royals

I'm new here so please to meet you.

I have some questions about adjusting a (digital) photo in photoshop. I have read some books about this but everybody says something different. What things should i use to improve my photo? Should i use Curves, Levels, Color Balance, Selective Color or only one or two of these? Wich of these should i use first and wich are more important and wich not?

Also I would like to know if CMYK or RGB color is better for printing. And if there's some general diagram about how much magenta, yellow and cyan one photo should contain? Or is this just personal taste.

I already know Unsharp Mask is the last thing u should add to an image in most cases. But i'm really curious about color adjustments because i want to make the best image possible.

I would really appreciate your help. Thanks in advance.

Lee

Welcome, Royals

All the things you mentioned are important but rarely does one image need them all. If you could post the image your are talking about we could better help you learn to improve what need some adjustment.

royals

Hi Lee,


I posted an image as an example. This one is good to work on i figure.

ronmatt

Personal taste, of course, is what you are trying to appease. No matter what adjustments you make, if you don't like the image, then you don't like it. As to whether you want RGB or CMYK depends on your printer. What does the machine require. If you refer to 'printing' on a press, then,CMYK. There is no such animal as a RGB print press. Everything you questioned begins and ends with the quality of the photo initially. As an example, the photo you've posted leaves much to be desired. It's flat and it's grainy. I don't have much use for any of Photoshops 'auto adjustment' features so I'd use adjustment layers and curves to enhance this one. One very important piece if advise though would be, if your going to be making color adjustments, make sure your monitor is calibrated properly and I'd recommend 'not' using any color profiles. or ICC tags. A lot of people also prefer adjusting color in LAB colorspace. I've downloaded the image and I'll mess with it later and post the results with a note as to how I did it. As to unsharp masking, if you change the size and or resolution, it's best to use 'some' throughout the process.

royals

Thanks Ronmatt, hope to see the end result [righton]

MindBender

I highly recommend getting or at least reading through Katrin Eismann's books on photo retouching if you are serious about doing retouching. One of the best resources for techniques you'll find. Not always beginner stuff, but easy enough to follow after you've gone through it a couple times.

For printing, ronmatt is right, there aren't such things as RGB printing presses per se... but there are times when you would want to leave your files in RGB. Unless you're doing logo work or something similar, work with photos in RGB and then save a copy in CMYK to go to the press. The editing environment is better in RGB and you will have access to more tools as well as a larger color gamut. Working in L*a*b is okay too, if you know how.

Now, if you're planning on printing off a desktop printer like say, and inkjet (especially inkjets) do not convert to CMYK, always work and stay in RGB. Desktop printers assume that you are working with RGB files and if you send them CMYK, they will convert them as if they were RGB and the colors will look like they are made of mud. Just stay in RGB.

This is one of the most important things you can do. If you are planning on doing a lot of work where color accuracy is important, then you should consider purchasing a hardware color calibration setup. At the very least, you should run through Colorsync or Adobe Gamma to generate a calibrated profile for your monitor. This is more important with CRT, which have more accurate color but can shift more. While LCD can shift color, there isn't as much that you can do with it... and Adobe Gamma wasn't really designed to work with LCDs so the more you shift away from it's default the less luminosity you will have. It's a little tricky to do LCD with visual calibration like Adobe Gamma also due to their inherent visual color shift, even on new LCDs.

This is just plain wrong. Unless you are specifically told not to work with color profiles by a printer (usually because they haven't updated their equipment in the last 20 years) then you should always work with a color profile. That color profile should be a standard color profile too (do not use your monitor color profile). Use something like sRGB or AdobeRGB1998. Working with a color profile is what allows all that careful calibration to continue on to the out put device. That profile is what tells the hardware what colors are supposed to look like. Without that, there's hardly any point in calibrating or adjusting because you're just relying on luck at that point. Any modern print service bureau should be dealing with color profiles, and honestly, if they aren't... find a new one. For home printing, it's slightly less important since your desktop printer is directly hooked to your computer, but it's still helpful. And should you ever want to take the file somewhere else, you won't have to worry about correcting the color again (hopefully).

Photoshop automatically does a special unsharp mask every time you scale an image. So whenever you change the size it's automatically applying it to attempt to preserve the per pixel contrast. If the image needed sharpening of highlights before, then it will after a resize too. If it was okay before, then it will still be okay most likely. (Although, most images can use a little sharpening of some kind or another). Changing resolution has no effect on what the pixels look like. Resolution is purely a print conversion tool. You can have a 72dpi image and convert it to a 1200dpi image... and not one pixel will change. When you check "resample" you are actually scaling the image to fit with a resolution choice.... that is scaling... not changing resolution. Understanding resolution is a tough one for beginners and seasoned graphic artists alike. Sooner or later it will throw you for a loop... especially when dealing with multiple output devices.

ronmatt

mindbender... you caught me. I was wrong about the icc tags. However, Pitstop will only remove them and replace them with the printers own profiles. The thought here was if you set to Adobe sRGB, don't mess with it.

It's also been a rule of thumb to reduce in size, in even numbered steps and to 'sharpen' some at each step. You tell us we don't have to anymore? good news...

MindBender

Hence the qualifier... if they can't work with industry standard... don't use them. Or at least, if you have a very compelling reason to use them, expect the worst and prepare yourself for lots of trial and error before you get what you want (if ever heh).

Photoshop, for a few versions, has don't a light unsharp type procedure. It's not as extensive as what can be accomplished by using the unsharp mask itself, but it's the same basic procedure on the pixel level from what I understand. That is the whole concept behind "stair-stepping" an image (prior to the bicubic resampler et al). You would step down, it would sharpen, you would step down, it would sharpen. Staying within the limits of what the imperfect interpolation engine (read that as, sacrificing quality for speed) that they use allows Photoshop to more accurately maintain the detail elements of the image while you scale. From what I've heard this isn't necessary with CS2 if you're using the new resampler, but prior to that it was the method of choice.

royals

Thanks for your advice Mindbender, I appreciate it.


"And if there's some general diagram about how much magenta, yellow and cyan one photo should contain?"


I was curious about this because with photo-printing, for instance black and white printing, there's always a rule about how much grey, black and white there should be in the pic. So that's why i asked what numbers are best to use with color adjustment (like 50% Magenta, 30% Cyan & 60% Yellow).

docilebob

How much of any one color is in an image depends entirely on the subject matter of the image. If you use a formula for color amounts, all images would look like they were taken in the same setting, at the same time of day, by the same camera, and so on.
Viva la differance !