Book Printing Service Reviews

April, 2005

I sent three test images out to different web photo book publishers: mypublisher, sharedink, and fastbackbooks. I got back one of the test images from asukabook. I also printed two of the test photos on three different inkjet printers: HP 8450, Canon ip6000d, Epson R320. My goal was to find the best photo reproduction quality for a coffeebook style presentation of my travel photos.

The photos I sent out are:

  1. A picture of several folks wearing bright bicycle jerseys in neon green, red, and yellow, sitting in a restaurant in France. This image has some good shadow details on the bicycle shorts. Taken on my Minolta A2 camera as an 8 Megapixel image.
    Original sRGB image: booktst1.jpg
  2. A picture of a religious man in India, which has some strong saturated saffron highlights, shadow details, and good near-white highlight details. Taken on my Minolta A2 camera as a 5 Megapixel image.
    Original sRGB image: booktst2.jpg
  3. The PhotoDisc test target with the babies on the bottom linked from
    Original AdobeRGB image: booktst3.jpg

I followed the submission instructions that the publishers gave me via email or their websites, making sure to use the color profiles (sRGB or AdobeRGB), and pixel dimensions they requested. I did color profile conversions and scaling in Photoshop CS.

I did visual comparisons with the image on my Sony LCD monitor, calibrated with a Gretag Eye-One Display 2 to gamma 2.2 and 6500k. I evaluated the prints under mixed lighting consisting of a 500W halogen, 50W halogen, and some incandescent lights. I tried various combinations of those lights and it didn't seem to make much difference in my evaluations of the color.

Mypublisher didn't have any means for ordering individual prints, so I ordered a "pocketbook" with 20 pages for $10, 6" x 8" size, actually about 5 7/8" x 7 3/4", full bleed. I used their bookmaker tool to create the book, pointing it at the full resolution images. I converted the photodisc image from AdobeRGB to sRGB since they only accept sRGB. Their semi-glossy paper was the thinnest of the bunch, but just fine for me.

The color was the worst of the bunch; it was too cool. Things that should be white were just a bit cyan. This was most noticeable in the faces of the babies, which ended up looking quite sickly and unacceptable. The photodisc image seems to have more of a cyan cast than the other two images I submitted, perhaps just because there are more neutral whites in the image.

Yet the causasian skin has more rosiness in the mypublisher prints than any of the others. I think they do some kind of processing just like the instant labs that try and make everyone look like they have a glowing tan. The saffron robe of the religious man seems to have blown the gamut slightly, and the saffron highlights are a bit lacking in detail.

I also had 17 other images in the pocketbook to look at, and the worst color issue I saw in the mypublisher shots was the darker greens. I can't really explain what is wrong with them, but my guess is that there is some low level greyscale nonlinearity that causes color shifts in the darker trees and foliage. Also perhaps the inks they use don't quite fall into the center of the spectrum band that our eye's green receptors cover.

It might sound from these descriptions like the color was horrible, but really it was fairly decent, just not great. The color reminds me of what you would see in a quality magazine that had been in storage for say 10 years so that the colors started to change.

The real concern was the shadow details. All the shadows were too dark. On the photodisc image, you could just barely make out detail in the kodak gray scale (lower right test swatch) in the #12 cell, whereas on my monitor I see two more shades before it goes to black. The dark gears in the top left of the photodisc image don't show their details. The pants of the bicyclists are almost completely inky black, and the detail visible in the monitor isnt visible there. It looks like mypublisher has cranked up the contrast control in order to make unadjusted pictures look more "snappy", but at the expense of the shadow details. And they did look more snappy, but in an unprofessional way.

The smoothness and sharpness of the various publishers is difficult to compare since I ended up with different sized images from each. Frankly, that isn't a real concern for me; all of them produced adequate sharpness and smoothness for my standards, just a bit less smooth and sharp than a typical magazine. So I'm not using those factors in my selection process.

In smoothness they are all very close, but mypublisher is slightly more speckly than the other two.

When I emailed sharedink about getting test prints, they enrolled me in a free 45 day trial of their special photographer program, which is not documented in the public portion of their website. As far as I can tell, the program doesn't necessarily give better image quality. It's a way of ordering special covers, foil lettering, and letting you use your logo rather than theirs. The photographer program also allows you to order test prints for $5.95 for the first and $.50 per additional minibook sized image, which is what is what I did. I submitted the three images in their native color profiles (AdobeRGB for the photodisc image and sRGB for the others), 7.875" wide x 6.0" and 300dpi jpeg format. The images that I got back were 7.875" x 5.875", on paper 9"x6", so there were white borders on the test prints. The semi-glossy paper was the heaviest, similar to the thickness you would get from a minilab print.

The color was better than mypublisher. The saturated bicycle jerseys looked more natural, not as blown out as in the mypublisher shots. Whites and the overall color cast was very neutral, matching well with the image on my monitor. Skin was still a bit overly rosy, but not as much as the mypublisher shots.

Sharedink seems to have the same problem with dark details that mypublisher does. The black bicycle shorts were missing detail. On the kodak gray scale from the photodisc image, there was one more square of grey; 13 was the last cell of grey, but it was very close to black. The left part of the Kodak/Heidelburg color checker showed less shadow detail than the sharedink. Sharedink uses an HP Indigo UltraStream printer.

The smoothness of the Sharedink shots was perhaps the best of the publishers, with less speckliness than either of the other publishers. But its halftone pattern also was slightly more apparent, giving a very slight horizontal banding when examined closer than you normally would.

When I emailed fastback, they offered to do three test prints for a very reasonable price, so reasonable that I don't want to share it here for fear that they may get innundated with requests from people who want the same deal. They wanted the images in AdobeRGB, 4"x6", 300dpi, tiff format. So that is what I gave them. What I got back were 3.8" x 5.9" images on 8.5" x 11" paper. Fastbackbooks has an appealing creative feel, very small and just starting out. The people seem friendly and helpful.
The prints from Fastbackbooks were on matte paper, which puts them at a disadvantage to the others who printed on glossy paper. I asked them about glossy, but they confirmed that matte is the only surface they do. Actually it has just a bit of shine; they described it as "almost eggshell". The prints look more dull than the other glossy prints, just because the blacks reflect more light. Frankly this is a big problem for me; I find I prefer the depth that comes from glossy paper.

The color is the best to my eye. Most importantly, skintones look very natural, not overly rosy. The baby skin looks closest to what I see on my calibrated monitor. The white highlights seem neutral or just a very bit warm. The bicycle jerseys are significantly less saturated then the others, looking very natural. Fastbackbooks uses a Xerox Phaser 7750 printer.

And the shadow detail is easily the best of the publishers. Just like on my monitor, I barely make out detail to cell #14 in the Kodak Grey Scale in the photodisc image. The detail in the black bicycle shorts is a bit less than I see on my monitor, but still much better than the other publishers. Shadow detail still is noticeably worse than the monitor image on the religious man's eyes; in the fastbackbooks image (just like the others) the eyes are lost in the darkness whereas they are very visible in the monitor image.

Unfortunately the matte surface of the fastbackbooks prints puts them at a real disadvantage to the other glossy prints, making the prints look somewhat dull. Nevertheless, the matte surface has a kind of arty, wholesome feel.

They took more than a week to return my images, the longest of the bunch. And their headquarters is just an hour's drive from my home, so it wasn't the mail. They said they had their busiest week ever.

Asukabook is a brand new option; they had only been open a few weeks when I checked them out. Asukabook did not allow printing individual test prints, and their books cost several times what the others charge. They are too expensive for me to order one just for testing. They did send me a brochure printed in the same way as their books are printed, and it happenned to have the same photodisc test image that I had been using to compare with the other publishers. They also emailed me the original file that the photodisc test image was printed from, and it looked the same on my monitor as the one I had sent out to the other publishers.

The Asukabook paper was good quality, and the glossiest of any of the publishers, although still much less glossy than glossy chemical or inkjet prints. The shadow detail was decent but not great. The kodak gray scale went to cell 13. The dark gears in the photodisc image lacked details.

The white cell on the grayscale target doesn't look white, it looks grey. There are some highlights that are white, like the specular highlights on the metal canister, but overall the highlights come across as a bit washed out, too grey. This tends to lend an overall dullness to the prints. I'm assuming they made this choice in order to make sure not to lose detail in wedding dresses and other white wedding clothes, but I think they went overboard; all of the other publishers and printers have pure white in the end of the kodak greyscale. The halftone pattern that asukabook uses is slightly bigger and more complex than the other publishers, more similar to what you would see in an actual magazine. It tends to give the impression of diagonal lines when examined up close.

Asukabook's color is okay but not great. The babies had a slight cyan-bluish cast to their skin, roughly halfway as strong as the cyan cast that mypublisher had. And the babies faces looked darker than in all the other prints, probably due to the grey-highlight issue. I noticed some minor false-color (greyscale nonlinearity) on some other faces on the flyer.


Inkjet printers

For comparison purposes, I wanted to see what a good inkjet printer could do. I went to the San Francisco CompUSA store with a stack of Epson glossy photo paper and my test images on a compactflash card. The printers I tried were the HP 8450, Epson R320 and the Canon ip6000d. On each of them I printed two images on a single 8.5x11 piece of paper: the sRGB photodisc image and the bicycle jersey people.

The inkjet printers blew the publishers away. The HP and Epson printers produced outstanding prints: Shadow detail to cell 13-14, nice white highlights, and great color. Detail and smoothness were much better than all the book publishers. The HP had slightly better color than the Epson; the napkins and table matts in the restaurant image look white in the HP print, whereas the Epson made them slightly pink. The Canon print had poor shadow detail (cell 12). Facial features and the napkins were too reddish on the Canon, but it still had great highlights, detail and smoothness. The Epson photo paper I used had much higher gloss than the paper from any of the book publishers.

Since I was testing floor model inkjets at a retail store, where the ink levels are unknown, it's quite possible that the issues I had with the Canon were just due to low ink levels or poor setup, and should not be looked at as reasons to avoid that particular printer or brand.



Shadow Detail
mypublisher 12: Poor Great Middling Semi 3 days
sharedink 13-: Middling Great Good Semi 4 days
fastbackbooks 14-: Great Great Very Good Almost Matte 6 days
asukabook 13: Good Too Grey Barely Good Semi 5 days
HP 8450 inkjet 14-: Great Great Great Glossy instant
Epson R320 inkjet 13+: Good Great Very Good Glossy instant
Canon ip6000d inkjet 12: Poor Great Middling Glossy instant



All of these publishers produced decent output, and anyone who just wants to print up some snapshots for grandma would have no problem with any of them. I am sorry to report that I haven't found the perfect publisher for my needs; they all have different strengths and weaknesses. Fastbackbooks would be my first choice if they used glossy paper, but the almost-matte paper they use seems a bit dull. Sharedink is a strong contender, but their shadow detail is not so good. I wonder if I could get good results from them if I lightened my photo files before sending to Sharedink. High priced Asukabook was a bit of a disappointment, but they are very new as of this writing and maybe they will improve over time. Inkjet prints are clearly superior to the book publishers, and the HP 8450 printer lived up to its reputation for producing perfect color right out of the box. The best image quality would clearly be obtained by printing your own inkjet pages and having them bound.