Post Processing #1 - Very Basic Contrast Tutorial

[Scottes]

So this is the picture I shot in one of my attempts to take a landscape.

http://www.itsanadventure.com/postimages/5337_original.jpg

Yuch. Looks pretty washed out, and it isn't very straight.


The very first thing I'm going to do is switch to 16-bit mode by clicking on Image... Mode... 16-bits/Channel. Because I'm go to be mucking with colors and stretching them, squishing them, and otherwise distorting them I want as many variations of colors as possible.

Then I'm going to set this to Adobe RGB color space if it's not already there. Again, pretty much the same reason as above - more colors. Adobe RGB is a larger color space then sRGB - that is, it actually has more colors available to use. (It's a very long story so trust me for now.) Choose Image... Mode... Convert To Profile and choose Adobe RGB (1998) as the Destination Space. If the Source Space is already Adobe RGB then click Cancel, otherwise click on OK.


Next I'm going to tackle the washed out look by adjusting the Levels to get some contrast in there. Image... Adjustments... Levels and here is the histogram.

http://www.itsanadventure.com/postimages/5337_prelevels.gif

One of the ideas behind using Levels is to utilize the entire range of colors from pure black to pure white. This histogram shows that my image is only using about 2/3 of the range, all in the middle. I'm going to stretch out the colors to use as much range as possible - well, as much as necessary to make the image look good.


The first thing I'm going to do is grab that bottom left slider - the black one - and drag it over to where the blacks seem to start, right at the beginning of the hump. By doing this I'm telling Photoshop to make my blacks start at the point above the slider - or in other words, make this point pure black. Anything blacker (to the left) will be made black, too, but there's not much black in that range so it won't make much difference.

Then I'll grab the right slider - the white one - and drag it over to the left, again to the beginning of the hump. Same idea - make this pure white as well as anything brighter.

Here's the histogram:
http://www.itsanadventure.com/postimages/5337_midlevels.gif

And the resulting image:
http://www.itsanadventure.com/postimages/5337_midlevels.jpg

But I didn't like that, because it really increased the glare at the top of the gazebo, and I lost detail. If you look at the histogram above you'll see that there *are* some whites all the way to the right edge - very few, indicated by the very thin black line, but they are there.

Here's the detail I lost (original on the left):
http://www.itsanadventure.com/postimages/5337_details.jpg

To get the details back I'm going to move the right slider back all the way to the right.

Come to think of it, I lost some blacks in the same way when I compressed them by moving the left slider. But with blacks it doesn't seem so noticeable. Losing the whites by compressing them did noticable damage, so I want those details back. The blacks I destroyed just became more shadows, and the eye doesn't seem to notice that. (Well, in *this* picture it didn't. Sometimes it's the other way around.)


Now the image doesn't look quite right. It's a little dark. I'm going to move the middle slider - the midtones - around until it looks good to my eye. Moving it to the left will make the image brighter, and to the right makes it darker.

This step is totally subjective - play with the middle slider until it looks good.

This is what I eneded up with:
http://www.itsanadventure.com/postimages/5337_levels.jpg

And this is the histogram:
http://www.itsanadventure.com/postimages/5337_postlevels.gif


And that's it. The picture doesn't looked washed out any more, and we're using almost the entire range of color. What's important is that it looks good to me.

OK, the *colors* look good to me. :)

[Mills]

Thanks. Like I have said, I always seem to learn when I stop by this forum.

[tommykjensen]

Thanks, a lot of reading I have to do when I get back from work.

[FJC]

Excellent tutorial - it is so much easier to grasp with the various pictures and screenshots.

I just managed to learn this in the past couple of weeks by reading a ton of other sources. I should have just waited and read this, and saved myself a few days. :) Now I find that the Levels command is easier to use than the other brightness/contrast controls.

Quick hint - CTRL-L brings up the Level's applet quickly. :)

[Jesper]

Another hint: Hold down the ALT key (don't know the equivalent on a Mac...) when you drag the left or right sliders in the Levels dialog. The image will turn black (or white) and while you drag, you'll get to see which pixels will be clipped to black (or white).

[Scottes]

Another hint: Hold down the ALT key (don't know the equivalent on a Mac...) when you drag the left or right sliders in the Levels dialog. The image will turn black (or white) and while you drag, you'll get to see which pixels will be clipped to black (or white).

Hey, THAT is sweet!

I did this while adjusting the blacks and saw that only scattered pixels would be clipped. This explains why clipping the blacks isn't so noticable in this image.

But when adjusting the whites I saw that a couple of areas would be clipped - notably the gazebo roof and the stone path to the left. Given that so many clipped pixels were clumped together it's easy to see that I would be losing detail in those areas.

Nice tip. Thanks Jesper!

[FJC]

OH, now that tip rocks! Thanks much!

[IanD]

Thanks for the time and effort to post this Scott. I'm sure a lot of folks are going to be reading the entire series, starting with me first.

[RCJ]

[quote="Jesper"]Another hint: Hold down the ALT key (don't know the equivalent on a Mac...)

That would be the alt/option key on a mac.

[Jmurman]

Hey Thanks! Great job!

[BDM]

That was a very interesting post. I'm relatively new to Photoshop and I have a lot to learn. So far I have learned a few things not to do. At some point I must begin to learn what to do.

I believe you performed a great service to those of us who are just getting our feet wet. Perhaps you could write additional ilustrations from time to time.

Bruce

[Scottes]

Thanks Bruce. And I'm glad you enjoyed it. Note that we have several tutorials in the top "sticky" post in this forum.

As to writing more... I wish that I had time. With summer here, busy weekends are constant, and I barely have time to edit pics and spend a few hours a week on my website (due out in June of 2008 at this rate).

I've been thinking about writing smaller, "quick tip" tutorials, though... Maybe...

[RinkRat]

Thanks Scottes!

Very cool!

Forget the website. When is the book of tips coming out?? :)

[tofuboy]

I'm just starting to learn PS, had it for years but after discovering how great it is to shoot in RAW decided it's time I learn some useful techniques. I spent some time flipping through some PS for photography books at the bookstore and found a few useful hints, including this one. You can accomplish similar effects using curves... perhaps we can get a tutorial on them soon ^^;;

Anyways, one thing to add to this one. You can modify the image by using 'Adjustment Layers'. This is a non-destructive method, as it adds the effect to a layer rather than the actual image. The effect is the same, and the tools are the same. The real plus to using these is you can later change the effect or delete it without losing your other changes. To use Adjustment Layers, go: Layers -> New Adjustment Layer -> (choose one of the options).

[Jesper]

That was a very interesting post. I'm relatively new to Photoshop and I have a lot to learn. So far I have learned a few things not to do. At some point I must begin to learn what to do.

I believe you performed a great service to those of us who are just getting our feet wet. Perhaps you could write additional ilustrations from time to time.

Bruce

If you want to learn how to correct and improve photos in Photoshop, go to a bookstore and look up one of the many books about the subject. I have this one and I like it: The Adobe Photoshop CS Book for Digital Photographers (http://www.amazon.com/exec/obidos/ASIN/0735714118/qid=1093075015/sr=ka-1/ref=pd_ka_1/104-9361414-7211916) by Scott Kelby. It contains a lot of tips like what Scottes showed here.

[Jmurman]

Jesper. I just got in 2 of Kelbys book os CS. I like them both!

[gkma01]

A really good tutorial !
A small point:

The levels feature can be applied to any PART of the image which is selected using any of the tools -- crop, magic wand etc.-- thus avoiding changes to those bits of the image which are OK

Myles

[jrm]

Another hint: Hold down the ALT key (don't know the equivalent on a Mac...) when you drag the left or right sliders in the Levels dialog. The image will turn black (or white) and while you drag, you'll get to see which pixels will be clipped to black (or white).

The Mac equivalent is the "Option" key, and it will do the same thing as the Alt key.

This tip also works when processing RAW images in PSCS (Not sure about earlier versions). Hold the Alt/Option key when dragging the "Exposure" or "Shadow" sliders and you will have a preview of clipped pictures.

The Alt/Option key is used quite a bit in PS. Let's say you play with the Levels and really screw up the image (before you hit "OK"). No need to click cancel and then open "Levels" again... just hold the Alt/Option key - the canel button will turn into a "Reset" button, bringing you back to where you started.

This same "reset" technique works in most dialog boxes. Hold down Alt/Option most anywhere to see what features it changes.

[Jmurman]

A really good tutorial !
A small point:

The levels feature can be applied to any PART of the image which is selected using any of the tools -- crop, magic wand etc.-- thus avoiding changes to those bits of the image which are OK

Myles

how do you do this?

[gkma01]

A really good tutorial !
A small point:

The levels feature can be applied to any PART of the image which is selected using any of the tools -- crop, magic wand etc.-- thus avoiding changes to those bits of the image which are OK

Myles

how do you do this?

In Photoshop open image file. Select eg Rectangular Marquee tool. Apply it to area you wish to submit to levels adjustment. Use levels (as shown in original tutorial). Job done !

[Jmurman]

A really good tutorial !
A small point:

The levels feature can be applied to any PART of the image which is selected using any of the tools -- crop, magic wand etc.-- thus avoiding changes to those bits of the image which are OK

Myles

how do you do this?

In Photoshop open image file. Select eg Rectangular Marquee tool. Apply it to area you wish to submit to levels adjustment. Use levels (as shown in original tutorial). Job done !

cool....thanks!

[Scottes]

In Photoshop open image file. Select eg Rectangular Marquee tool. Apply it to area you wish to submit to levels adjustment. Use levels (as shown in original tutorial). Job done !

A seclection - especially done with Select... Color Range - is a cheap Shadows or Highlights recovery for those without PS CS.

Using Select... Color Range choose a very dark pixel and then run the Fuzziness slider up to select a good wide range. Hide the selections with Ctrl-H. Then use Levels to add contrast or just adjust Gamma with the middle slider. Instant Shadow Recovery and looks very good.

PS: If it doesn't look so good on the edges of the selection, Undo and choose Select... Feather and use a small number.

[Mathiau]

So easy! and so great! just did on pic and it looks so much better!

[Riken]

Great tutorial. Definitely need more of these.

I prefer doing adjustments in a separate layer. That way you can turn them on or off, and mask out bits too. To do this, instead of going Image->Adjustments->Levels try Layer->New Adjustment Layer->Level.

[Mathiau]

^^ thanks for that tip - i am always hitting ctr+alt+z :)

[Sharpy]

Hi Guys,
1. How do you produce those screen shots?
2. How do you produce the grey highlight area showing a previous quote?
(still limited knowledge of the computer side of things, learning all the time though by
watching these excellent posts.)
Regards Alan

[Scottes]

1) Alt-PrintScreen, then into Photoshop and File... New... Edit... Paste
2) In the botton right-hand corner of everyone's post is a "Quote" button. It quote their text in grey and puts you into make a post.

[BillMarks]

Very cool! But it is my understanding that you can't really gain anything when you start with 8 bit color and switch to 16 bit. If you only captured 8 bits of color info that's all you'll have. When you switch to 16 bit those additional 8 bits are just empty. If your camera captures 12 bits in RAW, you only get 4,096 color levels in 16 bit (2 raised to the 12th power), not 65,536 (2 to the 16th).

Anyone know if my info is correct?

[jfrancho]

By switching to 16 bit, you minimize impact when editing color since the extra detail is available when making adjustments.

[Sharpy]

1) Alt-PrintScreen, then into Photoshop and File... New... Edit... Paste
2) In the botton right-hand corner of everyone's post is a "Quote" button. It quote their text in grey and puts you into make a post.

Thanks Scottes I think I've got it.
Regards Alan

[Scottes]

Re-reading my reply makes me think I should stop replying at 5:30 in the morning... I have no idea what language that was.
:-)

[BillMarks]

By switching to 16 bit, you minimize impact when editing color since the extra detail is available when making adjustments.

But if you capture in 8 bit, there IS no extra detail, right? Shooting in 8 bit records 256 levels of luminocity. After switching to 16 bit mode, you still only have 256 levels of information (I think).

If you shoot in Canon Raw (at least with a 10D) you capture 4,096 levels of luminocity. If you process in 16 bit, you actually can use all that information. If you switch to 8 bit, the file looses 3,840 pieces of information per pixel.

[barrettbonden]

If you want to learn how to correct and improve photos in Photoshop, go to a bookstore and look up one of the many books about the subject. I have this one and I like it: The Adobe Photoshop CS Book for Digital Photographers (http://www.amazon.com/exec/obidos/ASIN/0735714118/qid=1093075015/sr=ka-1/ref=pd_ka_1/104-9361414-7211916) by Scott Kelby. It contains a lot of tips like what Scottes showed here.

great book. I use deke mclelland's bible for basics and this one for more specific info. There are also great dvd's from both mclelland and kelby. in my opinion, videos are the best way to learn, it's just as is you had a friend showing you how to do things in your computer

[jfrancho]

But if you capture in 8 bit, there IS no extra detail, right? Shooting in 8 bit records 256 levels of luminocity. After switching to 16 bit mode, you still only have 256 levels of information (I think).

If you shoot in Canon Raw (at least with a 10D) you capture 4,096 levels of luminocity. If you process in 16 bit, you actually can use all that information. If you switch to 8 bit, the file looses 3,840 pieces of information per pixel.Editing in 16 bit/channel mode allows greater sensitivity irrespective of source bit depth. If it wasn't an issue, 16 bit mode wouldn't be available since printing is in 8 bit.

[BillMarks]

Editing in 16 bit/channel mode allows greater sensitivity irrespective of source bit depth. If it wasn't an issue, 16 bit mode wouldn't be available since printing is in 8 bit.

Well, not that I don't believe you, but i'm having trouble understanding the mechanics behind "switching up."

With 8 bit capture, each pixel records a number from 0 to 255. That's 256 discrete values. 16 bit editing allows for 4,096 discrete values per pixel. But there is no way to map 8 bit capture information into 4,096 usable values. You could assign every value in the 8 bit source to every 16th value in the 16 bit range, but you are still only left with 256 discrete values (the 15 levels between each 16th value end up being empty). And you can't split a value from the 256 available in 8 bit into 16 pieces, assigning each to a different value 16 bit has available.

So I don't see how 16 bit editing treats an 8 bit source any differently than 8 bit editing would.

My understanding of the basics of editing computer code is that it is essentially matrix algebra: an image file is a big matrix with X rows and Y columns (determined by the sensor chip pixel dimensions) and a depth of Z (determined by the bit-depth of each pixel).

Old versions of, say, PS, had computer code to handle various sensor sizes but a bit depth of only 8 bits (i'm guessing this was because that's what early chips could capture). PS CS has some 16 bit processing available--which meant that someone re-wrote the code of some editing options to handle matricies with a depth of 4,096 levels. PS CS2 has even more 16 bit editing options--meaning that more options were converted to handle matricies of 4,096 levels deep.

My understanding is that 16 bit editing offers more sensitivity because it can manipulate/registed/recognize more color levels (4,096 vs. 256). But if you have only 256 to start with, it can't offer that sensitivity.

To me it's like if you pour 8 pints of water into a 16-pint bucket, the 8 pints of water don't expand to fill the 16 pint volume of space in the bucket--they just leave empty space in the bucket.

Can you help me out here on how this works?

[jfrancho]

Your paint bucket analogy won't work here, since color data at this point is non linear. I believe it has been explained clearly enough in previous posts, including the first post. Try this: take a piece of paper and cut into 3 strips. On one, mark off 12 ticks, evenly spaced and number them 1-12. On another, mark off 8 ticks, evenly spaced and numbered 1-8. On the third, mark of 23 tickes, randomly spaced, and unnumbered. Line up all three with the unumbered strip in the middle and compare the ticks. Assign a value from the closest tick on 12-bit strip to the random strip at each tick. Do the same with the 8-bit strip. Now slide the center strip (image data) to the left a few inches. Reassign values to the random ticks on the center strip. Do you see the difference in precision now? Now imagine that when you edit the image, what was once 8 pints of water, is now 1.337 Imperial Gallons vegetable oil.

[BillMarks]

Sorry. Your explanation just doesn't add up (to me). It's like you are making gold out of lead.

[jfrancho]

Then by all means, bind early to small gamut workspaces and do all your editing in 8-bit mode. At least the chrome filter will be available to you. What you fail to see is that you are adding and removing and otherwise changing the source data when editing in a program like Photoshop. Common activities such as sharpening, noise reduction, color correction all change the info. Would you rather do the fine cuts with a 16-tooth blade or an 80-tooth blade? It sounds as if you are going to resist any reasonable explanation, so I'll leave it to the others teach you.

[Scottes]

So I don't see how 16 bit editing treats an 8 bit source any differently than 8 bit editing would.
Sometimes it's been seen through example....

Pull up a web-sized picture that contains a nice pretty blue sky. Duplicate it. Make one 8-bit and the other 16-bit.

On one, play with Shadows & Highlights. Do the *exact* same thing to the other. Then play with Levels, pulling both sides in quite a bit (I'm exaggerating here to make sure the difference is visible). Do the *exact* same thing on the other picture. Pull a hard curve, an exaggerated one. Do the *exact* same thing on the other. Add some Saturation, something noticable like +20. On both, of course. Find a couple other things that change the colors like these.

Now compare the Histograms of both. Your 8-bit version should be pretty darn spikey, showing posterization, which lessens the number of unique colors in an image. The 16-bit one should not be spikey.

You NO LONGER have a full 8-bits of data in your 8-bit file - the processing posterized the image and you lost data. You lost data. Bad. Your 16-bit image didn't lose data.

Now zoom in a little bit (200% or 300%) and check out the sky. The 16-bit image should have no noticable transitions from one shade of blue to the other. The blue should change smoothly across the scene. The 8-bit image should have noticable bands where one shade of blue changes to another shade.

When processing in 16-bit PS is able to smoothly transition from one shade of blue to another because there are 256 times more shades of blue between two colors. 8-bit mode can't do this, and will suddenly change from one shade to another.


If you pick a pic with a blue sky and do some exaggerated processing steps all this should be very noticable.


Note that it's not so noticable with other picture, or even parts of the picture. If the picture you used had some trees or grass in it, zoom in to that section and compare. You most likely won't see a difference, or if you do it will be much less noticable than the sky.

[BillMarks]

Then by all means, bind early to small gamut workspaces and do all your editing in 8-bit mode. At least the chrome filter will be available to you. What you fail to see is that you are adding and removing and otherwise changing the source data when editing in a program like Photoshop. Common activities such as sharpening, noise reduction, color correction all change the info. Would you rather do the fine cuts with a 16-tooth blade or an 80-tooth blade? It sounds as if you are going to resist any reasonable explanation, so I'll leave it to the others teach you.

Geez. I'm not resisting your explanation. I JUST DON'T UNDERSTAND IT. I get that 16 bit editing on a 16 (or even 12) bit source is advantageous vs. 8 bit editing. I'm just having trouble understanding the mechanics of switching up.

[BillMarks]

Scottes:

Thanks for your example. I did as you outlined and I saw the effect you predicted.

The histograms of the original 8 bit image and the duplicate 16 bit conversion were exactly the same after the conversion.

Then I applied +25 contrast (just to keep it simple, with no compound effects). At that point, the histograms were different. The 16 bit histo was shifted less to the right compared to the 8 bit. 8 bit had a higher mean and median compared to the 16 bit image. This indicates clipping or posterizing or both going on in the 8 bit image that is not hapening in the 16 bit image--due to the extra "room" available in the 16 bit processing space. I'm not sure I buy that NO damage is occuring in the 16 bit processing, but it is certainly less damage than the 8 bit processing.

[Mathiau]

going from 8-bit to 16 is like going from black and white to color..... you have a larger palette of colors to choose from vs 8bit color.

On your computer monitor - put up a wallpaper with transitions of colors - put your monitor to 16bit color - now put it to true color (32bit)

i can see the difference clearly.

Think someone may need to post a side by side pic of an image done in 8bit and 16bit.

So my questions is...

You change your 8bit image to 16bit to do some image editing - THEN you convert it back to 8bit to save it as a jpeg - how does that work since 8bit obviously can not provide the color's 16bit does but all your editing is done in 16bit.... quality loss?

[Scottes]

Yes, there's some quality loss, of course, but not as much as there would if the entire processing were done in 8-bit. Again, that depends on the image.

And with the eye of the beholder, now that I think about it. I've shown lots of people things in PS and they honestly can't see the difference, whereas the difference is black and white to me...

[Mathiau]

I am like tha myself on many things - i often see, and hear things other's dont - sometimes i hate it, like when someone has a CRT monitor @ 1600 x 1200 and 60mhz refresh rate, to walking into a house and being able to tell you if a T.V is on (with no sound on) - they just make a faint noise i can hear.

[Jaymz]

I can hear the "whine" most TVs make when they are on, also. It drives me nuts when a TV is left on. heh

[Mathiau]

^^ thas what i am talking about :)

[Rezarf]

great tutorial, never noticed the small line at the bottom of the levels before always dragged it to the bottom of the curve.

Always learning.

Thanks
Rezarf

[jameslcross]

Good stuff

[adas]

Well, not that I don't believe you, but i'm having trouble understanding the mechanics behind "switching up."
[...]
So I don't see how 16 bit editing treats an 8 bit source any differently than 8 bit editing would.

It's all about mathematical precision. That's why they use 16bit instead of 8 bit. Hey, I would prefer 128 bit SSE2 floating point precision, because my processor has this capability. Everyone's computer has it. But then, one image would require 16 times as much RAM to fit in.
There's no color palette broadening when you go from 8 bit to 16 bit. Some comments here are quite missleading.

Scottes, nice work you put here, although I'm not fully agreed with it.
Sometimes later I'll post my JPG recovery workflow.

[Jaybob]

Great tutorial Scottes. I'm learning a lot from this site. I know this post is a little late but is that picture from Atikokan Ontario? It looks extremely familiar.

[Scottes]

Tower Botanical Gardens, near Worcester, MA

[creature_]

Thanks for the tutorial! :D