You have to login first in order to be able to post messages.
Click here to login or register if you are a new user.


Page:
1
Showing posts 1 - 10 of 10

Short tutorial: Press or non-press printing

Author Message
Costa007
414 posts
84
Message posted at 09/06/2006, 02:38:27 AM by Costa007 - member is an admin
The most usual image modes, reguarding colors, are RGB, CMYK and gray. RGB mode is the most common colour mode when printing on glossy paper (witch includes the fotographic paper) and the CMYK mode is most common when printing on paper - newspaper. The most important difference between these colour modes is that the RGB mode uses more sparkling colors and CMYK more pale. The big problem which the newspapers confrunts is that the amount of ink in the image is crucial. If the "black" color in the CMYK (cyan, magenta, yellow, black) is 400% (100% for each colour layer), the ink printed on one side of the paper, would probbably resorb also through the other side, and that is a mess. The RGB colour is more easy to print, and most photographers, print their work on glossy. If you have to print one of your images on a newspaper, the tip is: after you prepared your photo (colour layers, contrast, brightness) in RGB, flatten all layers and resample the image to CMYK, or better, edit your photo directly in CMYK mode (always checking the amount of ink used for each "final" color), using the settings in "colour settings" to apply with the printing machine from the typography ("dot gain", "resolution", etc). The black color is better to be 100% black and eventually 20% another color. Many “experienced” DTP-ists make their advertising posters for newspapers with 300% black and over. The usual-good newspaper is made for 240% - 270% ink for each (printing) side, over this limit - problems begins to appear.







Tip for all the fotographers who use digital editing:



When you shoot jpeg and edit your photos in photoshop (like most do), even is a question of brightness/contrast, after you open the image, convert it from 8 bits/channel to 16 bits/channel, to increase the amount of information in the image. After that, you can do the editing and, when you're finished, convert the image again to 8 bits/chanel. This is a simple thing to do, and preserve the image (colour dept and pixel/details information) quality as much as possible.
Sharp eyes, opened mind, two cameras, paper, pen, books, mus...

Uploaded files:3868 | Total Sales: 16403
78
Message posted at 09/07/2006, 14:19:05 PM by Littlemacproductions
I will have to try that. Interesting. Thanks for the tip.



"When you shoot jpeg and edit your photos in photoshop (like most do), even is a question of brightness/contrast, after you open the image, convert it from 8 bits/channel to 16 bits/channel, to increase the amount of information in the image. After that, you can do the editing and, when you're finished, convert the image again to 8 bits/chanel. This is a simple thing to do, and preserve the image (colour dept and pixel/details information) quality as much as possible."
Two brown eyes and an open mind.

Uploaded files:1569 | Total Sales: 27956
Red
1637 posts
Message edited at 09/07/2006, 15:32:56 PM by Red
Converting an 8-bit image to 16-bit doesn't automatically increase tonal range. You are still only using the 256 colors available in an 8-bit image and spreading them out - you end up with the same 256 colors with gaps between the colors and this introduces noise. And, if your image doesn't have all 256 colors in it to start with you are making it even worse by spreading less colors out, leaving more gaps. Unless you do something to the image to add color information in the empty spaces any manipulations you do in 16-bit will just spread and compress the original 256 levels. Take it back to 8-bit and take a look at your histogram. You will find small gaps throughout the image. You can try adding a bit of Photoshop>Filter>Noise to the image, or a bit of Gaussian Blur to the worst channel to fill in the gaps but that only introduces new problems. In essence, you cannot create extra colors simply by going from 8-bit to 16-bit.


Uploaded files:212 | Total Sales: 1564
Wysiwygfoto
671 posts
76
Message posted at 09/07/2006, 18:39:54 PM by Wysiwygfoto
Hmmmm...makes for a good argument for uploading RAW files as an additional format and leaving the options available to the buyer/printer.


Uploaded files:6 | Total Sales: 1095
78
Message posted at 09/08/2006, 05:04:48 AM by Littlemacproductions

Originally posted by Red:
Quoted Message: Converting an 8-bit image to 16-bit doesn't automatically increase tonal range. You are still only using the 256 colors available in an 8-bit image and spreading them out - you end up with the same 256 colors with gaps between the colors and this introduces noise. And, if your image doesn't have all 256 colors in it to start with you are making it even worse by spreading less colors out, leaving more gaps. Unless you do something to the image to add color information in the empty spaces any manipulations you do in 16-bit will just spread and compress the original 256 levels. Take it back to 8-bit and take a look at your histogram. You will find small gaps throughout the image. You can try adding a bit of Photoshop>Filter>Noise to the image, or a bit of Gaussian Blur to the worst channel to fill in the gaps but that only introduces new problems. In essence, you cannot create extra colors simply by going from 8-bit to 16-bit.




Thanks for that tip as well Red!
Two brown eyes and an open mind.

Uploaded files:1569 | Total Sales: 27956
Afagundes
3243 posts
<10
Message posted at 09/08/2006, 22:29:41 PM by Afagundes
Red,

I think what Costa007 meant was that when you convert the image to 16bits, and do your "magic", photoshop will be more accurate on its calculation, for example, transitions will be more natural when using some filters (like USM).

Of course, you wont get any more information than you already had, since the picture was taken with 8 bits, the diference is the same as using a calculator with more digits, it will truncate less than the other with less digits.

I will try a math example:

Say you want to add 1/3 to 2/3 and subract a half of the result, than multiply it by 2. The result should be one (or as close to it as possible).

If you use a calculator with only 1 digit the result would be:

(0.3+0.6-0.5)*2=0.8

Now if you were using a 4 digit calculator the result would be:

(0.3333+0.6666-0.5)*2=0.9998

Much closer to one.

So, if you start with an 8 bits picture, adjust with cuvers, levels, use the USM filter and so one, you will notice a degradation in your picture quality.

Now, if you change it to 16 bits, and do the same adjustments, you might not have the same degradation, even if you change it back to 8 bits in the end.

Does it make sense to you?

Now, the best is to convert it from RAW straight to 16 bits, of course...
Canon EOS 5D Mark II Canon EF 17-40mm USM f/4L Canon EF 24...

Uploaded files:9567 | Total Sales: 15044
Fleyeing
935 posts
66
Message edited at 08/06/2009, 14:41:12 PM by Fleyeing
Whoops, I revived a 3-years old thread. :-p




Originally posted by Red:
Quoted Message: Converting an 8-bit image to 16-bit doesn't automatically increase tonal range. You are still only using the 256 colors available in an 8-bit image and spreading them out - you end up with the same 256 colors with gaps between the colors and this introduces noise.


Excuse me, but this is incorrect. 8-bit per color channel gives indeed 256 values per channel, but since a perceived color of a pixel is the combination of 3 channels (R,G,B), the actual number of colors is still 265x256x256=16,777,216. What you are talking about is GIF or PNG-8 with 256 distinctive colors = 1 byte/8-bit per pixel. JPG is 3x8=24 bits per pixel = 16,777,216 colors.



The "noise" you see at 8-bit is not caused by "gaps" in the color space, since those fall below the perception threshold. Nevertheless, 8-bit can cause the dreaded banding phenomenon in monochrome gradients, like in cloudless blue skies getting darker from horizon to top. The reason is that all 3 channels go up (or down) 1 step (of 256) at the same spatial point, so you'll get a monochrome luminance step/border of 1/256.

The human visual system detects edges almost purely by luminance, not by color, and its luminance difference sensitivity is 1/500 to 1/1000, well above 1/256. That's the reason you can see banding at 24 bits (3x8) per pixel, but this is purely a luminance phenomenon.



The "gaps" you mention in the Photoshop histogram are caused by luminance stretching, like when it's flat left and right (the image looks dull) and you expand the luminances to fill the entire histogram (contrast enhance/optimize).

Anyways, you could never see color gaps on an histogram of 500px wide on your screen when you go from 16-bit per channel to 8-bit. For that, your screen should measure 16,777,216x2 px wide at least. ;-)



It's true that spatial calculations (involving adjacent pixels, not just one pixel like with contrast, luminance or color space changes) like blur, USM, sharpen work more accurate on 3x16 bits than on 3x8 bits. But to start, the 3x16 bit info should be there, not converted from an 8-bit JPG to a 16-bit bitmap in Photoshop. You can make an omelet out of eggs, but you can't make eggs out of an omelet. ;-)



If you start from the JPG out of your cam, you already bought an omelet. To do real 16-bit work, you should start from RAW (that minimally is 16-bit per channel) and save it in Photoshop as a 16-bit per channel bitmap (done automatically in memory).

When all post processing is finished, save this bitmap in memory as a 16-bit TIFF (or PSD), and let this be your master copy for always. The conversion to 8-bit per channel JPG should only be done as a last step, just before uploading. Never start from this JPG again to do additional post processing: not only you'll lose the 16-bit, but you also suffer from generation degradation since JPG is a lossy compression format and TIFF is not.
Canon 5DMKII, 24-70, 70-200, strobes

Uploaded files:1186 | Total Sales: 4971
67
Message posted at 06/27/2010, 08:49:25 AM by Digitalexpressionimages
To the OP.



There is no such thing as an RGB press or printer. RGB is an entirely different colour mode and no matter how you print the image it has to be converted to CMYK first. When you send an RGB image to an inkjet printer you are letting the software RIP do the conversion which is fine, but do not make the mistake of thinking you are printing RGB. Open the printer and you will find a cyan, magenta, yellow and black ink cartridge.



And for the record: the issue of printing 100% of each plate (colour) is called ink density. There are many problems with having too high an ink density, bleed through is only one of them.
Canon 5D Mark II, Canon 40D, Canon EF 24-105mm L, Canon EF 1...

Uploaded files:649 | Total Sales: 1502
Wjrodel
68 posts
73
Message posted at 07/14/2010, 01:17:31 AM by Wjrodel
haha, yes I would also like to see some one order additional RGB ink for their printer haha, coz it wont happen. but yes, sometimes printing directly from an RBG image is better because some RIPs dont mess up the curves as badly as someone converting them to CMYK in another programme, sometime altering a colour mode can loose information (gamma) if it is not done correctly.
A never resting mind turning millions of ideas

Uploaded files:116 | Total Sales: 144
67
Message posted at 02/22/2011, 14:50:57 PM by Digitalexpressionimages
Yes Wjrodel is right. For some reason, allowing the RIP to do the conversion usually results in more vibrant color. Keep in mind though that the CMYK gamut (range of colors) is in the neighborhood of 16,000 colors while the RGB gamut, regardless of the color profile you're using (sRGB - Adobe 1998) is up around 16,000,000. You're going to find a few colors that do not reproduce exactly as they look on screen. It's been that way since gutenberg invented the press.
Canon 5D Mark II, Canon 40D, Canon EF 24-105mm L, Canon EF 1...

Uploaded files:649 | Total Sales: 1502
Page:
1
Showing posts 1 - 10 of 10
Current Server Time: 08/01/2014, 11:19:12 AM