Category Archives: Technique

Making huge prints

Printing

Tiling is the answer.

Background:

Back in 2005 I wrote about making Really Large Prints. That meant 13″ x 19″, a veritable wallet size compared to what I am doing here. We are talking 36″ x 48″ or seven times the area.

On occasion I have regretted buying the 18″ wide HP DesignJet 90 printer, wishing I had instead paid a little more for the 24″ DJ130. But when it came to making a 36″ x 48″ print the other day, neither could have done the trick. At least not in one pass. And making something really huge for my DJ90’s seventh anniversary seemed the right thing to do.

With my DJ in perfect tune after its annual checkup I was all set, comfortable that perfect colors would result following a recent colorimeter profile session.

Why not just delegate the work?

There are so many variables you cannot control in delegating this task that I discounted the idea in seconds. They include:

  • Finding a good print shop.
  • You have someone making minimum wage, charged out at $100/hr, giving tender loving care to your baby, in between beer and pot breaks. Yes, they will have a jumbo printer which makes this on one sheet. So?
  • Getting the colors matched correctly to your calibrated display image – good luck with that.
  • Getting the big print home undamaged. USPS/UPS/Fedex? Ha, ha, ha! They will fold it to get it into your mailbox. Don’t laugh – that has happened to me before.
  • Longevity. How can you believe the claims made by the print shop for the inks and mounting board? I use original HP Vivera inks which have been tested fade-proof for 80 years by Wilhelm Research. As I’ll be 6 feet under, pushing up the daisies – or maybe fertilizing them – in under half that time, that seems long enough. My mounting board is certified, acid free, Bainbridge Alpha 3/8″ foam core, and will probably outlive King Tut. So there.
  • Satisfaction. Sure, it’s a menial task and one you might not want to do daily. But it is satisfying as it gets doing this yourself and doing it to a higher standard than any agent ever will. I despise the saying “If you want a job done well, do it yourself” because it’s asinine in the extreme, but this task makes for an honorable exception. “Printed by the artist”, don’t you know.

Get my point?

Choosing the right image:

I had a large expanse of wall space viewable from the landing on the upper level but had never quite found a picture to fill it. As I started working on my Golden Gate Bridge series one image immediately suggested itself. Taken from the GGB overlook at the site of the old WW2 battlements, the image is simple, surreal and uncluttered. Just the ticket.


Simple, surreal, uncluttered. The ultrawide 20mm lens used adds to the effect.

Tiling application:

Next it was time to shop around for a good tiling app for the Mac. Tiling apps allows a print to be made over any number of abutting sheets of paper, the result being trimmed and joined just so for display. You can do this in Photoshop but it’s a pain. I settled for the inexpensive SplitPrint which had good reviews and indeed proved capable and easy to use. $5.99.

Preparing the file for tiling:

Before you do anything, examine your file square inch by square inch for sensor spots at 1:1 in your photo processing application, looking especially hard in big, continuous tone areas. Nothing will ruin your day more than finding you have missed some blobs once the print(s) are made. Retouching them now will save a lot of heartbreak later. Just about every sensor will accrue dust and oil spots with use, regardless of whether it has a cockamamie ‘sensor cleaner’ built in or not.

Thereafter the process is simple. I exported the image at the original pixel size of the sensor in the Nikon D3x as a TIF file to Photoshop CS5. Then I ‘uprezzed’ the image to the final print size desired, which would be 36″ x 48″, or four sheets of 18″ x 24″ paper through the printer – the largest it will take. Resolution was set to 240dpi, the maximum the DJ30/90/130 series of printers can handle.

Uprez dialog from Photoshop.

The file was then saved to the Desktop as a JPG, highest quality. SplitPrint at this time only accepts JPG files. No matter. The file was some 45MB in size.

Then into SplitPrint and the required settings were made. I created a custom setting for 18″ x 24″ paper in the Print dialog of SplitPrint and told it to spread the image over four panels thus:


The SplitPrint tile settings.

Dry print runs:

Now 18″ x 24″ paper is not cheap and neither are ink supplies, so I first tested SplitPrint making a 16″ x 20″ print on my office monochrome laser printer using four sheets of 8 1/2″ x 11″ plain paper to see if things aligned and that margins were properly handled. You do not want the margins swallowing any content and neither my laser printer nor the DJ90 can make full bleed borderless enlargements. Margins are unavoidable and the inside ones will have to be trimmed off when assembling the four printed panes.

All went well but, ever cautious, I printed the first panel in color to the DJ90 on a test piece of 13″ x 19″ paper to check that everything was as expected and to verify that colors were true.


SplitPrint printer dialog.

This test confirmed good colors with a perfect match to my display thanks to the EyeOne colorimeter, so I loaded up the printer with four sheets of 18″ x 24″ premium glossy, checked that all the cartridges had ink (running out part way through will not make your day ….) and let her rip, telling SplitPrint to print all 4 images after setting it to use HP Glossy paper with bidirectional printing (‘Best’ quality). This takes a while, some 12 minutes a page, so I did the only rational thing possible. I took my assistant, Bert the Border Terrier, for a walk after the first page started coming out.

Tell me all you want about the romantic days of darkrooms, red lights, blackouts and poison chemicals, for me nothing beats the thrill of seeing a full color image emerging from the DesignJet in broad daylight.


The first pane emerges.

The final prints:

Here’s the finished article, with Bert the Border Terrier for size. I will mount these on acid free board, after trimming the borders (SplitPrint permits small overlaps and draws the cut lines on the images though I had no need of these) then attach all four to the wall, ready for a faux wooden frame. SplitPrint did a perfect job of aligning the images on each page. If it’s of interest I will paint the wooden frame in the colors of the Golden Gate Bridge itself! The paint codes appear below.

Bert and the untrimmed print(s).

Trimming and mounting:

Once the print borders are trimmed (carefully!), the prints are tacked to the mounting boards with the matching sides at the edges of the board. I do not own a 24” trimmer so have to remove the borders with a straight edge and Stanley knife. This is by far the riskiest part of the whole operation, rife with opportunities for error and injury. As the boards are very accurately cut this approach confirms that everything has been trimmed squarely. It’s either right or it looks schlocky. You can opt to leave the margins in place and install these as discrete panels on the wall but I find that sort of thing has been done to death and avoid it:


Tacked prints and boards assembled for dry run to check fit. 26″ bike wheel for reference.

Then it’s off to the dry mounting press to permanently heat mount the prints. Once that is done much of the risk of damage to the fragile prints fades.


In the dry mounting press at 200F. Border Terrier at 102F.

It’s at this point that you start thinking that Michelangelo had it easy daubing that ceiling.

The prints are dry mounted and ready for wall installation:

Ready for hanging.

I use glossy paper. Sure, it’s a pain to work with, fragile as heck and demands very high standards of cleanliness for it shows every blemish, but the result is unrivaled among paper prints for detail, resolution and punch.

Golden Gate bridge paint color codes:

Here are the color codes for those of you into the most beautiful bridge in the world; you can have the shade mixed at your local hardware store to paint any wooden frame you decide to install:


Color codes for the paint used in the Golden Gate Bridge. Click the image for details.

Speed:

There really are no shortcuts here. It’s hard manual labor. SplitPrint makes generating the four panes trivial but after that it’s an exercise in concentration and attention to detail. The next time I do this I doubt I’ll shave more than a few minutes off the total production time, which is several hours. Trimming and wall mounting are by far the most time-consuming steps.

How does it look:

In a word, stunning. While the viewer can get no closer than 10 feet to the print from the second floor landing, even at 2 feet resolution and detail rendering are to die for.

Gear:

Camera? Nikon D3x with the 24mp sensor. But, frankly, a 12mp D700 or any number of similar sensors (FF or APS-C) would be just fine for this setting, though the image would show lower resolution at the same ‘nose in print’ distance. There is more undiluted swill written about the need for high pixel count sensors than you could possibly imagine.

The lens? No magic sauce here either. An old 20mm Ai-S 20mm f/3.5 MF Nikkor – that’s the tiny one which fits in a jeans pocket, handheld at f/8. Mine is 1982 vintage and ran me $215. I added a CPU for $29. You spent how much on that plastic AF-S zoom?


Installed. Wooden frame to come. Prints below are 18″ x 24″.
Cotton gloves on the window sill are used to handle the prints.

HP DesignJet annual checkup

Printing

A little goes a long way.

My HP DesignJet 90, the 18″ carriage model, was commissioned March 14, 2006, so it’s approaching seven years in age. One recent print with a dark black silhouette showed less than perfect blacks and deep, lustrous blacks are one of the many strengths of this excellent printer. Amazingly, B&H still lists the 24″ DJ130 for under $1300 new, and all ink cartridges and printheads remain available on their site, though you may have to hunt about a bit for the special swellable paper which absorbs the ink dyes used by the machine. Regular modern pigment ink papers do not work.

The HP DJ30/90/130 series is blessed with truly outstanding diagnostics and a quick checkout was all it took to find the cause.

First, I dialed up the HP Maintenance Utility which uses online software at HP. For Mac users you have to use OS Snow Leopard or earlier or an even older PPC machine, as HP never updated the software to run on Intel machines. Snow Leopard comes with Rosetta, the PPC emulator software and will run the HP Maintenance Utility fine. Apple recently re-released Snow Leopard on DVD and if you want to run the HP Maintenance Utility on a modern Mac it’s your best choice, though whether it’s even installable on the latest Macs I rather doubt.

I use a decade old iMac which runs Tiger and uses a PPC CPU. Unlike its modern day descendants it does not overheat and refuses to die.

You can run all these print jobs using plain paper in your printer – 8.5″ x 11″. Here’s the Image Quality Diagnostic – this one can unfortunately be run only by using the online software and is the best for determining if a printhead is failing:

No real issues are disclosed here but printhead alignment is called for, judging my some of the colored squares in the center section. The full interpretive section for the above appears here.

Then I ran a printhead alignment which can either be done using the online utility or using the button presses illustrated here:

This did disclose a problem with the black printhead:

The large ‘X’ mark above testifies to a worn or blocked head.

Before deciding on cleaning or replacement of the head I ran the ‘Information Pages’ printout (see above) and got this:

Gaack! The black printhead is 2,534 days old, meaning 6.9 years. It’s the one which came with the printer when I bought it new in 2006! So rather than trying to clean it, I simply replaced it. Pigs get slaughtered.

After replacement of the black printhead – the new one has been on my shelf for ages and is already out of warranty! Some users claim that heads over 30 months old will not work but obviously my experience does not bear this out.

When a printhead is replaced in these machines, they automatically run a printhead alignment which takes some 10 minutes and requires one sheet of plain paper.

Here’s what I got:

All is well.

Finally, out of curiosity, I ran the ‘Paper Usage’ report:

Some advice on older DesignJet printers:

Would I buy one of these used? Only if I could see the Usage Reports and Diagnostics shown above. Many were used by printshops which have beaten the heck out of them. New heads – there are six – run $35 each and cartridges cost a similar amount, so all new heads and supplies total $420. Add $35 for feed tubes. No bargain. Further, if the printer has been unplugged for any period of time, reckon on changing the clogged feed tubes as I had to do when mine went into storage for a few months when I moved a few years back. I explain how to do that here.

Bottom line? I would not pay more than $200-400 for a lightly used HP DJ90/130 (18″/24″) printer, anticipating that some parts will have to be replaced.

Spare parts:

I get mine from Spare Parts Warehouse. The ink feed tube assembly runs $35 and is easy to replace.

Would I buy a new HP printer? Hell NO. I would not buy anything from America’s worst run business whose customer service is a joke. Buy an Epson. The 24″ model runs $3,000 but they will fix it for you when it breaks.

Result:

Success. Perfect blacks were restored.

Here’s the print which was giving me problems:

In the extract, below, you can see a tear sheet of the old print, before the repair, superimposed on the new – night and day:

Printhead failure and analysis:

I have illustrated this before but it bears repeating. Right after the annual checkup, above, the DJ started printing everything with a green cast. This indicates printhead failure as ink levels were fine.

Here is the analysis chart:

I ran the diagnostic report using the online HP utility and this is what I got:



Diagnostic report showing printhead failure.

Comparing with the above chart, you can see that the color patches at A1 (should be magenta), A2 (should be purple) and B2 (should be red) are faulty.

The chart states that A1=M, A2-C+M and B2=Y+M. Note also that the central patch in the left middle section is wrong – it should be magenta. M (magenta) is the common factor to all four error conditions, so I concluded that the Magenta printhead was faulty. $35 to B&H later and it was replaced (a 30 second task) – do insert plain paper when doing this as a printhead alignment chart will be automatically printed when a head is replaced, and the printhead alignment will be performed automatically. It takes some 10 minutes, so be patient. Sure enough, re-running the diagnostic report showed all is well and the DesignJet is back to perfect operating condition.

Pixel peeping fallacies

Lightroom, Nikon bodies, Technique

Know what you are looking at.

When I migrated from the 12mp Nikon D700 to the 24mp D3x, I did a bunch of thinking about the justification for more pixels.

If you do not propose increasing your print size or cropping more severely, more pixels will likely not serve you well. I contemplate making both larger prints and cropping more when needed. Thus, the higher pixel count sensor makes sense for my contemplated needs.

When I first uploaded D3x images from the D3x to Lightroom, I naturally previewed images at 1:1 and remember thinking “What’s the big deal? This does not look any better than the files from my D700 at 1:1.”

The problem, of course, is that I was not comparing like with like.

Here’s a simple table to illustrate the issue.

I have compiled data for four common Nikon sensors – the math is brand-independent, it’s just that I know these bodies and have RAW images from all. I enlarged these original images using the 1:1 preview function in LR4 and measured the image width on my 21″ Dell 2209WA (1650 x 1080) display. So in the table above, using the D2x as an example, the 12.2MP sensor delivers an image which, if printed 1:1, would be 47″ wide.

What does Adobe’s Lightroom mean by 1:1? It means that images displayed 1:1 are displayed at 90 pixels/inch – you can confirm this by dividing the ‘Sensor – W’, the pixel count across the width of the sensor, by the ‘Width at 1:1 in inches’ and in each case you will get 90 dots per inch. That’s good for an LCD display or for prints looked at from a reasonable distance. If you want to stick your nose in the print, then you want to limit the pixel density to 240 pixels/inch, which is the same as dividing the above ‘Width at 1:1 in inches’ data by 2.7. So a 240 pixels/inch print from the D800’s sensor, for example, would be 31″ wide (83/2.7). But in practice, you do not need that high a density in huge prints.

As you can see, comparing a D700 image with, say, a D800 image, is not fair if identical 1:1 preview ratios are used. You are comparing a 46″ wide image with one almost twice as large at 83″. To make the sensor comparison fair, you need to preview the D800 image not at 1:1 but at 1:2. That will yield approximately the same reproduced image size, making for an objective comparison of resolution and noise if the same lens and technique are used for both.


Preview options in Lightroom.

Yet, I suspect, many snappers fall afoul of these erroneous 1:1 comparisons concluding:

  • I need better lenses with the newer body
  • My images are blurred, I need to use faster shutter speeds
  • My focus is out, there’s something wrong with the camera

All of the above lead to much time and money wasted in fixing the unfixable. Bad data.

It is indeed quite likely that your new sensor out-resolves the limits of your older lenses at 1:1. It’s also reasonable to expect motion blur to be more visible at the same shutter speeds if you use faulty comparisons. And the chances are it’s your technique not your hardware which accounts for poor focusing, the errors only becoming visible at double your former preview magnifications. But, unless you contemplate making crops to one quarter of the area of your previous sensors or making prints 7 feet wide instead of 4 feet wide, your sensor upgrade is only causing you needless pain.

My first conclusion with the D3x compared to its D700 predecessor was all of the above, until I figured out what I was looking at. Some comparisons are easily drawn. It’s clear for example, that the D700 has lower noise than the D2x for the same image size, hardly surprising as we are comparing a recent FF sensor with an older APS-C (D2x) one. The total pixels and 1:1 print sizes are almost identical. On the other hand, comparing the D700 at 1:1 with the D800 at 1:2, for like print sizes, shows little difference. It’s only when you double preview sizes with the D700 to 2:1 and the D800 to 1:1 that you see the greatly superior resolving power of the D800, as the number of pixels you are looking at in such a comparison is tripled in the case of the newer sensor.

Nikon has not helped the situation. After their affordable high pixel count FF bodies – the D600 and D800 – came to market, they started publishing pieces intimating that only their very costliest and newest lenses were ‘good enough’ to extract the best from the new sensors. The rest of the sheep writing purportedly critical analysis followed right along. It’s called sales and makes little sense. Some of Nikon’s highest resolving power lenses were made ages ago, long before digital sensors existed – any Micro-Nikkor macro lens pretty much qualifies (55, 105 and 200mm) – as do a host of pre-Ai lenses, many over four decades old. If you like the latest and greatest (and costliest) have at it. But don’t believe everything you read from such conflicted sources. Their primary focus is not on your image making capabilities but on your wallet, be it through sales (Nikon) or click-throughs (the whores who parrot this stuff as if it was technically proved fact).

So before you chuck out your old lenses and start buying costly superspeed exotics which allow the use of faster shutter speeds, while contemplating return of the body to Nikon for repair of focusing errors, ask yourself what you are really looking at when you preview those enlarged images on your display.

Practical implications: It’s not like you can avoid buying new gear with lots of megapixels by trying to save money on something with fewer. Everything has lots of pixels today. 12MP is hard to find at the lower limit. But the practical implication of this rapid technological advance is that, for those on a budget, substantial savings can result from buying the previous generation of hardware, comfortable in the knowledge that while 8-12MP may not be a lot, it’s more than enough for 99% of needs. DSLR bodies like the Canon 5D, Canon 5D MkII, Nikon D700, Nikon D2x, Nikon D3 and others no less capable from Pentax and Sony offer tremendous savings just because they have been replaced with something that measures better in a comparison table. Heck, a lightly used 6mp Nikon D1x can be had for under $250 and will offer tremendous capability, outfitted with a $50 mint MF Nikkor, far in excess of the abilities of most. The barrier to entry to good hardware has never been lower. 16″ x 20″ prints? No problem. Why do I say that? The D1x’s sensor is 3,008 pixels wide, so for a 90 pixel/inch print (what Lightroom shows at 1:1 preview) you would get a print sized 33″ x 22″. Unless you stick your nose in it, it will show just fine.


Nikon D1x. Add Nikkor of choice.

AeriCam

Technique

A fascinating device.

Waiting at the traffic lights in San Francisco the other day, my eye was caught by this plaque so I quickly made a snap.

It turns out that AeriCam manufactures a state-of-the-art miniature helicopter which permits a movie camera or DSLR to be attached for aerial photography. It costs $12,000 ready to fly. They also manufacture a gimbal mount for an additional $2-3,000 which helps minimize vibration and the results are quite extraordinary in their smoothness and professional quality.

Results from their six bladed ‘hexacopter’ can be seen by clicking the image below.


San Francisco from the Bay Bridge. Click the image for the video.

If you can overlook the extraordinary car skills displayed by the driver, focus on the smoothness of the aerial shots. It’s not quite clear to me how one gets a video feed from the camera to the operator to permit accurate composition, but at $15,000 this device appears to be far cheaper than helicopter rental, and is available in moments to any cinematographer. The rather sparse web site refers to a ‘video transmitter’ so I assume that beams a live picture back to earth to the included ‘9″ SonyMonitor’. Bad weather? Put it back in the box and come back another day.

The recommended payload is 10lbs which will easily accommodate a big DSLR and lens. The device is GPS capable so programming locations should be easy. Order lead time is 6-8 weeks. American ingenuity at its best.

Snapseed for OS X improved

Photographs, Technique

Welcome enhancements.

Snapseed is a simple and effective application for both iOS and OSX for conferring special effects on pictures. I mostly use it to add grunge in those snaps which merit the treatment.

Now the two major issues I had with the OS X version of the application have been fixed by Nik Software. First, you can export to Snapseed directly from Lightroom and second the application will now accept TIFF files rather than just JPGs. I have tested both and can confirm the revised version does both perfectly, and when you save the file it’s stacked with the original in Lightroom.

Here’s a snap processed in Snapseed with grunge, vintage effect and frame added:

On Broadway, SF. D700, 105mm f/2.5 Nikkor-P at f/11.