment - Making the subject stand out

rutt

Here is an example from my archives of an interesting subject, but a bad picture. It was taken at the Living Desert near Palm Springs, CA. The mountain lion's camauflage is working too well. She blends in instead of standing out. It's sort of the opposite of the distracting background from the last assignment. The cat also has a greenish cast, making it hard to differentiate her with some sort of curve steepening. Last assignment, we (I especially) learned how to make an effective selection, but I think the desired effect here is pretty subtle -- correct the cast and make the cat stand out.

As usual the image is a link to the original.

rutt

OK, I took a stab at this to get things rolling:



This took less work that I guessed.

1. Correct the cast on the cat with curves in LAB. Assymetric steepening of the A and B curves to get neutral colors on in the cat's nostrils and the white fur on her snout. [Oh, I do love LAB...]
2. Sharpen on the L channel as long as we are in LAB.
3. Crop.

hutchman

You've got me hooked!


My attempt.

Color corrected by setting the black adjust in levels
Neutralized the lightest color in the back ground.
Adjusted the midtones. Maybe a little too much red, but he jumps out!
Nothing else.




I think I like him!

Hutch

rutt

hutchman wrote:

he jumps out!

Hutch

You don't what him jumping out for real! Anyway, we both seem to have skinned this cat with different tools and the same effect. In this case, the cast in the original was so bad that it obscured the fact that the cat really isn't that well camaouflaged.

Since this turned out to be so easy, let's up the ante. This is an image that Dan Margulis used in his 2 day course. It came at the end as part of his "house of horrors." It's sort of what I was after, but the lion was too easy. This time, the camouflage really is good and not just a matter of cast.
[Dan, if your are out there, I hope it's all right that I posted this image here.] At least for the people in the class and for Dan himself, this is really a challenging image.

So, make the muledeer jump out:

cletus

Tried something quick and dirty:

1. Added a curves adjustment layer - set white and black points
2. Made a mask around the deer
3. Coppied the deer onto a new layer
4. Ran a Gausian blur of 1px on the background layer
5. Made a Levels adjustment to lighten the background layer

pathfinder

Rutt - Here is my shot at the cat - I color corrected by threshold and then dodged and burned to get the relative light levels more in line - then I selected the cat with a magnetic lasso, sharpened the cat in LAB and then inverted the selection to adjust hue/saturation in the grass - I dialed the grass down a little and then blurred in with Gaussian blur. Then after looking it over, I dodged the cat again to lighten it up some more. How'd I do?

pathfinder

rutt wrote:

Since this turned out to be so easy, let's up the ante. This is an image that Dan Margulis used in his 2 day course. It came at the end as part of his "house of horrors." It's sort of what I was after, but the lion was too easy. This time, the camouflage really is good and not just a matter of cast.
[Dan, if your are out there, I hope it's all right that I posted this image here.] At least for the people in the class and for Dan himself, this is really a challenging image.

So, make the muledeer jump out:

For this one again I adjusted coor balance by threshold and then selected the deer with a magnetic lasso quick and dirty. I increased the saturation of the deer and sharpened it and then inverted the selection and used Gaussian blur and adjusted the hue/saturation on the background. Still not too thrilled with the result, but the deer defintely does stand out more. So John , lets see yours now. We've showed you ours.

pathfinder

I got in too big of a hurry and I decided to do a little more dodging and burning - I am an old darkroom kind of a guy - and increase the saturation of the grass - Is this better or worse?

rutt

pathfinder wrote:

Rutt - Here is my shot at the cat - I color corrected by threshold and then dodged and burned to get the relative light levels more in line - then I selected the cat with a magnetic lasso, sharpened the cat in LAB and then inverted the selection to adjust hue/saturation in the grass - I dialed the grass down a little and then blurred in with Gaussian blur. Then after looking it over, I dodged the cat again to lighten it up some more. How'd I do?

Beautiful. You get an A.

rutt

pathfinder wrote:

I got in too big of a hurry and I decided to do a little more dodging and burning - I am an old darkroom kind of a guy - and increase the saturation of the grass - Is this better or worse?

This is certainly better than the original and better than anything I've ever been able to do with this image. I keep hoping there will be an approach to this that won't require a selection or other local manipulations. In this case, that just might not be true...

rutt

rutt wrote:

This is certainly better than the original and better than anything I've ever been able to do with this image. I keep hoping there will be an approach to this that won't require a selection or other local manipulations. In this case, that just might not be true...

I made a stab at this, and I still think it's hard. Here's what I have so far. Clearly there is room for improvement (Pathfinder's work is an existence proof of this.)

My approach was to focus on the deer and ignore the rest of the image at first. I wanted to make the deer look as good as possible. Later on I'd figure out what to do with the disasters this caused in the rest of the image. So I corrected and enhanced the colors with LAB curves; this time I used the L channel as well to enhance the contrast. The deer still didn't have enough detail so I moved the image to CMYK and blended the green channel from the original into the ugly yellow channel. THen I sharpened on the M+K channels. Now there is some detail in the deer:



So I was on track. The deer looks better, but the rest of the image is really ugly.

To my eye, the worst of it was the bright yellow patch of grass just to the right of the deer. I used selective collor to add some cyan to yellows, thereby making it more green. But the oversharpened tree branches were still terrible. I made a duplicate layer with a mask. I darkened the underlying layer and gaussian blurred it. (Probably this should have been done in reverse order?) Then I used the graiant tool on the layer mask to fade the topmost (unblurred) layer around the deer's head. The circle still included some distracting background areas, so I used a big soft brush to paint out some more of the unblurred layer. Here's what I got:



Not so great, huh. I think Pathfinder is currently the muledeer leader. I would probably have done better to have made selection or used the extract tool or something.

hutchman

It's a little early out here on the left coast and my eyes are barely open, but here tis...



This was really a tough one.

I used the extract tool with a lot of touch up.
Adjusted color in the background.
Burned the dear for some contrast with the background.
Made a levels adjustment on the dear.
Used a very slight gausian blur on the background.

?????????????????

Hutch

ian408

pathfinder,

The result for the cat is beautiful.

ian

rutt

hutchman wrote:

It's a little early out here on the left coast and my eyes are barely open, but here tis...



This was really a tough one.

I used the extract tool with a lot of touch up.
Adjusted color in the background.
Burned the dear for some contrast with the background.
Made a levels adjustment on the dear.
Used a very slight gausian blur on the background.

?????????????????

Hutch

Your deer is very pleasing and does stand out from the background. Perhaps you went a little to far? The contrast doesn't look completely natural to me. This deer's camouflage wouldn't really work, would it?

Still, it's much better than my efforts. I've spent a lot of time on this image - in Dan's class, after Dan's class, and just now. I find it pretty frustrating.

I have some of Dan's favorite efforts with this image, his own and his students collected over years of classes. I havent looked at them since last spring when I took the class (too depressing and who knows, maybe I'll have my own good idea.)
When we've all bored with this assignment, I'll post these and we can see how we are doing against the pros.

hutchman

Yeah,

I think he took a bath in a vat of red dye!

Hutch

cletus

hutchman wrote:

I used the extract tool with a lot of touch up.

I thought about using the extract tool also. I decided to just create the mask by hand. Lassoed most of the dear, created a channel and touched up the rest with a brush.

pathfinder

rutt wrote:

This is certainly better than the original and better than anything I've ever been able to do with this image. I keep hoping there will be an approach to this that won't require a selection or other local manipulations. In this case, that just might not be true...

I am at the office this morning and the thing thta really stands out is how bad these images look on my office monitor - It means that we have to be very careful in assessing color balance on the web. I know this is not news - but to my eye the deer still looks cyan this morning, but did not at home. I think many monitors tend toward the cool side. This one I am looking at is a Sceptre 19" LCD -- Fine for the web, but poor for correct color interpretation.

rutt

cletus wrote:

I thought about using the extract tool also. I decided to just create the mask by hand. Lassoed most of the dear, created a channel and touched up the rest with a brush.

Somehow I missed your original post. I think it's the best so far, actually. Your touch was light enough that the selection isn't intrusive. The deer does stand out a lot more and the colors are more natural than in the original (or in the other efforts.)

Still, this is a frustrating image. I wonder if anyone can actually make it look good (as opposed to just better.) Every time I work on it, I end up just wanting to cut the Gordian Knot and crop the hell out of it. But that would definitly be cheating, don't you think?

pathfinder

rutt wrote:

I have some of Dan's favorite efforts with this image, his own and his students collected over years of classes. I havent looked at them since last spring when I took the class (too depressing and who knows, maybe I'll have my own good idea.)
When we've all bored with this assignment, I'll post these and we can see how we are doing against the pros.

I've been thinking about the muley picture and the stated desire to make the deer stand out. But I am not sure that should be our goal. Deer that stand out do not last long in nature. I was just in Palo Duro canyon last month shooting pictures of deer at distances of less than 20 feet and deer definitely do not stand out. They DO blend in - I have a whole series of pictures that demonstrate how well they blend in. For example - and in this one the deer stands out MUCH more than most of the ones I took



I have not really done any post proceesing here to emphasize the deer - just threshold color correction and levels. Deer blend in very well. Please post the images you have from your class demonstrating the pros' success with your deer image Rutt.

Part of the problem besides color with the your deer image is that the deer is in front of the tree and her sillouhette is not clear, as a result the deer is hard to see and this cannot be easily corrected - UNLESS WE WERE TO CLONE IN NEW BACKGROUND! ( I know that I am yelling for effect here - this is an attempt at humor ) I know.....I Know.....I sound like I am repeating myself, don't I !!

DoctorIt to the rescue - pull out your background images of the woods in autumn and try cloning that in behind the muley......It might work!?

rutt

pathfinder wrote:

I've been thinking about the muley picture and the stated desire to make the deer stand out. But I am not sure that should be our goal. Deer that stand out do last long in nature...

You get an A again. I agree with you 100%. Your deer is beautiful and doesn't really stand out. THe #$%@% muledeer is a badly flawed image and the deer is doing what deer are supposed to do: be camouflaged. After wasting another hour or so on it, I looked and the versions that Dan included on the CD from his class, and I don't think anybody did any better with this image than you or Cletus did. Either the deer stand out too much and looks fake, or the background takes over the image and it looks ugly. I'll convert them and post them, but don't expect too much.

So maybe this image can't be saved or isn't worth saving. I remember concluding this after the last time I wasted a day on this image. But it would be great to be proved wrong, wouldn't it?

Anway, I think we have learned something. The cat was a strong image obscured by an ugly cast. Fixing the cast made the cat stand out. After that, Pathfinder got extra credit for making the cat extra double beautiful with some local moves. The muledeer image might be too poor to save. Pathfinder's deer doesn't stand out in terms of color, but it does in terms of composition. Also it has great depth and texture. So that image works.

I think we need some new images to work with. Does anyone have a problem image that doesn't work because the subject blends in with the background?

pathfinder

rutt wrote:

You get an A again. I agree with you 100%. Your deer is beautiful and doesn't really stand out. THe #$%@% muledeer is a badly flawed image and the deer is doing what deer are supposed to do: be camouflaged. After wasting another hour or so on it, I looked and the versions that Dan included on the CD from his class, and I don't think anybody did any better with this image than you or Cletus did. Either the deer stand out too much and looks fake, or the background takes over the image and it looks ugly. I'll convert them and post them, but don't expect too much.

So maybe this image can't be saved or isn't worth saving. I remember concluding this after the last time I wasted a day on this image. But it would be great to be proved wrong, wouldn't it?

Anway, I think we have learned something. The cat was a strong image obscured by an ugly cast. Fixing the cast made the cat stand out. After that, Pathfinder got extra credit for making the cat extra double beautiful with some local moves. The muledeer image might be too poor to save. Pathfinder's deer doesn't stand out in terms of color, but it does in terms of composition. Also it has great depth and texture. So that image works.

I think we need some new images to work with. Does anyone have a problem image that doesn't work because the subject blends in with the background?

You want images with subjects that do not stand out......hmmmm... Perhaps deer in the brush might work...... I bet it might

Some of my shots I have not posted because I thought they were of too poor quality - Shot as jpgs, not RAW - (my CF chip was almost full when I saw these rascals and so I switched to jpgs rather than taking the time to change chips. If I had to tried to change chips the deer woud have been long gone. ) I was trying to crawl - sneak through the brush as you can see from the second image. I have done no post processing of these two images other than to color correct by threshold and add a border and my coyright data .......
If you wish to edit these to make the deer stand out more be my guest.

rutt

pathfinder wrote:

You want images with subjects that do not stand out......hmmmm... Perhaps deer in the brush might work...... I bet it might

These both seem likely, but it's frustrating to work on reduced sized images. Please, can you post more pixels? It's OK if you use lower quality jpegs, but it's easier to work with more bits.

pathfinder

rutt wrote:

These both seem likely, but it's frustrating to work on reduced sized images. Please, can you post more pixels? It's OK if you use lower quality jpegs, but it's easier to work with more bits.

The images are about 800Kb and 500Kb and the originals are available at http://pathfinder.smugmug.com/gallery/63356/8

rutt

OK, I'll bite.



I'm not really sure this is an improvement, but it gave me an opportunity to try out a trick that one I learned from one of my fellow students in Dan's class. I used a gradiant in a layer mask to separate the deer from the background. Then I could apply different curves and unsharp mask/gaussian blur. It might be kind of a hack, but in this cases, I think it mimics the experience of focusing on the deer in a complex environment.

This isn't really my kind of "photoshop sheneniagan". I liked the lion, which was kind of a puzzle. Once you had the key (removing the cast) it was possible to get an excellent result. I'm not sure about this image. Either it was excellent as shot or there is no hope for it. De gustibus...

wxwax

pathfinder wrote:

How'd I do?

Superbly, I'd say.

pathfinder

rutt wrote:

OK, I'll bite.



I'm not really sure this is an improvement, but it gave me an opportunity to try out a trick that one I learned from one of my fellow students in Dan's class. I used a gradiant in a layer mask to separate the deer from the background. Then I could apply different curves and unsharp mask/gaussian blur. It might be kind of a hack, but in this cases, I think it mimics the experience of focusing on the deer in a complex environment.

This isn't really my kind of "photoshop sheneniagan". I liked the lion, which was kind of a puzzle. Once you had the key (removing the cast) it was possible to get an excellent result. I'm not sure about this image. Either it was excellent as shot or there is no hope for it. De gustibus...

I'd like to know more exactly how you did the layer mask - One of the reasons I thought this might be a tough imge is that it is so difficult to paint over the deer or around the deer because of all the brush in the way. I think you have done a commendable job with a difficult subject. Here's an attaboy!

Isn't it amazing how the autofocus picked out those eyes in the midst of all that brush? Just amazing. But they are dead center and that is the ONLY focal point I permit my 10D to use.

And now I think this is a pretty nice image - one that makes you look at it a little while to decide just where the deer is and the brush isn't.

rutt

pathfinder wrote:

I'd like to know more exactly how you did the layer mask - One of the reasons I thought this might be a tough imge is that it is so difficult to paint over the deer or around the deer because of all the brush in the way. I think you have done a commendable job with a difficult subject. Here's an attaboy!

Isn't it amazing how the autofocus picked out those eyes in the midst of all that brush? Just amazing. But they are dead center and that is the ONLY focal point I permit my 10D to use.

And now I think this is a pretty nice image - one that makes you look at it a little while to decide just where the deer is and the brush isn't.

About the image: Yes, I noticed how nice the focus was and wondered at it. Keeping that one focus point enabled alone worked well for you here. It seems it would have some downsides, though. And what if you had a fancy pro level camera with a zillion focus points? How would you use them? I've been thinking it might be useful to become adept at switching FPs. I think Minalta once made cameras that were supposed to detact where you were looking...

About the correction: I took a trick from your book to make the layer mask. (Remember the dog?) I made a white layer mask and used the gradiant tool with the radial gradiant option to paint white round gradiant centered on the deer. I had to play around for a while with the customized gradiant to get it to fade out the way I wanted. I still couldn't tell you exactly how this works. Here is the top layer alone so you can see how it fades (to white in this case, but photoshop it fades to transparent, because I made the gradiant in the masK).



Does this answer your question?

lynnma

[I had a quick go at the deer... NOW THAT I'VE DISCOVERED THE "EXTRACT" FUNCTION!!!! after all this time.. I really should go to school or something... I'll have another go at it tomorrow when I have more time...

pathfinder

rutt wrote:

About the image: Yes, I noticed how nice the focus was and wondered at it. Keeping that one focus point enabled alone worked well for you here. It seems it would have some downsides, though. And what if you had a fancy pro level camera with a zillion focus points? How would you use them? I've been thinking it might be useful to become adept at switching FPs. I think Minalta once made cameras that were supposed to detact where you were looking...

The Canon EOS 3 ( and probably the 1 as well) does have the facility to chose the aiming focal point by evaluating where your eye is looking. You have to do some training with it - but that is what Canon says it will do. I have no experience with it tho so I can not swear to its accuracy.

I tried switching FPs but I find it much slower than locking focus and recomposing - works for me usually...

pathfinder

rutt wrote:

OK, I'll bite.



I'm not really sure this is an improvement, but it gave me an opportunity to try out a trick that one I learned from one of my fellow students in Dan's class. I used a gradiant in a layer mask to separate the deer from the background. Then I could apply different curves and unsharp mask/gaussian blur. It might be kind of a hack, but in this cases, I think it mimics the experience of focusing on the deer in a complex environment.

This isn't really my kind of "photoshop sheneniagan". I liked the lion, which was kind of a puzzle. Once you had the key (removing the cast) it was possible to get an excellent result. I'm not sure about this image. Either it was excellent as shot or there is no hope for it. De gustibus...

HI Rutt.
I was wandering around my files and found a couple more deer shots that I had forgotten ... see if you can count the deer - these were shot on the run with a 28mm lens - but if I had tried to change lenses they would have been in the next county - these are Hoosier deer, not Texas deer and they know when it is time to leave....Just to show that deer can be hard to see in the woods....





There are at least 2 deer in each picture - perhaps more as the herd was at least 6 or 8.

rutt

pathfinder wrote:

HI Rutt.
I was wandering around my files and found a couple more deer shots that I had forgotten ... see if you can count the deer - these were shot on the run with a 28mm lens - but if I had tried ts change lenses thwy would have been in the next county - these are Hoosier deer, not Texas deer and they know when it is time to leave....Just to show that deer can be hard to see in the woods....





There are at least 2 deer in each picture - perhaps more as the herd was at least 6 or 8.

Looks like the subject is the sign.