4x6 Prints from Digital Images:
Comparison of Some Ottawa Photofinishers

2006 October

Here's the story: I had about 100 digital images to print and was new to digital image printing; who should I get to print them? I gathered some test images from web sites and tried a few local photofinishers. I made this web page to share what I did and what I learned.

(This page was last updated in 2006. It might still be useful for the methods used, and to highlight how quality can vary among photofinishers.)

SUMMARY (4x6 prints)

Best value: PhotoLab (Richmond Rd) Good quality and good price ($0.19)
WalMart (South Keys) Good quality and good price ($0.19)

Good choices: FutureShop (South Keys) Good quality but higher price ($0.29)
Ginn Good quality but higher price ($0.39)

Also: Black's ($0.39), and Throop ($0.49)

(Prices fluctuate; check their web sites for current prices)

Not tested but reportedly worth trying: Costco (Nepean) ($0.18)

(The background colour of the box above should look 'beige' or 'pale yellow'. If it looks nearly white, you're probably using an LCD not able to display pale colours.)

Background

All photofinishers tested here use light-sensitive photographic paper to make their prints, just as photofinishers have been doing for decades; but now, instead of shining light through a negative onto the photographic paper, the light is generated by a computer according to information in the digital image you give them. Thus the prints produced by this method, on photographic paper, should last as long as traditional prints produced from film. (Home photo printers typically use inkjet, dye sublimation, or laser printing technology.)

The quality of a print is influenced by the quality of the photographic paper and the photofinishing (which may include digital image processing, such as sharpening).

Some web sites with information about photofinishing and digital printing:

Method

Glossy 4x6 prints from several Ottawa photofinishing labs were compared using a test image (shown below) based on a PhotoDisc target image (as modified by Dry Creek), plus some colour bars I generated using software, and other components described below. To download this test image, click here to view a full-size version (about 2Mb) and then save it to your computer (by right-clicking on the full-size image).

 

The test image above has these components:

Cropping indicators Perfect alignment of their printer with the edges of the paper is too difficult to maintain, so instead photofinishers project your image so that it slops over the edges slightly. So even if you provide correctly-proportioned images, you'll lose a small amount from all edges of your image. The red wedge-shaped figures at the four edges of the test image above will measure how much is cut off ('cropped') by the photofinisher. The center line of each arrow has tick marks every 10 pixels, and because the angled lines are 45 degrees, the distance between the angle lines and the center line is equal to the amount cropped. Lab people I spoke with say that the amount cropped depends on the particular machine's settings, and should be stable over long periods. Cropping information gives you a rough idea what to expect, and if it matters, an idea about how much to pad your images (eg., using Photoshop). Remember to scale correctly -- this test image is 3600 pixels wide, so if 10 pixels were cut off when this image was printed, that will be 10/3600 * [pixel width of your image] for yours.

Skew detector The thin black line at the outer edge of the image makes it easy to notice whether the image is printed squarely with respect to the paper, or is skewed (tilted).

PhotoDisc image The bulk of the image is the PhotoDisc Inc test target. The four faces should appear natural, and details of the image should appear as in the original target.

Colour space target The target in the top-right corner detects problems with colour production. Problems show up as ripples in the colours, or shapes that are different from the original pattern.

Colour bars The bars at the upper-left are generated by software. The pixel values for each square increment by 8, ranging from 7 to 255 across the 32 squares (a pixel can have one of 256 possible values). Ideally, the right-most square should be white, the left-most square nearly black, and all other squares should be distinguishable. If squares at either end are indistinguishable (or identical to the solid bar beneath them), it means all colour values within the indistinguishable range have been rendered identically, reducing the range of colours that printer can represent.

The image file is 3600 x 2400 pixels, which is the correct ratio for a 4x6 print.

The test image also tests your computer's display! How does it look? Is your display able to show you all the graduations in the colour bars (left side of the target above)? Probably it is able to do a better job than prints, which typically cannot achieve the same dynamic range.

Results (alphabetical, by paper type)

 Crop loss (pixels)
(short red bar is best)
PhotofinisherPaperPriceLeft/Right Top/BottomTimeComments
 A Black'sFuji Crystal Archive$0.39 Same dayEfficient uploader
BPhotoLabFuji Crystal Archive$0.19 Same dayEfficient uploader
CWal-MartKodak$0.19 Same dayEfficient uploader
D Future ShopKodak Royal$0.29 Same dayWell organized, good uploader
EGinnKodak Royal$0.39 3 daysKnowledgable, helpful staff
FThroopKodak Royal$0.49 1 day

(Rest your cursor on a red bar to see the actual pixel count cropped, of 3600 pixels wide.)

Below are thumbnails of scans produced by an Epson 3490 Photo scanner of prints from each of the surveyed photofinishers. The 600 dpi scans are as provided by the scanner, without colour adjustment or sharpening. Dark paper behind each print provides contrast so that you can see the edges. Click on any of the images to see a larger version.

Images from the scanner were not rotated in post-processing to avoid possibly introducing artifacts; what you see is straight from the scanner.

(A) Black's (Rideau Centre)
(B) Photolab (Real Canadian Superstore,
Richmond Rd)
(C) Wal-Mart (South Keys)
(D) FutureShop (South Keys)
(E) Ginn Photographic
(F) Throop Photographic

Analysis

Side-by-side comparisons make differences more apparent and can be useful even if your monitor is not colour-calibrated. In this analysis section, fragments of the test image from each photofinisher are compared side-by-side. The label A to F identifies the photofinisher, as defined above. Dust specks are obvious on some of these scans; they are on the scanner glass or on the print; they aren't photofinishing flaws.

The results have been grouped by paper type. A and B (the top two) are Fuji Crystal Archive paper, C is Kodak paper, and D, E, and F are Kodak Royal paper.

The digital original is included in many of the image sets below, but remember that no print is capable of producing the range of colours and brightness in a digital image; print technology cannot produce the sharpness, brightness, and colour range of computer displays.

Colour bars Yellow dots mark the approximate range that the print was able to produce. The wider the range, the better, because levels outside the range are all rendered identically. A and B are Fuji paper; note the limited range in green and blue; we can expect Fuji prints to have exagerated greens and blues (because low levels of those colours will be rendered 'brighter-than-real', according to the results here). C, D, and E have a good range.

Similar information is provided by the synthetic gradients (which contain colours that are 'out of the gamet' -- out of the range -- that printers are capable of rendering). The digital original is at the bottom. Note also the differences in sharpness of the text; B, C, and D are sharp, while F and E are fuzzy.

Skin tones Faint letters A-F at the left of each row of faces indicate the photofinisher. The bottom image is the digital original. F has a yellow colour cast; A has a slight red cast; C may have a slight magenta cast. B has perhaps the best match with the original.

Colour spaceThe bottom image is the digital original.

Skin tones, Sharpness Below, colour casts (F, A, C) are evident when comparing the results to the digital original (at the bottom). Note also how sharply the kernels are rendered on the cob of corn (B is good, F is poor).

Sharpness Below, sharpness differences are evident in the rendering of the electronic circult board (the digital original is at the bottom). This set of images was post-processed (using Photoshop) to bring up the levels. B is good.

(If the digital original is below the result images, widening your browser window may bring it up to the right.)

Sharpness, Alignment Below, sharpness differences are evident in the rendering of text. The colour casts might be due to misalignment of colour generators, causing a colour fringe that gives a colour cast (C and F).

Brights Below, bright sections, eg., the light filaments, are rendered differently (C and E, perhaps B, are closest to the original). Note also sharpness differences. The digital original is on the right. This set of images was post-processed to shift down the levels. F lacks sharpness, A seems 'blown out'.

Miscellaneous observations When viewed with a hand lens, a grid-like half-tone pattern was observed in print C (Wal-Mart). It isn't visible when viewed normally.

All prints were well-aligned with the paper; this year, none were skewed.

None of the prints were manually adjusted (according to my interpretation of the processing information printed on the back of the prints).

In the 2005 results, it seemed that some photofinishers applied sharpening algorithms, which improved the images. Artifacts at the boundaries of colour bar cells may indicate sharpening was applied by some photofinishers this year too.

'Kodak Royal' paper seems thicker than the 'Fuji Crystal Archive' or 'Kodak' papers.

Discussion

Despite some minor flaws when examined closely, prints from all the photofinishers were good; none were so far off as to be objectionable. Photofinishers are offering great improvements in value over years ago.

Prices ranged by 250% (from $0.19 to $0.49), but quality was much less variable. Some of the higher-priced prints had significant flaws. With quality similar, it makes sense to shop by price.

Many people prefer images that are 'better than real', and would prefer such prints over ones that more accurately reproduce the original. For example, exaggerated greens and blues might make an outdoor scene look better. (However, if you already adjusted a digital image to exaggerate the greens and blues, you might not want the photofinishing to exaggerate them further.)

All have good uploaders this year (providing the ability to send photos via the internet). I particularly liked the uploaders used by FutureShop, Wal-Mart, and Black's. Most use an ActiveX object (with an option to use a less-functional interface without activeX); Ginn and Throops have an application ('fotoDesk') that is downloaded and run in your computer. All had good follow-up by email (order confirmation, notification of order ready). None but Wal-Mart seem to make use of my email address for subsequent advertising (Wal-Mart sends perhaps-monthly email about promotions, price changes, etc).

Ginn's staff seem well-trained and informed; they are helpful and knowledgeable. Ginn's staff is willing to help customers get up to speed with digital processing. FutureShop also has helpful staff on-site. Throop Photographic is a traditional full-service photography store, like Ginn Photographic.

In previous years, some photofinishers produced their prints at central sites, but this year, it appears that all of them used on-site printers -- allowing most of them to have orders ready within an hour or two. Quality doesn't suffer; there was no apparent correlation between time-to-process and quality.

There are price fluctuations and sales: In the post-christmas period, Wal-Mart lowered prices (to $0.12) for a period; PhotoLab and FutureShop responded by discounting prices.

The higher-priced photofinishers often have volume-discount plans. For example, Ginn offers a card for 100 prints for $29.

Dry Creek (referenced above) offers a profiling service to photofinishers, so that their customers can download ICC profiles. Costco is the only Ottawa photofinisher in their database (click here to see their entries; find Canada, then Ontario, then scroll to Ottawa).

Dry Creek and others recommend Kodak Royal and Fuji Crystal Archive paper. (Kodak also makes a paper labelled just 'Kodak', which apparently is not the same a 'Kodak Royal'.)

Print dimensions: The 4x6 prints were all 4 inches high, but none were 6 inches wide. Most were 5.885 inches (14.95 cm) wide; Wal-Mart and Throop were slightly wider, at 5.930 inches (15.05 cm). (An aside: The actual dimensions of 5x7 prints vary much more; some places produce 5x7.5 inches, the same dimensional ratio as that of a 4x6; check ahead if it's important to you.)

Caveats: Only one trial was done. All prints were processed on-site; results may differ at other retail locations of the same company. Photofinishers may have taken extra care with the obvious test print (though most/all? have highly automated processes). Although none of the prints have indications that they were colour-corrected manually, there may have been some automatic image sharpening by some of the photofinishers (no special instructions were given).

I didn't try Henrys, Japan Camera, Galaxy, or others. If you try them, I'd be interested in what you find. Costco (Nepean) says they have an on-site Noritsu printer using Fuji Crystal Archive paper, for $0.18 per 4x6, one-day service.

It's great to have so many good photofinishing choices at such reasonable prices. It's convenient to be able to upload via the internet, or walk into a shop with your camera's memory card, and get good prints in short order.

Jim Elder
2006 Oct
Comments?