Printing Size

[Ferry Zievinger]

Advertisement (gone after registration)

I have a request from a restaurant owner to sell several of my photos to use as prints in his is restaurant, the photos are taken with the X1, RAW format at max. resolution, my questions:

 

-What is the maximum size prints I can get from these files? These prints will be seen from a distance of about 4-5 feet.

 

-These will be printed by a commercial printing company, do I provide the files as JPG or TIFF?

 

Thank you for taking the time to answer and if you need more info let me know.

 

Ferry

[algrove]

I always print my files from TIF. JPG I only use for the internet. It depends on the images, but if they are viewed as you say, larger than 13x19 (A3+) I would suggest since I view my (A3+) at about 3-4 feet. So just as an approximation, it seems 17x22 would be better for the 4-5 feet you are suggesting.

 

Look around your island to view prints already existing on walls and then see which ones you view at 4-5 feet and then measure them to see their size.

[hlockwood]

I have a request from a restaurant owner to sell several of my photos to use as prints in his is restaurant, the photos are taken with the X1, RAW format at max. resolution, my questions:

 

-What is the maximum size prints I can get from these files? These prints will be seen from a distance of about 4-5 feet.

 

-These will be printed by a commercial printing company, do I provide the files as JPG or TIFF?

 

Thank you for taking the time to answer and if you need more info let me know.

 

Ferry

 

Ferry,

 

Algrove's comment should be your first consideration. Convert your dng image to an uncompressed tiff. Jpeg is a lossy compression, and pixels will be discarded. Not good for a large print.

 

After that, things get a bit complicated. My approach would be to avoid resampling, especially up-sampling by the print driver. This means choosing a proper fraction of the native resolution of the printer which, in turn, will dictate the non- resampled print size.

 

In the case of an Epson printer, the native resolution is 1440 dpi (not ppi). So, 1440/n, where n = 1,2,3,4,5,6, (not 7) 8,9..., and the choice of one of these print resolutions will yield a print size that will not be resampled.

 

For example: Noting that the max resolution for the X1 is 4272 x 2856 ppi, let's do the arithmetic.

 

1440/6 = 240 dpi and 4272/240 = 17.8 for a non-resampled print of 17.8 x 11.9 in.,

 

1440/8 = 180 dpi and 4272/180 = 23.7 for a non-resampled print of 23.7 x 15.8 in., and

 

1440/9 = 160 dpi and 4272/160 = 26.7 for a non-resampled print of 26.7 x 17.8 in. and so on.

 

If a higher resolution, say, 360 dpi were chosen for any of the above examples, the print driver would add pixels according to some (unknown) algorithm; a lower resolution would cause the subtraction of pixels. Maybe noticeable, maybe not; you be the judge.

 

Finally, the viewing distance. Subjectivity reigns here. Again, Algrove's suggestion to go out and look at some prints from the expected viewing distance is a good one. At 4-5 feet, the last example above (26.7 x 17.8) would come close to Algrove's 17 x 22 suggestion. (But keep in mind that some viewers will get within 4-5 inches, not feet. )

 

Not an Epson printer? Start over.

 

Harry

[sanyasi]

I'd talk to the printing company and see what they want in terms of resolution, color space, etc.

 

Jack Siegel

[jaapv]

Good idea. May I point to the printing thread immediately above this one ?

 

 

http://www.l-camera-forum.com/leica-forum/digital-post-processing-forum/115989-digital-printing-pixels-resolution-resampling.html

[hlockwood]

Jack Siegel's (sanyasi) recommendation to ask the print company what they want (for the image file) is, of course, quite correct. But let's keep in mind that not all "print companies" are created equal. I'd go much further and ask for details about the final print itself. Will the inks be dye based or pigment based? And what specific paper will be used? What are its archival properties? And since the final print is likely to be resampled upward, does the print company take special measures to assure the best possible outcome by using software optimized to that end. RichC's (very informative) post recommends avoiding resampling (as I do) when possible, but, when not possible, he provides suggestions for software dedicated to that task. I couldn't agree more.

 

Just a bit of background: I have been a serious amateur print maker for many years. All of my work has been in B&W. I started with MIS quadtone (4 shades of gray) inks, and then moved on to Jon Cone's K7 (7 shades of gray) inks. These are carbon pigment inks intended for maximum longevity and, when printed on pure cotton rag paper, assure an archival print of tonal stability and paper-color constancy for many decades; carbon on cotton.

 

I have no competence (and little interest) in color printing. My post dealt only with the details of resampling, applicable to both color and B&W images. Perhaps Jaap can elaborate on his rather cryptic response. Was my post too detailed? Redundant? Incorrect? I do always look to the experts for guidance.

 

Harry

[jaapv]

Advertisement (gone after registration)

No criticism at all. Just a reference to an useful thread about resizing.

[tobey bilek]

My prohouse accepts tiff or jpeg. More importantly only Adobe RGB. They correct density and color balance as they see fit so that works well if there are reference colors. To get it perfect without reference colors, I need a calibrated quality screen, soft proof to their profile, and tell them not to color correct.

 

From 4/5feet, 180ppi is sufficient for laser printer. Send them an 8x10 test at 180 ppi and see the result. Sharpen file after resize but not so much as to get any halos. This need not be the full file, but a section.

 

Ink jet prints are different probably needing more. Matt and glossy require different settings.

 

If your original is 1800 pixels wide, it will make a 10" wide print. 1800 divided by 180 is 10.

 

A test is key here.

[delander †]

The thing about printing is that the printer driver expects a file of a certain resolution.

In the case of the later Epsons this is 360 or 720 ppi. If you send a 180ppi image to an Epson printer the driver will uprez to 360 and then print. Alternatively if you uprez your file, in PS for example, to 360 you will have control over the uprez and the printer driver will not resample your image. I always use PS to to resample to 360 ppi and then print.

 

Jeff

[hlockwood]

My prohouse accepts tiff or jpeg. More importantly only Adobe RGB. They correct density and color balance as they see fit so that works well if there are reference colors. To get it perfect without reference colors, I need a calibrated quality screen, soft proof to their profile, and tell them not to color correct.

 

From 4/5feet, 180ppi is sufficient for laser printer. Send them an 8x10 test at 180 ppi and see the result. Sharpen file after resize but not so much as to get any halos. This need not be the full file, but a section.

 

Ink jet prints are different probably needing more. Matt and glossy require different settings.

 

If your original is 1800 pixels wide, it will make a 10" wide print. 1800 divided by 180 is 10.

 

A test is key here.

 

Tobey,

 

Did you mean to supply a link in that last sentence of your post. If so, could you provide it?

 

Harry

[hlockwood]

The thing about printing is that the printer driver expects a file of a certain resolution.

In the case of the later Epsons this is 360 or 720 ppi. If you send a 180ppi image to an Epson printer the driver will uprez to 360 and then print. Alternatively if you uprez your file, in PS for example, to 360 you will have control over the uprez and the printer driver will not resample your image. I always use PS to to resample to 360 ppi and then print.

 

Jeff

 

Jeff,

 

Your recommendation contradicts the latter part of RichC's stickey; it contradicts the entire prescription in my post in this thread, and it contradicts the last part of tobey bilek's post in this thread. Why? Because you make your recommendation independent of print size. (And you do mean dpi not ppi, right?) Would you care to comment further on this issue?

 

Those of us who are serious print makers will benefit greatly if we were to get these divergent opinions to converge on the correct answer to this part of our workflow. After all, we are all striving to make the best print possible.

 

Harry

[delander †]

I mean ppi the resolution of the digital image. There is a lot of information over on the Luminous Landscape Forums on this subject. I can't find an explicit post but look for the postings of Jeff Schewe – Adobe guru and insider. There is even more info in the LuLa video 'From Camera to Print and Screen'. Also see this link Adobe Lightroom 1 - Print Tutorial

 

where below figure 5 it mentions the Epson 3800 printers native resolution as 360ppi.

 

The more recent Epson printers have two native resolutions, 360 and 720ppi.

 

As I understand it if you send a 240ppi image to the recent Epson printers the driver will uprez it to 360ppi. If you send anything above 360ppi the driver will uprez it to 720 and you need to have 'finest detail' selected.

 

I prefer therefor to uprez in PS or LR so that I can view the image or perhaps sharpen it before printing. That way the printer driver makes no changes to the image.

 

Jeff

[delander †]

Harry,

 

See reply number 9 in this thread, it is what I do whether printing from LR or PS.

Clarification on Print Resolution

 

Jeff

[hlockwood]

Harry,

 

See reply number 9 in this thread, it is what I do whether printing from LR or PS.

Clarification on Print Resolution

 

Jeff

 

Jeff,

 

I did see your post and disagreed with it in my response. And I went to the luminous landscape link in your current post. What I found was total confusion; almost as many differing suggestions as posts.

 

I'm tempted to walk away from this discussion because both here and over there people are just talking past each other.

 

Harry

[delander †]

Yes it is confusing, not made any easier by the fact that Epson and other printer manufacturer's keep information to themselves, in fact I dont think I've ever seen an Epson recommendation on native ppi. However that being said I trust the opinions of Jeff Schewe, primarily because he is an Adobe insider, although he does not work for them. I have his book on Adobe Camera Raw.

 

Anyway 360ppi works well for me although I suspect that we are looking at small differences.

 

Jeff

 

PS post number 9 I meant Schewe's post on Lula, not mine in this thread.

[tobey bilek]

I did not mean to provide a link, but I use AiProLab.com. They will mail final work to you. Profiles are in down load tab, main page. Their order input is thru ROES. Very nice once you master it. Instructions provided. They do wedding and event photos mostly, but will do other types and they have only a pro base of repeat customers. They shop is in an industrial area and normal consumers will never find it. It may take you a while to get calibrated, but their work is top drawer once you coordinate. Neil will answer your E-mails

 

180 is for photo paper with a laser printer. I would not recommend that for an inkjet like epson. Someone did not read carefully. Also 180 is for 4/5 feet like you requested. I use 240 for 11x14 and sharpen myself and they are as sharp as I want.

[CalArts 99]

Yes it is confusing

 

I don't think there ever was any confusion. And there is no real magic number, either. You are correct about the upsampling.

 

The print driver will upsample the file to its native size it needs to work with whether it's spraying its ink dots at 1440 dpi or 2880 dpi or whatever. It does a second algorithm for the actual dithering (which is pretty good with the newest printer drivers.)

 

Ideally one would probably want to do the interpolation with something pretty robust. It's usually better to let LR or PS do the interpolation even though the latest Epson drivers can do a decent job. Of course if one is using a RIP then one would normally let the RIP do the interpolation for the printer (since that's part of the RIP's function, to override the factory driver's upsampling.) Epson's native numbers are 360 and 720. With Canon and HP it's 300 and 600 (unless that's changed recently.) Here is a better and more descriptive explanation: The Qimage Print Quality Challenge

 

All this is dependent of course on substrate, print size, your eyes, and most of all: subject matter. The resolution issues are kind of meaningless unless you consider subject matter. I think we know that from analog printing, too. Certain images made from high ISO grainy film can look okay blown up in a large prints, while others (a detailed landscape, e.g.) might look awful. The bottom line is WYSIWYG. And everyone will evaluate a photograph differently depending on their own criteria.

 

I think what one really should be doing is always making tests and making their own decisions on what works best in any given situation. Personally I feel a good thing to do is make test prints with different interpolations on different substrate with different subjects. Do it on a contact sheet (several different kinds of subjects on the same substrate and with different sampling rates.) Write the info on the back and keep them as a permanent reference. You can do a few with different interpolation algorithms from LR, PS, third party software like Qimage, and some with the print driver alone. Use a loupe to satisfy yourself.

 

fwiw, I worked commercially in a studio (drum scanning, inkjet and lightjet printing) and each file was tested and processed specifically depending on the printer, print size, substrate, and subject matter, etc.. The Lambda and Océ printers have their own native resolutions, too.

 

I think as far as the OP's question is concerned, they should really communicate with the person doing the printing, Every printer has their own workflow and requirements that they use based on the printer being used and the final product.

[hlockwood]

CalArts 99,

 

Lots of interesting stuff in your post. I'm looking forward to printing some of the challenges offered in the Qimage link provided there. Thanks for that.

 

Evidently, the maximum (although not terribly useful) resolution of the Epson printer is 1440 dpi. You cite the "native numbers" as 360 (1440/4) dpi and 720 (1440/2) dpi. Can you provide a precise definition of "native" and why, say, 288 (1440/5) dpi or 180 (1440/8) dpi are not "native numbers"? I.e., what makes any particular resolution native, a term we bandy about rather casually?

 

Harry

[CalArts 99]

The output is 1440 or 2880 dpi. These output numbers reflect the 'accuracy' of the printer to spit out its ink drops and not 'actual' resolution. Comparing ppi and dpi is not quite a one to one sort of thing, an ink droplet is about volume and is not a pixel dimension. The Epson driver does use 720 ppi as its resolution (for vector graphics with finest detail enabled.) From that it produces its 1440/2880 dpi (ink drop) 'resolution' output.

 

This 'native' file resolution is likely the best way for the Epson software to work when it processes the algorithms for its error diffusion dithering for the 1440/2880 dpi output. (The Océ Lightjet uses 300 ppi for its output, as another example.) And sending it a file that has been interpolated by LR/Adobe (or other software) can often give better results then letting Epson do it. Bear in mind that all this presupposes that one has done proper sharpening for print output and only after any interpolation and at final print resolution (we know sharpening for print output and sharpening for web is vastly different.)

 

(And of course, using 1440 or 2880 is up to variables like substrate, print size, viewing venue, subject matter, etc.. For most images and B+W, it's too time consuming and wastes ink to print at 2880 unless it's warranted depending on those variables.)

 

I agree it's unfortunate that manufacturers won't clearly document everything, but it's no doubt partly because it's a propriety sort of issue. And probably using the dpi metric as resolution sounds a lot better for marketing (and after all, that is what we care about.) RIPs will often advertise their driver's specs because that's what they are all about. Epson won't tell us exactly how their printer drivers work and instead promote more tangible stuff like inksets and color gamut and longevity, etc..

 

Various commercial third party RIPs have their own exclusive dithering and screening patterns. Some are designed primarily for a continuous tone photographic look (like ColorByte's ImagePrint) or for versatility of job queueing, trapping, and speed of production (like Onyx Graphics' PosterShop.) They often use rendering algorithms that are not solely dependent on a specific file ppi (much more complex software with more memory than an Epson driver.)

 

Try some print tests by sending already interpolated files directly to the driver and see what you think. Do some in 240, 360, 720, etc.. Then try some at different ppi (288, 180 as you mentioned) and let the driver do the interpolation and see what you think. You can also try some third party software like Genuine Fractals or Qimage, etc.. You might like what you see with certain subject matter on certain substrate. Or maybe not. Again, a lot is dependent on those variables already mentioned.

 

btw, here's an old article by somebody who was frustrated in trying to get info on the actual ppi resolution from their Epson printer. Despite the age, it might be a helpful explanation: Inkjet Resolution

 

And here's Eric Chan's classic FAQ: Epson Stylus Pro 3800 FAQ

[Advertisement]

Hallo,

Look here: Interesting Leica gear

[CalArts 99]

I just wanted to add one last thing about RIPs. They aren't a cure-all for printing and all RIPs vary in what they can and can't do. And it used to be that the Epson driver was pretty basic but things have changed a lot now.

 

With the latest Epson drivers, I tend to agree with Eric Chan (see the link I posted above) about ColorByte's ImagePrint which tends to be popular with many photographers. Just for the record, Chan is a principal scientist at Adobe and his main product assignment is LR, DNG, and ACR. (Eric Chan)

 

The IP RIP was an important piece of software when the earlier drivers were all that was available. With the newest drivers, its value has diminished somewhat. What you are really getting now is a library of profiles and versatility in layout with batch printing capabilities. Like Chan, I also used to use it with the older printers. But I found its use value diminishing over time (and I have access now to making my own profiles. Yes, it's boring and time consuming but I also only use a few different brands/types of substrate anyway.) It's an expensive product (although nothing like the cost of commercial RIPs such as Onyx) and using a dongle is kind of a pain. And it's not really robust enough in its layout and queuing capabilities for big production commercial use. It's kind of an 'in-between' product.

 

It's a product I'd test out very carefully first and compare with some custom profiles and the Epson driver before making a commitment. I find the screening of the Epson to be very good and its become more difficult now to discern the final product from the final product of the IP RIP.

 

All of them are different and you need to personally decide whether there's enough value for you in even using one at all.