Correct / Largest image size??

[tasadam]

OK It likely exists somewhere, but I cannot find it.
I shoot RAW with Nikon D200, and PP in CS3.

I open my raw image. The workflow options at the bottom say
Adobe RGB (1998); 16 bit; 3872 by 2592 (10.0MP); 300ppi
I understand from something I read somewhere when I learnt (how little I know) about this that this is the correct settings because it matches tha native resolution from the D200.
It is possible to change the crop size setting in Workflow Options to 6144 by 4113 (25.3MP)
When I have finished my adjustments in RAW mode (typically Basic, Tone Curve, Detail, and Lens Corrections to fix CA), I then may or may not tweak the unsharp mask before turning it into an 8 bit file and saving as a JPG.

I am editing my files for sale on Redbubble or for printing myself through one of the online print companies such as Digital Works or RGB Digital.

Questions -
1. Is it correct to change tthe crop size setting to the largest one if I am looking for the highest quality print?
Or should I leave it as 3872 x 2592?

2. When saving the JPG, the jpeg options screen comes up - under Quality, if you choose maximum the value (size?) goes to 10. But it is possible to push it to 12.
What advantage is this giving my image? I can't see it, apart from a larger file.
Like, if I use 6144 x 4113, a 5042 kb jpg at 10 will give me a 12,690 kb file at 12.
And if I use 3872 x 2592, a 2534 kb jpg at 10 will give me a 7508 kb file at 12.
Ordinarily I thought a bigger file would hold more data, but where is it getting the data? A 10.2Mp camera cannot prodice +25Mp files, can it?

3. What else can I do in PP to give me the largest clearest images possible from the raw data from the D200?

Any help would be very much appreciated.

[damead]

I'm sure there are others more knowledgeable than I, but my two cents' worth (FYI, I also use a D200):

When sending the image to a commercial photo lab to print, I refer to the lab itself for instructions whether to send it with or without interpolating to the ordered size. I recommend that you pose the resizing and jpg quality setting questions to Digital Works and RGB Digital if they don't cover it in their instructions; you may well do it two different ways for the two labs.

"A 10.2Mp camera cannot produce +25Mp files..." No, but resizing can and does by making up data between the real data. If I do resize in Photoshop, I have complete confidence in its interpolation feature. On the other hand, years ago I sent a jpg from my 6.1 Mp D100, 2000x3000 pixels, to Bay Photo and they printed a stunning 20"x30" canvas print with edges so sharp you could shave with them - and no noise.

Other suggestions for max quality: Calibrate your monitor, on exposure push your histogram as far to the right as possible without losing detail (brightening an underexposed image posterizes dark areas, whereas darkening an over-exposed one does far less damage), and do advanced (separate channel) noise reduction where practical. But you know all that.

[Mr Darcy]

1. If you save at a larger size than the original, then the software doing the resize makes up the extra information. Some software does a reasonable job of guessing, other S/W does a hopeless job. But remember that it is ALWAYS a guess. ("Interpolation" is just a fancy word for guess) If there was a narrow feature in the real world that fell between the pixels in the original, it will still be missing in the resized photo.

2. JPG is an inherently lossy format. Every time you save something in JPG something gets thrown away. If you do it often enough, the photo will become unusable. If you choose quality (12) over size, something is still being thrown away, it will just take longer till you get to the point where you may as well throw the photo away; but you will get to that point if you continue to open and save the file. If you only save the file once, you probably won't notice the difference between the RAW and the JPG.

3. As damead says, talk to to your lab. Ask them what they want. If they say "Huh?" go to a different lab. A good lab will probably ask for TIFF ( a common lossless format) and CMYK (printer ink mode) and the highest native resolution out of your camera. They should also be able to supply you with a profile so you can match your saved file to the printer and paper they use.

[aim54x]

talk to to your lab. Ask them what they want. If they say "Huh?" go to a different lab. A good lab will probably ask for TIFF ( a common lossless format) and CMYK (printer ink mode) and the highest native resolution out of your camera. They should also be able to supply you with a profile so you can match your saved file to the printer and paper they use.

Very good point Greg. The number of people who post process on uncalibrated screens and then scream and shout when they look at it on a photo teller or on paper just amazes me. I dont have my screen calibrated but I am aware that it will different. I must really get a better LCD and calibrate it, but even then without calibration to paper you are still going to get some variation. I guess this is why a lot of serious photographers print at home.

[Murray Foote]

To add a little to what Greg said:

in CS3 you resize prints through Edit/ Resize image. There are two kinds of resizing here:
(1) There is resizing without interpolation so that as your image size increases, your dpi count will go down. This does no harm to image quality as long as you stay above about 180dpi.
(2) You can resize with interpolation so that for example you increase your image size and stay at 300dpi. on the one hand there is a theoretical advantage in having a dpi that is an even fraction of the native printer dpi (whatever that might be if it's a lab) but in any case this is likely to be more than cancelled out by the effects of interpolation. So in most cases you would only upres for printing like this if otherwise you were in danger of falling below 180dpi.

What Cameron says is also true. If you do not have your monitor calibrated with a good colorimeter such as the Eye-One Display2 then your results are likely to be haphazard. If you pick the wrong lab then your results will be haphazard even with a well profiled monitor. If you have a good monitor profile and a good printer profile then you can print out at home and get pretty damn close first up. in fact you should generally be able to get better quality at home than you can get from a lab and depending on your output, it will be cheaper from somewhere round 10x8 up. But when you want critical quality for an exhibition print, pretty damn close may not be good enough and you are likely to need to appraise the print under good light in its own terms and do some tests. Probably only a few but as many as it takes. And that's another argument for doing everything yourself (as long as your printer can deliver the size you want).

Regards,
Murray

[Mr Darcy]

(1) There is resizing without interpolation so that as your image size increases, your dpi count will go down. This does no harm to image quality as long as you stay above about 180dpi.

Can't say I entirely agree with this. What it means is that if you print the size of a billboard, each dot will be the size of a golfball. If you look at it from 50cm away it will just look a big blobby mess. If you look from the other side of the street, it will look fine. (I am making the size up, but you get the point) Think impressionist, or better, pointillist painting.

[Murray Foote]

Well, that's true. I wasn't thinking about printing that large. I had in mind A3, A3+, maybe A2.

Regards,
Murray

[Aussie Dave]

Q1:
I would leave the crop size (when in ACR), to the one that matches your camera.
Should you then need need to resize your image to suit a desired output size, you can do so via the edit/resize tool (as indicated previously)

Q2:
The higher the number for JPEG quality, the "less" compression is used on the image.
Therefore, the same file saved at a quality of 10 will be somewhat smaller (in Kb's) than the same image saved at a quality of 12.

Q3:
I always resize my images to the exact (print) resolution and dimension that is required....and then give it to the lab for printing.

In order to do this, you need to know:
a) what resolution (in dpi) the lab prints in for the size you are choosing them to print for you.
Typically anything smaller than 8" x 12" will likely be at 300dpi...bigger than this may reduce to 200dpi....and the larger the print gets, the smaller the dpi usually drops to.

b) what size print do you want ?
the simple formula of [length in inches x dpi] and [width in inches x dpi] will give you the pixel dimensions for your image (at the chosen dpi).

eg. For an 8" x 12" photo (being printed at 300dpi), you will need to resize the image in CS3 to:
8" x 300 = 2400 pixels
12" x 300 = 3600 pixels

So, your 3872 x 2592 raw image would need to be "down-sized" (or cropped) to 3600 x 2400 (@300dpi) in order for the lab to print this at 100%

You could take the raw image size to the lab and they will convert and make it fit for you (but depending on software used, this may not give you the same quality).

One thing to keep in mind is that dpi does not really come into play when viewing images on your monitor, as it's the pixel dimensions that determine the size on-screen.....but when you talk about "printing" the images, dpi determines the size of the actual print, from the file you see on-screen (hence Dots Per Inch).

As a test, make an image that is 1800px x 1200px (@300dpi)....then make another image at 1800px x 1200px (@ 100dpi).

On-screen they will both be the same size and look the same.....but if you were to print the two out, the first would come out as a 6" x 4" print and the 2nd would come out as an 18" x 12" print.....however the 2nd image would have a lot less detail close up, but would look OK when standing further away (similar to how bill-boards work).

If you can understand the relationship dpi has to an image when printing, the rest is easy

Hope this helps...

[photoaffinity]

great answer dave

[tasadam]

As a test, make an image that is 1800px x 1200px (@300dpi)....then make another image at 1800px x 1200px (@ 100dpi).

On-screen they will both be the same size and look the same.....but if you were to print the two out, the first would come out as a 6" x 4" print and the 2nd would come out as an 18" x 12" print.....however the 2nd image would have a lot less detail close up, but would look OK when standing further away (similar to how bill-boards work).

If you can understand the relationship dpi has to an image when printing, the rest is easy

Hope this helps...

I've been scratching my head over this for a while, coming back to it now and then, and I think I have figured it out.

I couldn't figure out why the second one would print out bigger if it's at 100dpi when the 1st one is 300dpi.
Now I think I get it. I have always understood DPI - the number of dots in an inch - the more you have, the clearer the picture.
I was thiking for a while, did you mean that the 1st one prints out at 18x12 being it's the 300dpi one, and the 2nd at 100dpi would print out at 6x4. Like it all depended on the printer dpi capability or something.

But now it makes sense (I think...)
It's just a setting to tell the printer what to do. The image is the same. But the 1st, at 300dpi, (1800x1200) / 300 = how many inches (6x4).
And the 2nd, set at 100dpi, 1800 pixels at 100 per inch would make a 18 inch print.
Does DPI do anything else to the image?
I had a bit of a play and scrutinizing myself to see what I could detect, they look the same to me when I save at 100 or 300 dpi so surely it's just a setting for printing?

By the way, I only save to jpg once. Either I start from RAW again and use the XMP file, or lately saving more important stuff or stuff with more PP work as a photoshop PSD file. Boy, doesn't that take some disk space!
Is this a normal thing to do?
If I am going to be making changes to the image dpi for different print sizes, is it better to do it from the PSD file or can I do it to the final JPG file already produced?

The last step in my workflow is IMAGE - MODE - 8 bit then file - save as - jpg, quality 12 (was using 10).

Thank you to everyone for responding. Helps me to get a proper grip of this stuff.
Now I've gotta go learn what to do with an ICC file. Image - assign profile - icc file name that I have imported...
And I send that to the printer no matter how out of whack the colour looks?

Monitor calibration is a whole new ballgame. Expensive tools eh? More than a monitor.

[aim54x]

Monitor calibration is a whole new ballgame. Expensive tools eh? More than a monitor.

hmmmm not if your looking at 30" EIZO's....I wanted one possibly two with my next computer then decided not to think about it, but I will have to get into calibrating my current (and my next) monitor soon!

[tasadam]

on exposure push your histogram as far to the right as possible without losing detail (brightening an underexposed image posterizes dark areas, whereas darkening an over-exposed one does far less damage),

Do you mean you do this in the camera - take your photo, look at the histogram in the camera and have the bulk of it to the right without any running off the edge?
When shooting, I typically use the RGB screen to have any highlights on the shot flash at me.

Or do you mean you adjust the exposure level in Photoshop to get the histogram looking better?

and do advanced (separate channel) noise reduction where practical. But you know all that.

Um...
No.
I'm sure I will be able to google how to do separate channel noise reduction, I assume it's just one of the many billions of options in photoshop.

[Murray Foote]

on exposure push your histogram as far to the right as possible without losing detail (brightening an underexposed image posterizes dark areas, whereas darkening an over-exposed one does far less damage),

Do you mean you do this in the camera - take your photo, look at the histogram in the camera and have the bulk of it to the right without any running off the edge?
When shooting, I typically use the RGB screen to have any highlights on the shot flash at me.
Or do you mean you adjust the exposure level in Photoshop to get the histogram looking better?

Yeah. in camera. Most of the detail is in the brightest part of the image you capture and you don't want to waste it. So look at the camera's histogram. If you have a histogram that shows separate channels then look at that as well.

There are two caveats to this:
(1) If you push it too far and you destroy all detail in an area then that's it, you can't get it back.
(2) You can't determine this accurately from your camera's histogram unless you are merely shooting sRGB and JPG. If you are shooting RAW and processing in Lightroom (probably applies to other RAW processors), then you may be able to recover detail that seemed irredeemably blown out on your camera's preview (or on the default image on your PC). Up to a point. Difficult to nail that point in advance.

and do advanced (separate channel) noise reduction where practical. But you know all that.

Um...
No.
I'm sure I will be able to google how to do separate channel noise reduction, I assume it's just one of the many billions of options in photoshop.

I think he's being obscure here and passing it off as elementary. There are many ways to reduce noise but it's usually only necessary on modern DSLRs if you are really pushing the boundaries.
- Lightroom has reasonable effective chromatic (colour-based) noise reduction.
- There are also third-party utilities such as Neat image which were more important in the days of noisier sensors.
- You can also use the blur filters in Photoshop to a copy of an image on a separate layer and use a layer mask to apply blur (reduce noise) on specific parts of the image
- At a level of complexity beyond your question, you can potentially use edge masks to retain detail in some parts of the image while you blur others

Usually you wouldn't want to blur/ reduce noise on the colour channels because they contain detail - at least in RGB. What he is probably referring to is converting to LAB in Photoshop, blurring the A and B (colour) channels, sharpening the luminosity channel (say with the relatively new Smart Sharpener) and then converting back to RGB. This can often be an effective way of dealing with chromatic noise.

Regards,
Murray