Peter,
reading your message I think you are more informed about this subject then I am, but may be I can push things forward a bit.
What I understand from remove color is that the rgb colours are reduced to zero but the system is still set for colour. If under these circumstances you print an image you will finmd that the rgb cartridges are the first ones to go and not the black one!! Where as with greyscales you tell the soft ware to change colours into grey shades and the soft ware translates as it were the original rgb colours into greys, thus the result will be different.
I hope I made sense.
Robert
If you print using only black ink, you will probably get yet another look, in some cases I have found that gives me the look I am looking form. (Some of the old epsons let you chose just black instead of color)
Ralph
<
http://www.darkstar.us>
Follow up:
I’ve done a little experimenting and have some additional information. I’ve pretty much answered my own question for myself, but I’m still unclear as to WHY the software does what it does. So, that’s my remaining question: why is the software designed the way it is?
See the remainder of this message for what I found out about "the way it is"…
First, just to clarify: my question concerns strictly the on-screen, image editing behavior. While it’s true that there’s even more depth to the complexity when a printer is introduced, I don’t necessarily want to have to print out an image to get the image I want. 🙂
Anyway, my experiment involved drawing three solid rectangles, each one pure red, green, and blue respectively. I then used each method to remove the color.
Changing the color mode gave me the expected results: the rectangles all came out different shades of gray, with the green being the lightest, and the blue being the darkest. This is consistent with normal conversion practices.
Using the Remove Color command resulted in all of the rectangles being the same shade of gray. As near as I can tell, the red, green, and blue values are averaged, but without any weighting. Thus, each rectangle winds up as a 33% gray (since each started out with one color at 100% and the others at 0%).
I find it bizarre that a photo editing program with a pedigree firmly based in serious, professional graphic artistry would include such an amateurish color-to-black-and-white conversion feature (it’s the algorithm a freshman computer science major would use, before he learns about human visual perception).
However, since the mode conversion *does* do the right thing, I suppose it’s not that big of a deal. I can always export layers I want to be black & white to a new image, change the mode, and bring them back in to my original piece. Just seems a little silly to me, that’s all.
Anyone have any insight? In what situation would straight non-weighted averaging be a useful way to convert from color to black and white? I have to figure there are people out there using the feature…why do they consider it useful?
Pete
Peter,
I think you will get much better results with the following method, compliments of Scott Kelby. Weighting of colors can be addressed with this technique, it emulates the capability of full Photoshop’s "Channel Mixing".
* have your image in RGB
* create a Levels Adjustment layer above your image but DON’T make any changes, just click OK * create a Hue/Saturation Adjustment Layer above Levels and move the saturation slider all the way to the left, removing all color, click OK
* image now appears as a black/white photo, but it is not, it is RGB * in Layers Palette, click on Levels thumbnail to bring up the histogram * where the dialog box says RGB, use drop down to select one color channel * watching your image, move any/all of the three sliders * switch to the second color channel and tweak
* switch to the third color channel and tweak
Better? You have a lot more control with this technique, your eye is your guide as to how much contrast etc you want to build into the image. Technically this is still an RGB image, if you need grayscale, change the mode, it won’t change the image’s appearance.
Peter,
Last step,
* flatten image
Nancy, that’s pretty neat! Is that from Scott’s new book for Elements? I just ordered it this morning…
🙂
Chuck
Chuck,
Yes, Scott Kelby’s "Photoshop Elements for Digital Photographers"…it is a must-have in my opinion.
Nancy
Callum,
You don’t have Lab mode in Elements.
Nancy,
Thanks for the advice. That sounds like a very useful procedure, and I’m going to go check it out ASAP. I’m new to Photoshop Elements (and Photoshop, of course…but I don’t have that program), and eager to learn about all these things it can do.
That said, I was mainly trying to understand why the difference between the various methods to remove color. The color mode change does what I want, basically. I was just surprised the Remove Color didn’t. I think Chuck’s answer most directly addresses my original question (though it still doesn’t explain why that particular method is considered useful).
Thanks very much for everyone’s input. I don’t know that I’ve really found out what I wanted to find out, but I’ve learned lots of other interesting things in the process. 🙂
Peter,
The main usefulness of remove color to me is that by keeping the image in RGB it gives you the ability to maintain or add color back to an image. For example you have a photo of a red rose. You select the background around the rose and then use the remove color option. The result is a B/W image with a red rose. Can be very interesting effect. But for high quality grayscale conversion I would use the one of the methods using adjustment layers, like Russel Brown’s technique. You get the most control that way.
Rich
You select the background around the rose and then use the remove color option. The result is a B/W image with a red rose
Well, except that the B/W image left as the background doesn’t respect the perceptual differences between red, green, and blue.
To do what you describe, I think I would actually separate the flower and the background into two different layers, and then change the color mode for the background layer. This would have the same effect, but would keep the B/W image closer to the perceptual impact of the original.
More importantly, the Remove Color feature could have used the same algorithm that the grayscale color mode uses when converting. The usefulness you’re describing (which is very true) is in its ability to remove color from an area of the image, not in the algorithm it uses to convert color to grayscale.
"Nancy S" wrote in message
Peter,
Here is a variation on the former
<http://www.russellbrown.com/tips/photoshop.html>
Nancy,
Have you had success with the method Russell demonstrates?
I followed the instructions you credit to Scott Kelby, with wonderful results.
However, Russell’s method does not work for me in PE2. The adjustments made on the center Hue/Sat layer have no effect on the top Hue/Sat layer. In his vernacular, the "filter" does not affect the "film". If I turn off visibility of the "film", I can see the effects on the "filter". With both the film and filter visible, though, nothing I do on the filter layer causes any discernable difference.
I wonder if this is a "Big Photoshop" only process, or if I’m twisted up. (I hate resetting prefs…!#@$%)
Byron
Doh! I forgot to change the blending mode of the "filter" layer to Color… never mind.
"Byron Gale" wrote in message
"Nancy S" wrote in message Peter,
Here is a variation on the former
<http://www.russellbrown.com/tips/photoshop.html>
Nancy,
Have you had success with the method Russell demonstrates?
I followed the instructions you credit to Scott Kelby, with wonderful results.
However, Russell’s method does not work for me in PE2. The adjustments
made
on the center Hue/Sat layer have no effect on the top Hue/Sat layer. In
his
vernacular, the "filter" does not affect the "film". If I turn off visibility of the "film", I can see the effects on the "filter". With
both
the film and filter visible, though, nothing I do on the filter layer
causes
any discernable difference.
I wonder if this is a "Big Photoshop" only process, or if I’m twisted up. (I hate resetting prefs…!#@$%)
Byron
Peter, you can’t change the mode to grayscale on a single layer; it affects the whole image.
Chuck
Peter, you can’t change the mode to grayscale on a single layer; it
affects
the whole image.
BUT…having said that, here’s a potential workaround:
1. Take the image you wish to make partially grayscale and do an Image>Duplicate Image.
2) Turn the duplicate image into Grayscale (Image>Mode>Grayscale)
3) Turn the duplicate image back from Grayscale into RGB (Image>Mode>RGB
Color).
4) Move or Copy and Paste the grayscale-turned-back-to-RGB into the original image as a new layer.
5) Erase a portion of the pseudo-grayscale layer to expose the color below.
I think that will do it…
Chuck
"Chuck Snyder" wrote in message
Peter, you can’t change the mode to grayscale on a single layer; it
affects
the whole image.
Yes, I know. I took it as granted that would be understood, and that anyone reading my message would understand the layer in question needs to be copied to a new image, where it can be converted and then copied back.
Thanks,
Pete
Follow-up/Summary:
Well, I’ve had a chance to play with the techniques suggested, so I thought I’d write back one more time with my impressions.
First, I’ve learned a LOT about how to use the layers and what they can do. Some pretty flexible adjustments in there, that’s for sure.
I especially enjoyed playing with the Color blend mode using the Hue/Saturation adjustment layer type. One can accomplish similar results using the Levels adjustment layer type, but only by sacrificing color resolution. Using the Color blend mode and the Hue/Saturation adjustment, you get the best of both worlds. Neat!
Now, all that said: while one can approximate a true perceptual color-to-gray conversion (what changing the image mode to Grayscale does) using the techniques described here, they all ultimately fail in one important respect: loss of color resolution (or, since it’s a grayscale image, gray-level resolution).
For example, I can use the Level adjustment layer technique to adjust the relative red, green, and blue levels to match what the human eye perceives. But the only option there is to scale the levels downward. You can compensate by reducing the range of the input levels but that also loses color resolution.
So, only by using the mode conversion can you get a true weighted averaging of the color channels, taking full advantage of the full 8 bits of grayscale range available.
I really appreciate all of the techniques offered. Even though none match the color mode change for the particular application I initially asked about, they are incredibly useful for LOTS of other things. In fact, I suspect that the underlying ideas behind the techniques are the "bread and butter" of many Photoshop artists, seeing how much control one gets over the nature of the colors in the image with them.
Thanks so much for all the help, all of you…
Pete
Pete, one other technique to try would be a Gradient Map Adjustment Layer, possibly coupled with a Curves Adjustment Layer (an add-on you’re going to want if you don’t have it already). Just playing with them for a little while, I was able to get some way toward the Grayscale equivalent, although I don’t know how to use Gradient Map well enough to bring it to closure. Susan S. is our resident expert on Gradient Maps; perhaps she’ll offer her advice on its use. Just another one of those arcane features of Elements…
Chuck
"Chuck Snyder" wrote in message
Pete, one other technique to try would be a Gradient Map Adjustment Layer, possibly coupled with a Curves Adjustment Layer (an add-on you’re going to want if you don’t have it already).
Yeah, I saw the Gradient Map option. I have no idea how to use it though…haven’t had a chance to play with it yet.
Once I found out about the color-blending modes for layers, I had fun making my own color wheel though. It looks especially nice when you use a radial gradient fill for the circles, instead of just using solid colors. Lots of different effects that way.
Though, I did find myself wondering if there’s a way to change the blending mode for multiple layers at once. It got a little tiresome clicking back and forth between layers to repeat the blend mode selection for each.
Anyway, I guess I’ll have to look up the Curves adjustment layer too.
Thanks,
Pete
Pete, I also wanted to say ‘thank you’ to you for pointing out the huge difference between Grayscale Mode and Remove Color. I’m a little ashamed to admit I never noticed it, but in looking at what you had discovered and trying it for myself, I was amazed at the difference. I haven’t tried the comparison in Photoshop 7 yet, but I’m guessing it show the same anomaly.
Incidentally, in the add-on tools referenced in an earlier post, there’s one called Extract Luminosity, which presents a third (or more…losing count) choice for turning a color image to grayscale. It’s not as washed out as Remove Color, but is noticeably different from the Grayscale mode. Amazing program…
I don’t know about changing blending modes for more than one layer at a time….
Chuck
It’s not really an anomaly though. When you select Remove Color (which is the same as Desaturate in Photoshop), you still have three channels of data; R,G, and B. Just because these channels no longer have color, doesn’t mean they don’t contribute value. So with Remove Color you’re still seeing a composite image of data from three different channels. Elements doesn’t let you see channels, but they’re still there. If you convert to Grayscale mode, you only get one channel. The application sort of collapses the data from the three channels to make one Grayscale channel, similarly to if you flattened a layered image. That is why the results are different.
Elena, anomaly probably isn’t the right term, but the results are very different between the two methods. I didn’t realize just how different until I tried Pete’s experiment with three shapes: one pure red, one pure green, and one pure blue. The equivalent K (blackness) values of those three colors are quite dissimilar: 16, 50, and 64, respectively. Those values are maintained if you do an Image>Mode>Grayscale; however, when you do a Remove Color or a Hue/Saturation with sat all the way to the left, the values are 50, 50, and 50. Pretty dramatic shift – more than I would expect from flattening the image. I’m going to do a little more research; one of these Photoshop books on the shelves behind me must have an explanation…!
Chuck
Postscript on previous message: Dan Margulis has a chapter on the subject in his book, "Professional Photoshop: The Classic Guide to Color Correction." I’m plodding through it now…trying to get the concepts through my thick skull.
🙂
Chuck
Elena, thanks for a beautifully succinct explanation.
The blend that goes into grayscale is approximately 60 percent green channel, 30 percent red channel, and 10 percent blue channel – apparently based on the relative contributions of those channels to the average RGB image. So says Dan Margulis in the book I mentioned before. Seems logical…. In full PS and ‘juiced’ PE, you can do your own channel blending to get just the right mixture for the particular image with which you’re dealing.
I have now exhausted 125-150% of my knowledge on the subject….sorry I wasn’t succinct, Barbara!
🙂
Chuck
Um, that sounds like what I said… :-p
Okay, okay….Elena, you and Barbara win!
🙂
I thought it was fairies and pixie dust that made it work until I read thru all this. After reading you guy’s explanations, I’ve decided to stick with that ’cause it’s alot simpler to understand. 🙂
Seriously, even though it was way over my head, I enjoyed lurking while you guys hashed it out and even managed to learn something from it.
CR
Susan, I’m enjoying this thread immensely!
I tried an experiment with a rather flat image of a flower and a duplicate, turning one copy to grayscale and using the gradient map on the other. As you pointed out, the gradient map version had superior overall contrast, with the blacks and whites pushed almost to the limits. I say ‘almost’ because I then applied Auto Levels (or Auto Contrast – same on grayscale) to the grayscale copy and got slightly more contrast – pure whites and pure blacks – as measured with the eyedropper tool.
The gradient map layer is definitely moving the K values, though – different results for K (or B of HSB)for equivalent spots in the layered image with Gradient Map layer on vs. off. Very interesting ‘feature’!
Chuck
There is a tutorial I’ve seen by one of the Photoshop Evangelists, Julianne Kost, where she goes over several diffent methods for changing to B/W. I haven’t watched it in a while so can’t exactly recall all of the reasoning, but I remember she actually went over the pluses and minuses of each method and I think concluded that some images will benefit more from one than the other. Images with highly saturated and varied colors would generally benefit from one of the methods that give the most control, like using the channel mixer or Russell Brown’s method of 2 Hue/Sat adjustment layers.
Rich
Rich, thanks. We’ve gotten a little deep into the math and science of this one; I’m sure some of the artists in the group are wondering what the heck we’re doing spending so much energy on trying to figure out what’s underlying the techniques rather than devoting our attention to what creates the most pleasing image…! But at least we know now that there are a bunch of different ways to create a grayscale image that are significantly different from one another. Ya gotta love this program…!!
🙂
Chuck
Chuck,
There’s nothing wrong with having this kind of indepth discussions. They are as much part of a forum as the more straightforward discussions/questions.
The forum users that do not like this subject probably have left the thread at some stage.
Not that I understood all but that doesn’t matter to me, the discussion itself was interesting enough.
Some day it will come in handy.
Robert
Chuck, I agree with Robert. These discussions are a great help to understanding the way Photoshop works.
I saved important parts of this discussion (to Word the way Beth describes in another posting) for later referral.
I would like to express my sincere thanks to everyone who contributed to this extremely important and interesting discussion. This was the best I’ve seen on this forum.
Maybe an appendix to Jody’s book?
Leen
