Upload & Sell: Off
| p.1 #8 · Question on 1:1 Print Size |
Mr Joe wrote:
Viewing at 50% on the monitor is generally a good simulation of a print. Viewing at 100-200-300% will help ou assess your sharpening. Adobe's Jeff Schewe debunks the printer resolution math as being outdated voodoo in recent LuLa videos. He recommends setting prints at 360 in LR if the native resolution is under 360, and at 720 if the native resolution is over 360. I did some controlled testing at multiple sizes when printing 10 images for a gallery show recently and I'm glad I followed his advice.
On what factual basis does he support that claim? Is it an understanding of how the printers and objective tests including resolution targets (required if the objective is comparing resolution) or his subjective impression of what looks good in photo he prints on his Epson printer? (Epson printers have native resolution of 360).
The significance of native resolution is that where ink or photo diode meets paper the printer's RIP (raster image processor) must translate discrete RGB pixel values to the resolution of the print head. If the file dimensions are random relative to native resolution the results of how to the printer RIP does that will be more random. It's more a matter consistency in the results obtained image to image. Using 360 or 720 would be good advice for Epson printers but not perhaps for HP which have 300 / 600 native resolution. Also what works best for ink jets isn't best for photo paper based printers.
Ink jet printers use stochastic (random) screening and frequency modulation (how fast the ink gets spit out) to control the laying down of ink. Photo paper printers such as Noristu found at places like Costco use photodiodes to image the paper on a 1:1 basis hence the advice on the Dry Creek site to resize to printer native resolution.
Because a print, one way or the other, is created with discrete applications of CYMK inks or dyes there is an inherent conflict between what is ideal for rendering smooth flat tones without random variation (i.e. noise) and rendering fine detail. While it might seem that more resolution = more detail that's not the case perceptually. It's similar to music where silence is needed between the notes to make them discretely heard. That's why USM is needed / used. It artificially increases contrast at tonal boundaries which our brains perceive as sharpness. It's an optical illusion, but then so it seeing 3D on a 2D piece of paper.
The amount of USM needed for an image will vary with size and content. Enlarging a screen image to 200% will make it easier to see on the screen what how the application of USM affected the discrete RGB pixels on the screen, but it will not predict how much USM is optimal for a particular image when printed for several reasons. The most significant is that USM is needed to compensate for mechanical variables of the printing process. I know a bit about that aspect because I managed process control at printing plants most of my career. Secondly, the amount of USM which is optimal for an image varies with print size and viewing distance.
The driving factor is our eyes ability to resolve detail and how our brains discern 3D shape in 2D images. At reading distance high resolution of fine detail and texture s important because it can be seen at reading distance. But as prints get bigger, such as on a billboard, the brain shifts gears and gets its clues about shape from contrast differences.
One way to visualize and wrap your head around this is to think of image content not in terms of discrete pixels but how they would be rendered as a wire frame as in computer generated art. The smaller the planes are in the wire frame rendering the smoother the rendering looks. But if you look at a courser wireframe from further away it will look similar and often better in terms of discerning 3D shape because of the way the brain recognizes shape via tonal contrast of plane surfaces. This is something poster artists working with wood block or silk screen have long understood and exploited.
The dilemma with digital and printing is you can't make a really big print that looks optimal both from across the room and at reading distance. What looks great at reading distance will term to mush at greater distance not as a result of how well the printer prints or its resolution, but because of the limits of the eye's resolution.
The images which stand up best to extreme enlargement are those like a sail boat on the water where the viewer seeing the shape of the white sail and red hull immediately understand the content based on having seen boats in person. Images which don't stand up well to enlargement are those where detail is necessary to see / appreciate the content, like a fall foliage scene which subtle variation in color between leaves or dunes in a desert: viewed from a distance appearing as orange / brown blob.
So while it is possible to make generalizations getting optimal results in prints is a bespoke process in which the reproduction parameters need to be tailored to fit the content of the image and the distance at which it will be viewed (perceptual variables). You need to first understand how the printing process and the perceptual variable change your impression of the image between seeing it at reading distance on screen, and from across the room when hanging on the wall as a 24 x 30 print. The only way to tell with certainty which techniques will produce the optimal results is to try several different ones on prints of a section of an image, put them on the wall and look at them. If on the other hand you are making prints which will be seen at reading distance, that's the distance at which you'd want to evaluate. What will be the optimum printer resolution? The one that looks best at that distance
The history behind the 300ppi rule of thumb dates back to when the first laser printers were being created in the late 80s. B&W toner images like offset printing produce gray tones by changing dot sizes on a fixed DPI linear dot density. DPI = offset dots per inch and was in use long before digital reproduction was invented.
Each halftone dot on an early laser printer was constructed of an 8 x 8 gird pattern of 64 sub-dots which when hit with a laser beam would produce from 0 (white paper) to 64 (all filled in) gradations of tone. Creating a 150 DPI image on the laser printer required an imaging head with 8 x 150 = 1200 imaging cells per inch density.
Below 150 hard dots per linear inch (DPI) was long the standard for B&W offset (discrete dot) reproduction on coated papers because when the pattern is courser than 150 DPI the eye can see the dots at reading distance. Less resolution was possible for newsprint because uncoated paper absorbs and spreads the ink like a blotter, blurring the discrete dots. The relevance? The technology is driven by what is necessary to fool the eye into thinking the image was continuous tone.
When making halftones from continuous tone prints, which I did for a living at National Geographic for several years in the 70s, we used a screen with a ruled pattern. Digital images were already broken into pixels. Researchers experimented with various pixel > 150DPI printing dot sampling rates and found that between a 1.6 and 2.0 oversampling ratio of pixel to "hard" dots was needed to produce an acceptable image perceptually. 150 DPI x 1.6 = 240ppi and 150 DPI x 2 = 300ppi. Hence the conventional wisdom to use 240 - 300 ppi values for printing for offset, which is how most commercial photography was reproduced.
So that that rule of thumb applies to B&W offset and laser printing. It didn't apply to color printing. One of the things I learned managing offset printing process control is that with four different overlapping ink patterns in play less screen resolution is needed to full the eye into thinking a hard dot image is continuous tone like a photo print created from a negative — the standard against which printer resolution was compared. Where as 150 DPI was needed to make B&W with one set of dots look continuous tone, 133 DPI worked just fine for color images where four dots overlapped because in most area there were always 2-3 overlapping dots.
Ink jet printer resolution is similar in how it overlaps dots. The print head may be 360 jets per inch in density but the printer can move the paper under the scanning head in increments of 1/720th or 1/1440th inch with the stepper motor, overlapping the dots and laying down a more continuous tone pattern. What the addition of light magenta and light cyan does is allow the the printer to lay down a 100% "dot" of light cyan vs a 50% dot of the darker shade. The net effect is the same tone perceptually without any visible dot pattern.
Ink jets now spray ink on paper more like an air brush than a printing press, so the old rule of thumb doesn't apply. The print head has a physical number of imaging jets space 1/300 or 1/360th inch apart center-to-center but in the direction of paper travel through the printer the resolution can be increased by advancing the paper by a smaller increment between print head passes. If for example the native resolution of the print head is 360 and the paper move 1/720th inch between passes the ink pattern laid down by the head will overlap by 50% and lay down a more continuous layer of ink. That may not in all cases be the optimal way to reproduce fine detail in a reading distance image, but would work well for a larger one where the viewer is interpreting 3D shape form tonal variation and not resolving the fine detail. Leonardo didn't use a Wagner power painter to produce the Mona Lisa. But he could have if Mona was on a billboard on top of building in Times Square. Dan Margulis discusses how more resolution isn't always better in his "Professional Photoshop" book, devoting an entire chapter (15) to the subject.