OK thanks. I wasn't sure what to think of the final histogram. Why would photoshop change scale on the histogram?
The image definitely looks better, at least in my opinion.
It probably changes the scale to try to accomodate the spike of pixels at the far left end of the final histogram and that made everything else look lower. The area under the histogram is constant at the same scale so when the area suddenly looks a lot smaller, Photoshop must have changed the scale.
another question. How can I learn how to use Blend if setting?
I've been slowly going through the book and wanted to get through all of this thread before jumping in. I'm becoming an LAB convert. Must be some sort of religion.
I read no more than 2-4 pages a day and many are re-read and re-read. I'm somewhere around page 225.
It took me a while to figure out Blend if. While I'm still not comfortable thinking/applying it, this is the way it works.
First, you can select either the L, a or b channels to apply.
Picture two layers; one above the other. The top layer is all you can see.
You'll notice that the end points of the upper and lower sliders are spread to their maximum width.
By pulling in either end point of the top slider towards the other side, you allow the bottom to peek through anywhere outside the two points on the sliders.
Splitting an end point merely makes the transition from top layer to bottom layer smoother.
Now, moving one or more of the end points of the lower slider works differently. When you move an end point of the bottom slider, you force anything not between the end points of the bottom to take precedence over the top layer.
Now, an example. If you have two layers. On the top layer you have a green letter "T" and a magenta letter "T". On the bottom layer use the same colors but the letter "B" in both colors. Top letter directly over bottom letters.
Call up the Layers Options (double click the layer). Select channel "a" (Green/Magenta).
You should see to letters "T" (green and magenta).
If you slide the top left hand end point to the middle, you should exclude the top green color and the green "B" should show through (you excluded anything outside the top end points. Now slide that end point back to the left end. You should see both letters "T".
Slide the bottom left slider to the middle. Now, you should see a green "B" and megenta "T" (you forced the bottom green letter to show through by forcing anything outside the two end points to take precedence).
The top slider is used to allow the bottom to show through outside the end points and the bottom slider forces the bottom to show through outside the end points. To me that sounds like reverse thinking.
I'm blown away. I just ordered the book after discovering its existence in anothewr thread about blown out sky fixes. Today, I discovered this thread. What a great resource! Thanks all of you who contributed to this effort.
converting to B&W
hi guys, i have the book. can i ask a simple question. In chapter 6, Dan outlines a summary on how he created his B&W version of the buildings. This is where he apply imaged to the red channel and then made the green channel a bit lighter and then he said it converts to the b&w version.
when he said it converts, what does he actually mean? what do you do to convert? do i just go to image -> grayscale?
, it is those every day words used in a new context that can really confuse a person.
Glad to have read someone else with problems. I think my biggest problem in this stuff is learning the vocabulary. If I had it all down, I could zip thru a book or a tutorial.
A perfect LAB example...
If anyone wants a classic example of an image begging for a LAB retouch, go check out this thread. The image isn't mine or I'd post it here, though I offered a classic LAB retouch on it in the thread.
Grab the original from that thread and try it yourself. You only need to know chapter 1 of Dan's LAB book to do really nice things to punch up this image.
The image starts out with low contrast and pretty dull colors, exactly what LAB is really good at punching up.
If anyone wants a classic example of an image begging for a LAB retouch, go check out this thread. The image isn't mine or I'd post it here, though I offered a classic LAB retouch on it in the thread.
Grab the original from that thread and try it yourself. You only need to know chapter 1 of Dan's LAB book to do really nice things to punch up this image.
The image starts out with low contrast and pretty dull colors, exactly what LAB is really good at punching up.
Hi John,
I'm playing around with this image and also saw your take on this image in that thread you posted above. My colors are not as bright as yours. I was wondering by how much you brought in your A & B channels. Did you do it symetrically? I'd post my take here but I don't think that is a kosher thing to do since it is not my image, right? I brought them in by 10, then 15 then 20.
I'm playing around with this image and also saw your take on this image in that thread you posted above. My colors are not as bright as yours. I was wondering by how much you brought in your A & B channels. Did you do it symetrically? I'd post my take here but I don't think that is a kosher thing to do since it is not my image, right? I brought them in by 10, then 15 then 20.
Gracias,
I brought them in quite a bit on that image, more than I usually do because of the character of the image. I don't remember exactly, but I suspect it was probably around 30% on each end, perhaps even more. You just have to use your own eyes to judge how far to go. What you need to avoid is unrealistic colors that either look like an artificial color or look too brilliant to fit in the scene. I usually start with 20 in on each end and then back off if that creates something unrealistic or go more if that's not enough.
For images that do not have a color cast and you want to keep them that way, you will want to keep the middle of the curve fixed in the middle because that keeps neutral things neutral. The easiest way to do that is to just bring the two ends in evenly and that's what I did on that image.
The reason this image is so perfect for LAB is that the colors in the original are very dull and there are no bright colors anywhere in the image. If you look at the histograms for the A and B channels in LAB mode, you can see how all the pixels are concentrated right near the middle of the range. The symmetrically steep A and B curves just spread those colors out across a wider color range. When you combine this color enhancement with a bit more contrast on the L-channel, you can make a startling difference.
A more common case than this image is that there are some bright colors and some dull colors in the image. In that case, it's a little more difficult to keep from overdoing the bright colors while still enhancing the dull ones, while still keeping everything realistic. That typically requires some more than symmetric curves and is the subject of many of the more advanced chapters in Dan's LAB book. I mentioned this image because the simple techiques work so well on it.
I think you're probably right that we shouldn't post derivatives of that image here without the author's permission. In my experience non-pros are usually pretty good about granting permission for this type of educational use when asked if you ever want to do that.
I brought them in quite a bit on that image, more than I usually do because of the character of the image. I don't remember exactly, but I suspect it was probably around 30% on each end, perhaps even more. You just have to use your own eyes to judge how far to go. What you need to avoid is unrealistic colors that either look like an artificial color or look too brilliant to fit in the scene. I usually start with 20 in on each end and then back off if that creates something unrealistic or go more if that's not enough.
For images that do not have a color cast and you want to keep them that way, you will want to keep the middle of the curve fixed in the middle because that keeps neutral things neutral. The easiest way to do that is to just bring the two ends in evenly and that's what I did on that image.
...
Hi,
Ok - I increase the steepness and it worked better. I guess I thought I was going too far but in reality I wasn't going far enough. So much to learn...
Try going overboard on the A and B curves, way beyond what you think would be any good. Then dial it back using the opacity slider for the layer. This is a simple way to get a feel for how far you can go. Dan says that he is using this technique more and more in his actual work.
Try going overboard on the A and B curves, way beyond what you think would be any good. Then dial it back using the opacity slider for the layer. This is a simple way to get a feel for how far you can go. Dan says that he is using this technique more and more in his actual work.
Duffy
Cool! I hadn't thought of that. I'm going to try that from now on.
Is Anybody Still Here
I'm not sure anybody is still here but I'll go ahead anyway.
Following reading the LAB canyon book, LAB has become my colorspace of choice for my standard corrections.
My regular approach is to:
Go into LAB
Steepen appropriate portion of L if there is not already a great deal of contrast.
Steepen A and B in the range of 10-15% (I find more usually results in over-saturated colors)
Select the L channel and:
Highlight/Shadows to bring back area that are very dark and reduce the highs. I find that reducing the highs is the most important part as it remoze haze and glare and add crispness to the end result.
Add some sharpenning to the L channel.
= = = = = =
I now place a high degree of importance on the Highlight portion and am considering moving this to the first step.
- - - - - - -
Any comments and experiences would be much appreciated.
I just tried the LAB method and it works very well.
But how can I export an image to say sRGB for the Web?
When I try to do it all the color glory gets washed away.
What am I doing wrong?
“To consult the rules of composition before making a picture is a little like consulting the law of gravitation before going for a walk.”
― Edward Weston
I just tried the LAB method and it works very well.
But how can I export an image to say sRGB for the Web?
When I try to do it all the color glory gets washed away.
What am I doing wrong?
Are you using Convert to Color Space or Assign Color Space????
I simply saved as jpg then tried to select sRGB colorspace
(like LAB via menu). How can I convert to a colorspace like
you mention?
“To consult the rules of composition before making a picture is a little like consulting the law of gravitation before going for a walk.”
― Edward Weston
When you have an image in the LAB color space, you need to go to Image>Mode>RGB.
Next, unless your working space is sRGB, you need to go to Edit>Convert to Profile> sRGB.
Then you can save your image that is destined for the web.
Now it works. Thank you very much!
“To consult the rules of composition before making a picture is a little like consulting the law of gravitation before going for a walk.”
― Edward Weston
“To consult the rules of composition before making a picture is a little like consulting the law of gravitation before going for a walk.”
― Edward Weston
Comments
Homepage • Popular
JFriend's javascript customizations • Secrets for getting fast answers on Dgrin
Always include a link to your site when posting a question
See chapters 7 and 8 especially. My summary of chapter 8 might help.
I've been slowly going through the book and wanted to get through all of this thread before jumping in. I'm becoming an LAB convert. Must be some sort of religion.
I read no more than 2-4 pages a day and many are re-read and re-read. I'm somewhere around page 225.
It took me a while to figure out Blend if. While I'm still not comfortable thinking/applying it, this is the way it works.
First, you can select either the L, a or b channels to apply.
Picture two layers; one above the other. The top layer is all you can see.
You'll notice that the end points of the upper and lower sliders are spread to their maximum width.
By pulling in either end point of the top slider towards the other side, you allow the bottom to peek through anywhere outside the two points on the sliders.
Splitting an end point merely makes the transition from top layer to bottom layer smoother.
Now, moving one or more of the end points of the lower slider works differently. When you move an end point of the bottom slider, you force anything not between the end points of the bottom to take precedence over the top layer.
Now, an example. If you have two layers. On the top layer you have a green letter "T" and a magenta letter "T". On the bottom layer use the same colors but the letter "B" in both colors. Top letter directly over bottom letters.
Call up the Layers Options (double click the layer). Select channel "a" (Green/Magenta).
You should see to letters "T" (green and magenta).
If you slide the top left hand end point to the middle, you should exclude the top green color and the green "B" should show through (you excluded anything outside the top end points. Now slide that end point back to the left end. You should see both letters "T".
Slide the bottom left slider to the middle. Now, you should see a green "B" and megenta "T" (you forced the bottom green letter to show through by forcing anything outside the two end points to take precedence).
The top slider is used to allow the bottom to show through outside the end points and the bottom slider forces the bottom to show through outside the end points. To me that sounds like reverse thinking.
Whew!!! I think I need something to drink.
http://webwiz.smugmug.com
Does everyone always have to flatten their images before moving on to RGB again? To me this seems like a disadvantage? No?
http://www.samuelbedford.com
hi guys, i have the book. can i ask a simple question. In chapter 6, Dan outlines a summary on how he created his B&W version of the buildings. This is where he apply imaged to the red channel and then made the green channel a bit lighter and then he said it converts to the b&w version.
when he said it converts, what does he actually mean? what do you do to convert? do i just go to image -> grayscale?
thanks
thanks.
Yes, he just does Image->Mode->Grayscale
Glad to have read someone else with problems. I think my biggest problem in this stuff is learning the vocabulary. If I had it all down, I could zip thru a book or a tutorial.
ginger
Ive gone back to chapter one....
Had borrowed the book fro te library but decided to get my own copy.
Late to the aprtty but hope I can still join in
...pics..
If anyone wants a classic example of an image begging for a LAB retouch, go check out this thread. The image isn't mine or I'd post it here, though I offered a classic LAB retouch on it in the thread.
Grab the original from that thread and try it yourself. You only need to know chapter 1 of Dan's LAB book to do really nice things to punch up this image.
The image starts out with low contrast and pretty dull colors, exactly what LAB is really good at punching up.
Homepage • Popular
JFriend's javascript customizations • Secrets for getting fast answers on Dgrin
Always include a link to your site when posting a question
Me too! I just read the first 2 chapters and my head is spinning.
blog: www.rosaurasandoval.com
galleries: www.rosaura.smugmug.com
Hi John,
I'm playing around with this image and also saw your take on this image in that thread you posted above. My colors are not as bright as yours. I was wondering by how much you brought in your A & B channels. Did you do it symetrically? I'd post my take here but I don't think that is a kosher thing to do since it is not my image, right? I brought them in by 10, then 15 then 20.
Gracias,
blog: www.rosaurasandoval.com
galleries: www.rosaura.smugmug.com
I brought them in quite a bit on that image, more than I usually do because of the character of the image. I don't remember exactly, but I suspect it was probably around 30% on each end, perhaps even more. You just have to use your own eyes to judge how far to go. What you need to avoid is unrealistic colors that either look like an artificial color or look too brilliant to fit in the scene. I usually start with 20 in on each end and then back off if that creates something unrealistic or go more if that's not enough.
For images that do not have a color cast and you want to keep them that way, you will want to keep the middle of the curve fixed in the middle because that keeps neutral things neutral. The easiest way to do that is to just bring the two ends in evenly and that's what I did on that image.
The reason this image is so perfect for LAB is that the colors in the original are very dull and there are no bright colors anywhere in the image. If you look at the histograms for the A and B channels in LAB mode, you can see how all the pixels are concentrated right near the middle of the range. The symmetrically steep A and B curves just spread those colors out across a wider color range. When you combine this color enhancement with a bit more contrast on the L-channel, you can make a startling difference.
A more common case than this image is that there are some bright colors and some dull colors in the image. In that case, it's a little more difficult to keep from overdoing the bright colors while still enhancing the dull ones, while still keeping everything realistic. That typically requires some more than symmetric curves and is the subject of many of the more advanced chapters in Dan's LAB book. I mentioned this image because the simple techiques work so well on it.
I think you're probably right that we shouldn't post derivatives of that image here without the author's permission. In my experience non-pros are usually pretty good about granting permission for this type of educational use when asked if you ever want to do that.
Homepage • Popular
JFriend's javascript customizations • Secrets for getting fast answers on Dgrin
Always include a link to your site when posting a question
Hi,
Ok - I increase the steepness and it worked better. I guess I thought I was going too far but in reality I wasn't going far enough. So much to learn...
Thanks for the help!
blog: www.rosaurasandoval.com
galleries: www.rosaura.smugmug.com
Duffy
Cool! I hadn't thought of that. I'm going to try that from now on.
Gracias,
blog: www.rosaurasandoval.com
galleries: www.rosaura.smugmug.com
I'm not sure anybody is still here but I'll go ahead anyway.
Following reading the LAB canyon book, LAB has become my colorspace of choice for my standard corrections.
My regular approach is to:
Go into LAB
Steepen appropriate portion of L if there is not already a great deal of contrast.
Steepen A and B in the range of 10-15% (I find more usually results in over-saturated colors)
Select the L channel and:
Highlight/Shadows to bring back area that are very dark and reduce the highs. I find that reducing the highs is the most important part as it remoze haze and glare and add crispness to the end result.
Add some sharpenning to the L channel.
= = = = = =
I now place a high degree of importance on the Highlight portion and am considering moving this to the first step.
- - - - - - -
Any comments and experiences would be much appreciated.
But how can I export an image to say sRGB for the Web?
When I try to do it all the color glory gets washed away.
What am I doing wrong?
― Edward Weston
Are you using Convert to Color Space or Assign Color Space????
(like LAB via menu). How can I convert to a colorspace like
you mention?
― Edward Weston
Next, unless your working space is sRGB, you need to go to Edit>Convert to Profile> sRGB.
Then you can save your image that is destined for the web.
Moderator of the Technique Forum and Finishing School on Dgrin
Now it works. Thank you very much!
― Edward Weston
help each other
Actualy I've never researched this question until now. But it seems that you can choose between two answers:
a) Be a scientist (http://www.beerinfood.com/Franklin.html)
b) Join the party (http://www.talkbacktees.com/NewFiles/06-003.html)
― Edward Weston
I'd really like to see a thread on LAB get revived.
I'll check in a few days to say if there are any signs of life.