general questions about prints

I am just starting to think about printing on requests for my site. I am curious, its too late for the photos I already have, but I should start if....

Do you all use tiff rather than JPEG? Does it make a difference after editing in terms of print quality.

Also...for the digital shooters...does shooting RAW make that big a difference when printing? After PS editing?

Remember that JPG is a lossy, compressed format, while TIFF is lossless. I honestly see a HUGE difference in quality between shooting in RAW vs. JPG, so yes, RAW creates a superior print. So long as you're saving your editing in TIFF (since you can't actually print from RAW, you have to convert it to TIFF or JPG first) there shouldn't be any quality loss.

alright. I can't process RAW images through PS, because I don't have CS2, and the plugin on their website seems to only work with CS2. My Mac has a preview program that will convert RAW to TIFF. Is there a major difference between different applications converting? Or is it all the same? Thanks...

Sorry for the noob questions! yup..thats me...waving..im a noob! look at meeeee!

Just to let you know Jon, I'm curious to learn the answers to your questions, too. I've never tried preparing my photos for print. I also don't have CS2. I do shoot in RAW but convert to JPG (using Canon's Digital Photo Professional) to process in Photoshop Elements.

Another noob.

hi noob

There are any number of ways to convert from RAW into an editable format. ACR (the converter built into Photoshop) is just one alternative. You don't mention what camera you have so it's hard to offer alternatives.

RAW is simply the digitized readings directly of your camera's sensor with no processing applied. It's not really an image format in the sense of TIFF or JPG. You need to first convert from RAW into an image format that will allow you to apply editing and save a processed file.

Yes, JPG is a lossy format - it throws away image and color data in order to conserve space. Once it's saved you can never regain it. Moreover, JPEG only supports a 8-bit of color depth, vs. 16-bit in TIFF. This makes a substantial difference in images with very fine tonal gradation. This is why TIFF and other loss-less formats (like Photoshop's native PSD) are preferred.

Rather than describe every aspect of the process, I suggets you start at the beginning - there are a lot of good resources for raw conversion online to help you gain more familiarity with the workflow. I will then suggest a good book for whatever imaging software you use to process your work. If it's some version of Photoshop, I suggest "Photoshop Artisrty" by Barry Haynes.

Guy

Sounds like you are using a Mac...if not, a free program that I use for RAW conversion is called Raw Shooter Essentials by Pixmantec. When using this program any changes made to exposure, shadow contrast, highlight contrast, or any other settings are saved to a separate file, so that they do not affect the original.

I still use RSE 2005, but I was out on Pixmantec's site and they have RSE 2006 available for free (or a preimium version that you have to pay for). I have never wanted anything more than what is in the free version anyway though.

Nathan

How do you want to print? Ink jet in your home? Send out for lightjet? Most labs that use lightjets or something similar only take 8-bit jpgs. No one ever complains that the resulting prints lack 'fine tonal gradations', although I often keep the image 16-bit/uncompressed until I save it for printing.

With RAW you can fix exposure and white balance settings before conversion, and also avoid less than optimum in-camera sharpening. Of course you can decrease exposure and adjust color balance for jpg's in Photoshop as well. If you expose correctly and avoid over sharpening in camera then RAW might not get you any noticeable quality increase over jpg.

For RAW conversion, also check out Bibble - http://www.bibblelabs.com

Available for Mac and Windows, and inexpensive.

FYI Pixmantec was recently bought by Adobe; noone knows what will happen to Raw Shooter Essentials but it's likely it won't survive.

I do all my work in TIF or PSD files, then save as a JPG for printing at a lab. The lab I use requires JPG files. The results have been great. But I don't save as JPG until the image is ready to print.

I just saw what Nat mentions, the acquiring of Pixmantec by Adobe. I love RSE, that makes me sad.

Nathan

So do the majority of you folks use a lab rather than print inkjet at home?

I get my prints done at Colorado Camera on Kipling, and there is a gentleman there (I forget his name off of the top of my head) who has been very helpful with this type of information.

The DSLRs from Canon (I can't speak to any other brand) shoot at 12 bits. Processing at 16 bit saves all of the 12 bit data, but also adds 4 bits of blank data (which increases file size). Obviously, processing at 8 bit, you lose 4 bits of data. But all printers convert to 8 bit before printing, so how important are those 4 bits? Some people say they can see a difference. I can't. I am actually going through the process of converting all of my 16 bit files to 8 bit, to save space.

The formula for file size is File Size = (pixel dimensions x bit depth), or, on my 10D, 2048 x 3072 x 24. After you divide out to get MB, it equals 18 MB. If I end up with a file larger than 18 MB it means I have added information that was not there. Basically, PS (or whatever converting software is being used) is guessing what to do with the extra pixels that have been added based on the pixels around it.

The guy at Colorado Camera told me, when you resize a photo, you should do it by lowering or raising the dpi. What, try to print at less than 300 dpi? Yes, at a lab that uses a photo paper printing process, or has a pro ink jet printer with the algorithm to increase the 200 dpi to 7200 dpi (the dpi that the printer at the lab uses, meaning ALL prints go through the algorithm, unless you shoot at 7200 dpi). The way it was explained to me was that it is the same as enlarging a slide (or negative). The "file size" is what it is. You can't add "data" to the negative. When you enlarge, you are simply lowering the dpi. If a slide is the equivalent of 600 dpi (pulling a number out of the air, it is not an exact statement), then to double the size of the slide, you have to decrease the dpi by half, to 300 dpi. I have been shown examples of huge enlargements printed at 80 dpi, and you can't tell the difference from film. Try enlarging a 35mm slide to 20x30 and tell me if it is as sharp as the slide itself.

I haven't done this test myself, but I have seen tests where full size, lowest compression jpegs have been saved over and over, and it took many saves before there was a noticeable drop in quality.

My work flow has evolved over the years, with each new bit of information I learn. Right now, I shoot in RAW, then do my processing in 8 bit TIFF format. I save the RAW files (think of them as negatives), so I can always go back and start over. I save my TIFF files as masters, and then save a high quality jpeg for printing. I also will print from TIFF (as long as it is 8 bit) because Colorado Camera will print TIFFs.

David,

I hate to tell you but your lab "guy" has misinformed you a little.

It's true that all printers use 8-bit drivers (with the exception of the new Canon iPF5000, I think). The higher bit depth is not required for printing, but it makes a big difference for editing and processing the file. It makes for smoother and more detail interpolation (resizing) and gives you more range to adjust contrast and color before converting to your output device. If you start with 8-bits per channel of color information you run the risk of artifacts like banding and loss of detail, and in the case of JPG of compression-related artifacts.

The statement abotu resizing a photo by changing the dpi is simply not true. Every device has its own natuve resolution. If you feed it a file that is set to a different resolution, the device will interpolate it. By setting your file to the device's native resolution you are in control of how interpolation occurs and can preview the results.

It all comes down to how involved you want to be in decisions related to your images. In my case it's a lot, for others it may not be as much. When you capture JPG, you're letting the camera make decisions about color, contrast, sharpening and other parameters. When you resize your files by changing the dpi setting you're letting the printer decide how to interpolate the file to the actual size required.

Guy

Guy, you make good points. I have heard both sides of the 16 bit vs 8 bit discussion, and have seen examples. All I can really go by is what I have seen in person, and the prints I saw at Colorado Camera were outstanding, even at 8 bit processing, 80 dpi, larger than 20x30. I am by no means an expert, and am only regurgitating what I have been told, or have experienced.

I don't agree though, with your logic on interplotion. I am going to trust the quality of a program that costs more than what I make in two years (what is used at the lab) than anything I can afford to use at home. While you are correct that I am letting the printer decide, it is going to be more accurate and better quality than anything I can produce. And, if it is a good lab, they will do reprints if it is not to my expectations. While I actually enjoy the processing side of digital, when it comes to printing I absolutely don't want to deal with it. That is why I don't have a printer at home, and why there are labs. I have been very happy with everything I have had printed there, much more so than with Wolf or any of the consumer printers (Walgreens, etc...)

Your statement about JPEG is true, but it also does not mean that it cannot be corrected in PS. It is just a different workflow. I shoot RAW because I find it easier. I don't think one is better than the other, just different workflows.

I have been told by the folks at Photocraft that they trust their Lightjet's interpolation algorithms more than they trust people to be able to interpolate for the Lightjet through PhotoShop. I imagine in most circumstances that is true; the chances of your doing a better interpolation job than, say Epson's RIP can do knowing the printer's dithering mechanisms are slim, though not non-existant. Still, that's not to say that you don't appreciate your own interpolation more than what the RIP determines is "best"...

OTOH, maintaining the highest bit-depth image until the very end is very important for best results, especially if you do any manipulation of curves/levels or use the Highlights/Shadows tool in PS CS2.

Stacy - to answer your question, I think we have a mix of inkjet and lab printing people on the board. I used to print on Photocraft's Lightjet, but now I print most of my work on an Epson R1800 (I do almost exclusively color glossy...). The process is the same no matter where you print, though; stick to the highest bit-depth you have unless you need to downconvert (usually for printing...); similarly, keep to the widest colorspace you have available (usually Adobe RGB) until you have to downshift for printing.

One other comment. At Colorado Camera (and I am sure others) for any print at 20x30 or smaller they use a photo paper printer (not sure what the actual name of the printer is). My understanding is that it uses lasers to expose the photo paper, so it doesn't use dpi in the traditional inkjet printer way. You probably understand this better than I do, and can correct me. Like I said, I am really only repeating what has been told to me.

Hi Les,

I'm not sure that keeping the widest gamut space is the best way to go if your intended output device is relatively narrow gamut, i.e. lightjet, Chromira, etc. I think the argument is that there's no reason to waste space, so to speak, on colors that are out of gamut for your printer anyway. I could be wrong, and it probably matters little if your working space is 16-bit.

How do you like the R1800 for glossy? I got rid of my Epson 2200 because it didn't do a real good job at glossy prints.

David, the printer is likely Lightjet or Lambda, which use lasers. They are continuous-tone printers but they still do have a native resolution.

Tony - the number of bits has nothing to do with the working space. The working space sets the boundaries of reproduceable color. The number of bits determines the degree of gradation you get within that space.

Guy

I have the R1800 and it does a very nice job on glossy. There is still the potetial for bronzing differential between colors at high angles, but the gloss diferential is gone. I don't have experience with the 2200 or 2400 to know how it does in comparison to those printers, but I can say that it does a better job on glossy than any printer I'd personally used before.

That said, I still prefer to print on matte if it is going to be mounted behind glass for the simple purpose of reducing reflections.

"Do you all use tiff rather than JPEG? Does it make a difference after editing in terms of print quality. Also...for the digital shooters...does shooting RAW make that big a difference when printing? After PS editing?"

I'll throw my support in for RAW/TIFF captures for serious photo printing, or rather for processing before the print stage. It's a must, if you do ANY editing beyond the camera capture. I too send off a high quality JPEG for my large prints, and there is no discernable difference between a TIFF print and a JPEG with a "10" setting (after some quality processing). But saving as a JPEG is a final step. Honestly, my prints would look like crap if I started with a JPEG (specifically with color gradations, color noise and apparent sharpness). At least when directly compared to a print that originated from a 16-bit TIFF file ... even after it's been converted back to an 8-bit JPEG file. I say this because I've tested it. You really can't appreciate this fact until you make a print that is 16x20 and larger, and do some comparisons yourself.

Someday, when the technology catches up (like Guy mentioned with the new Canon printer), you'll appreciate having a master file that took full advantage of your imaging sensor. Soon, we'll be debating 16 vs. 32 and so on. 8-bit will be a thing of the past ...

Good question and a good time to make this decision.

Tony,

I love the prints I get on my R1800. From everything I've read it's leaps and bounds above the 2200, no matter what paper you're printing to. But if you're doing matte or lots of B&W, the 2400 is the printer you're looking for. On glossy paper it's [the R1800 is] untouchable right now.

As to color space, the same principles apply for gamut as for bit-depth; there are Photoshop operations you might perform that would bring colors from outside the gamut of your printer into gamut, just as there are operations you could perform to bring the low bits up to significance in the 8-bit gradation spectrum. No color space does a perfect job, though; the Epson R1800 colorspace is much, much broader in some of the greens than Adobe RGB or even the Pro Photo profiles provide, but is deficient in other spaces... But, e.g. I wouldn't pre-process my images for the LightJet in the LightJet/FujiCA profile; I'd wait until after I finished with my touch-ups.

Doesn't there become a point of overkill, though? I'm not saying we are at that point yet, between 16 bit and 8 bit, but it does get to a point when we just can't tell the difference. It's like mp3 encoding. It is just silly to encode an mp3 over 250, because most amp and speaker systems can't reproduce the frequencies that have been saved, and no one but a very small handfull of people would be able to hear a difference anyway (no matter what most people will try and say they can hear! ) When do we get to the point that the human eye can no longer see a difference, and we are only making the files bigger for our egos, not our eyes?

For the most part, large prints are designed to be viewed at a further distance than smaller prints. Even 35mm film, when enlarged to poster size, will show grain and fuzzines, but average people won't be looking at a poster sized print from 2 feet, it will be closer to 10 feet. I have read some articles that say we as photographers have gotten overly critical of the quality of digital prints, because we look at our prints (even poster size) inches from our face to search for defects, when we didn't do that as much with film enlargements. Take that for what it is worth, it is just an opinon I have read, and I know I do it! I mean, how good can it be if it isn't tack sharp at billboard size!

I have processed a lot of my pictures at 16 bit after being told by people I respected I should be doing it that way, but went back to 8 bit after I was told by other people I respect that it was a waste of space. I am going process a test RAW file at both 16 bit and 8 bit, to do a side by side comparison of the same image. I swear, it starts to be like getting advice on your golf swing! The more advice you get, the worse your swing becomes!!

Ah, but you aren't RIPping an MP3 so that you can mangle it for use in your own widely-released Rap CD, you're RIPping an MP3 to listen to it on (usually questionable) speakers. If I were just pulling pictures out of my camera and throwing them straight to a printer (say with a consumer camera and print system from Kodak), I'd be good with JPEG-fine at 8 bits, but I'm not; I process my digital images to give them some punch so I can compete with Velvia or similar film prints... As Guy points out above, if you're processing the image in PhotoShop (adjusting levels, fixing highlights/shadows, playing with layer masks...), you really want those extra bits - they even out the image significantly and fill in details that would otherwise be lost.

Once you're done processing your test file in 16-bit and 8-bit formats, take a brief stroll over to the Levels or the Histogram for your image and look at all the holes you've created in your image when processing it... Those appear as banding on your final image, and while some people won't mind, others will go "Eww!"; if you cringe when you see ugly compression banding on DVD movies, you're part of the second group...

I actually rip mp3s so I can listen to a lot more music on a CD, and have it a lot more portable. I have a media pc in my entertainment system, so I listen to them on a Yamaha amp with high quality speakers. I also have a Bose system in my car. Music is even more important to me than photography (if only slightly ). I have been playing in bands since I was 17, and still to this day write music. With all of that, I rip CDs at 192. At 128 I can hear the difference, but going over 192, there really isn't much difference, and over 250 or so, you just simply can't hear any more quality. The human ear just can't hear it.

Like I said, I am not questioning the difference between 8 bit and 16 bit. It was more a comment going further with Jesse's post, about eventually getting even higher bit rates. It could get to the point where it just doesn't matter. The human eye will just no longer to able to see the difference. It wasn't meant to make any points regarding whether 8 bit or 16 bit is better. I was just following the (off) track that the thread was taking.

For better or worse, I understand. My photo equipment only passed my audio setup last year when I picked up my D2X. Most people don't spend pro-camera levels of money on audiophile speaker systems And yes, there *is* a level where it's too much for most purposes.

But... I just did my first HDR image, and I have to say I was happy to have 32-bit image data during that process. Brought out all but the brightest (sun disk) highlights of the picture and at the same time preserved even the backlit sides of the seastacks I was photographing (though I didn't actually use that data...). So there is a ways to go before we reach "too much" for all uses.

David, my last comment about higher bit depth was somewhat tongue-in-cheek, an observation of the nature of digital camera technology debates that take place online.

I think the subject of 8-bit vs. 16-bit was indeed sparked here and is perfect for Jon's post. I know that I certainly regret using 8-bit for so many years (for real life reasons) when 16-bit was available - especially with the cost of storage dropping dramatically as camera file sizes increase.

Jon, if you are going to print and/or sell prints, definitely use the highest quality option your camera allows. And keep a master file of each photo in at least AdobeRGB and 16-bit. You'll will appreciate it later.

woah! thanks for all the replies everyone! It is currently 4:30 AM in Helsinki where I am....but I will read more when I am actually awake!

woah! thanks for all the replies everyone! It is currently 4:30 AM in Helsinki where I am....but I will read more when I am actually awake!

Guy -

Your comment actually supports my point.

You said: "Tony - the number of bits has nothing to do with the working space. The working space sets the boundaries of reproduceable color. The number of bits determines the degree of gradation you get within that space."

If the number of bits sets degree of gradation within a space, then wouldn't it stand to reason that in a wider space the gradations are further spread out? Isn't this bad? It seems to me that ideally you'd want the color space to cover the color range of your output format but not much more.

OK, had the chance to read this whole thing. From what I gather...best method....

1. Shoot in RAW
2. Convert to lossles TIFF (8 or 16bit)
3. Edit in TIFF, save as TIFF for "negative"
4. Convert to JPEG for posting/printing

The last thing I would love confirmation of. This is directed to the MAC users. It's cool to open the RAW in 'Preview' and then 'save as' to lossless 16 bit TIFF right? No problem with doing that instead of using a RAW program? I'm going to edit in PS anyways, and I'll save the RAW image seperate.

Kapeeeesh?

If you're editting outside of your normal RAW program, then save to 16-bit TIFF, yes.

Tony, if you make 16-bits worth of gradations in an inch, you can only represent an inch; anything wider than an inch is outside of your bounds. If you make 16-bits worth of data represent a foot, you'll have a lot more "width" to your data, but as you suspect, you'll have fewer points per unit of space. However, if your "output device" was an inch, but your input had data across a whole foot, you'd want to manipulate the input data so that it was well-represented within the inch-wide output space; that is where you'd want to maintain the entire original data space - so that you could manipulate it until it fit your output space to your liking.