Luminous Landscape Forum
Site & Board Matters => About This Site => Topic started by: amolitor on February 22, 2018, 10:26:57 am
-
I confess that I did not perform a DEEP read of this article, but it did not strike me as a particularly new thing at all?
People have been shining colored lights onto silver halide paper to make digital prints for a long time. I don't quite get how Lumejet is any
different from, say, a Durst Theta printer, which has been around for quite a while.
What am I missing?
-
....................
What am I missing?
The first half of the essay which describes their re-engineering of a long-standing process into one for which they claim superior outcome characteristics, and both Kevin and I determined does indeed produce very convincing prints.
-
I have a more detailed video and information coming in an article I am working on in the next couple of weeks.
-
With all due respect, Mark, if there's anything in the essay that acknowledges that previous versions of this kind of process exist, it is quite small. The first half reads exactly as if the Lumejet guys worked away in their lonely engineering garret for 15 years, and then came out with this, the first and only example of this amazing new technology.
The word "Durst" does not appear in the article, according to my search box (I *did* take a few pro forma steps to check for obvious things I might have missed).
Your comparison is with an inkjet printer, not with, say, a Durst printer, or with DSI.
Overall, the impression one gets is that this is a sui generis process, with no antecedents. This probably is not what you intended, and perhaps the antecedents are so strong and obvious to you that you cannot imagine how anyone could read it differently. And perhaps the LuLa readers are sophisticated enough that this is true.
-
and both Kevin and I determined does indeed produce very convincing prints.
As does a Lightjet, Durst Lamda and even a Chromira. but it's still just a C print exposed with LED light with all the longevity and gamut challenges of that process.
so when you say convincing, are you saying they look good (as do well made prints form those other devices) , or are you saying wow, this is really cool and new and is visually obviously superior to those.
-
Hi Wayne,
For avoidance of doubt, my review by intent and design does not focus on any other chromogenic process. My discussion has two basic components: (1) a description of their process based on what they say it does just so people know what it's about, and (2) a comparison of their prints with the prints that come out of my Epson SC-P5000; why - because I was interested in seeing, and thought readers would be likewise - how the output from this process compares with output from one of the finest inkjet printers on the market - i.e. comparing two very different and capable technologies. I also think such a comparison could be of interest to all those who want prints but don't want to print, so would send their files out for processing say to a good quality C-print lab, or have their photos made into C-print books which these folks also do a lovely job of.
So when I say their prints are "convincing", what I mean is that if you look at them in isolation they really do like fine, and if you look at them compared with output from a first-rate inkjet printer, they still look fine. This is not only my opinion by the way - when I produce these comparisons I invite other pairs of experienced eyes to look-see before I submit for publication. Now of course, a sample of prints isn't a whole universe of prints, but a purpose-driven sample can tell a lot.
Comparing their chromogenic process with other chromogenic processes is also a valid exercise, but a whole other talk-show and fresh piece of quite time-consuming research - which I would welcome others to undertake if they're interested. :-)
I agree with you that the gamut looks relatively challenging in theory when you examine their profiles, as I showed in the article, but in practice it turned out to be less of a big deal than I expected; but because of the gamut difference, different editing under softproof with bespoke profiles is needed between say a Lumejet output and an Epson SC-P5000 output on luster paper; this is pretty standard when trying to produce approximately similar image appearance from technologies or materials that have very different gamut or contrast ratios. Now maybe a different set of prints would tell a different story, but between Kevin and I, we did have them print quite a few different kinds of photos, as well as some standard evaluation targets, and they all looked very good. Longevity isn't my wheelhouse, but I think the general consensus is that they won't endure as well as pigment inkjet.
-
Thanks for the perspective.
I guess as one who has owned and operated LED based silver halide printers for over 20 years (my first one was the Kodak "Pegasus" LED printer which produced pretty remarkable results), I just don't see this as a new process. I think those that have never seen many digital C prints are surprised that the "limited" gamut and "meager" 300 dpi resolution can result in very sharp prints with great saturation, smooth gradations, and visually very competitive with high end inkjet prints.
So my curiosity is piqued ... have they really managed to improve the digital c print over previous technologies. Are they going to manufacture and sell the device to other labs? (seems that's would be their main goal, to market the technology). If you and Kevin can figure out how to test that it would be very enlightening.
-
FWIW Wayne, I agree with you 100%; nothing new about this print process, color gamut, archival qualifies and paper options not impressive in this century.
-
Thanks for the perspective.
I guess as one who has owned and operated LED based silver halide printers for over 20 years (my first one was the Kodak "Pegasus" LED printer which produced pretty remarkable results), I just don't see this as a new process. I think those that have never seen many digital C prints are surprised that the "limited" gamut and "meager" 300 dpi resolution can result in very sharp prints with great saturation, smooth gradations, and visually very competitive with high end inkjet prints.
So my curiosity is piqued ... have they really managed to improve the digital c print over previous technologies. Are they going to manufacture and sell the device to other labs? (seems that's would be their main goal, to market the technology). If you and Kevin can figure out how to test that it would be very enlightening.
Wayne, for the core technology they're using existing machinery and materials - their claim is that they have substantially improved upon the existing technology in the ways described in the article - higher resolution, sharper output, improved colour accuracy relative to the file values. Now, how much better it is relative to other C-print producers I agree is an interesting question that deserves an independent evaluation, but to answer it one would need to research which lab(s) to test them against and possibly set-up a dialog with them; it's a big job. I'll discuss with Kevin and see what he thinks. I'm wondering how large the pool of interest in the LuLa community.
Oh - and I should add - I don't know that they have in mind to market their proprietary technology. I think they are aiming at expanding the print and book offerings from their own facilities.
-
Let's not forget process control (consistency in output over time). I'm sure Wayne can tell us how much work it takes from the printers owner's end. But the end user expects (or should expect) the RGB values he output's today and in a year will look identical. And that's easy with modern ink jet printers. What's the dE drift when sending out colors today and in a month? Or among a lab using more than one machine. Again, with a modern ink jet (certainly Epson's and I've got plenty of colorimetric trending data), the differences are not visible.
The accuracy data metric is dE(00), comparing the file values of a 31 patch check wedge with the read values from the print of that check wedge (in Absolute Rendering Intent), provided along with the client’s print.
That sounds like marketing speak! If you send the IDENTICAL set of patches (and 31 is rather tiny) through a Spectrophotometer, you'll NEVER get a dE of 0.00! Impossible. There's noise in the individual readings of the same two sets of patches. Had the Marketing department who made this claim understood such a fact, they would have stated something like 0.04 or something like that (be happy to upload an actual colorimetric report of dE differences measuring the SAME target twice in a row).
-
Mark,
What interests me more than the C-prints is their "photobook printing". At present, they ask clients to use In Design instead of providing a template of their own and only create the book block which needs to then be sent to a binder. The description sounds like it may be an improvement over the quality produced by most online photobook producers. From my discussion with them by e-mail, I've learned that Kevin may be using this service. It would be helpful to a lot of us if an evaluation of this product could be made. Even if it is subjective as opposed to time-intense objective testing.
-
Mark,
What interests me more than the C-prints is their "photobook printing". At present, they ask clients to use In Design instead of providing a template of their own and only create the book block which needs to then be sent to a binder. The description sounds like it may be an improvement over the quality produced by most online photobook producers. From my discussion with them by e-mail, I've learned that Kevin may be using this service. It would be helpful to a lot of us if an evaluation of this product could be made. Even if it is subjective as opposed to time-intense objective testing.
Kevin did have a book made and I'm sure he'll be talking about it in his forthcoming essay on Lumejet, and most likely that discussion would be subjective. If they really confine their ability to handle customer layouts to InDesign I think they will be limiting the scope of their market, as there are most probably many more people who would like to get books made than who use InDesign.
-
Let's not forget process control (consistency in output over time). I'm sure Wayne can tell us how much work it takes from the printers owner's end. But the end user expects (or should expect) the RGB values he output's today and in a year will look identical. And that's easy with modern ink jet printers. What's the dE drift when sending out colors today and in a month? Or among a lab using more than one machine. Again, with a modern ink jet (certainly Epson's and I've got plenty of colorimetric trending data), the differences are not visible.
That sounds like marketing speak! If you send the IDENTICAL set of patches (and 31 is rather tiny) through a Spectrophotometer, you'll NEVER get a dE of 0.00! Impossible. There's noise in the individual readings of the same two sets of patches. Had the Marketing department who made this claim understood such a fact, they would have stated something like 0.04 or something like that (be happy to upload an actual colorimetric report of dE differences measuring the SAME target twice in a row).
Ya, they know a thing or two about process control. As mentioned in the article and on their website, they calibrate their machines every shift and provide step wedges with every job. So there is a high probability that what you print WITH THEM next year will come back looking the same as what you printed WITH THEM this year - but this doesn't mean that sending the same files to another kind of C-print lab will produce identical output. So on a closed loop basis it's probably reliable from job to job for the same file, but outside of that, not so sure.
dE(00) is an industry abbreviation for dE 2000. It has nothing to do with specific measurements - that's the dE formula they are using.
-
Yes not a very impressive review (also a tough read) gotta wonder why it even got reviewed...to warn people off?
-
It's always interesting to hear of new print services. I'll probably give them a spin, although I too am puzzled as to whether they are offering anything radically different from Lightjet or Lambda printing (which are well-catered for here in London). They operate from an industrial estate in Coventry, so I guess it's mail-order only. And their maximum print width seems to be 30cm (up to 100cm long) – so no exhibition prints, just books and (small) portfolios. To their credit, they've garnered some good testimonials. It's rather confusing that they have two websites:
http://www.l-type.com/
https://www.lumejet.com/
-
Ya, they know a thing or two about process control. As mentioned in the article and on their website, they calibrate their machines every shift and provide step wedges with every job.
That may be what they say, have you measured and reported output trending? Over what time frame and what average (and as importantly max) dE?
So there is a high probability that what you print WITH THEM next year will come back looking the same as what you printed WITH THEM this year
Any outside data to support that?
dE(00) is an industry abbreviation for dE 2000. It has nothing to do with specific measurements - that's the dE formula they are using.
It is, an industry abbreviation for dE 2000, where? Not seeing it here (http://www.colorwiki.com/wiki/Delta_E) or here (https://en.wikipedia.org/wiki/Color_difference) or here (https://www.xrite.com/-/media/xrite/files/whitepaper_pdfs/l10-001_a_guide_to_understanding_color_communication/l10-001_understand_color_en.pdf) etc.
So on a closed loop basis it's probably reliable from job to job for the same file, but outside of that, not so sure.
Probably, perhaps but we really don't know do we? Be kind of interesting to have them output actual targets of a lot more color patches over time and measure them independently.
-
The major thing that seems to be missing from the article and their web-site is the cost. So it wasn't clear to me whether I was supposed to think that this means I should no longer have an Inkjet at home, or if it was a low-cost better quality alternative to Blurb, or...
-
There are links to PDF price-lists at the bottom of this page:
http://www.l-type.com/order
An 8x10 inch print costs £9.48 (inc VAT).
Theprintspace (a popular London lab, using Lambdas - I think) charge £6.86 for the same.
But L.Type's thing seems to be mounting prints, either to thin board, or back to back. Theprintspace will do the former, but then the price goes up to £14.11 (whereas L-type remains £9.48). Theprintspace doesn't do the latter. It is this feature (back-to-back printing) which makes their service interesting (for c-type portfolios).
-
I will have a video and article on this printing process in the next week or so. Pricing, ordering and a more in-depth look at the products will be part of this. This is an interesting service for those that don't want to do printing or are looking for portfolios or books. I am traveling this week and this article is top on my list for next week.
-
The major thing that seems to be missing from the article and their web-site is the cost. So it wasn't clear to me whether I was supposed to think that this means I should no longer have an Inkjet at home, or if it was a low-cost better quality alternative to Blurb, or...
No, it's not "missing from the article" in the sense of AWOL :-) - I did say I was not covering commercial aspects, as all that is on their website and not the focus of my interest in reviewing the product.
-
I am posting this at the risk of revealing myself to be a member of an apparently tiny minority of Lula members who do not own nor plan to own an inkjet printer.
I certainly would love to have the superior color gamut of inkjet pigment printing, the print longevity, the apparent detail; but also the smoothness and seemingly infinite detail of the N-surface paper from large format film, and for it not to cost a fortune, and for the prints not to be damaged by a slip of the finger.
I don't have the room to house a top quality inkjet printer nor the patience to deal with various necessities for getting excellent prints (calibrating the printer, different profiles for each paper, keeping them in sync and up-todate, dealing with machine maintenance and periodic problems, and so on).
Those inkjet labs that I have tried either did not offer printing on papers that interested me for most purposes (e.g. WhiteWall) or others that I tried were too expensive, and the quality of output that varied from mediocre to poor.
What I would like to find is a printing process that will give me the quality I found so attractive with color prints made on N surface paper from large format film, yet to have that with photos now taken with full-frame or small medium format size cameras. I like to look at a print with the proverbial nose to the print. Not a practical way of looking, but that's what floats my last-century boat :-).
I'm sure inkjet printing has improved since I last tried, but I don't know what labs I should consider, nor the exact paper surface (although come to think of it there is probably plenty of info on appropriate papers already here on Lula).
I am interested in the Lumjet printing process and also am a little bit interested with WhiteWall's so-called HD process (although WhiteWall offers their 400 PPI process only on glossy paper which I generally don't like). Lumjet offers various papers types including a type of Matte which might be what I'm looking for.
While the Lumjet process may not interest the vast majority of those here, I am glad the article was published, and I may give Lumjet a try. While it doesn't seem like Lumjet would fulfill all my needs (restriction on sizes, mounting options), at least it may be useful for some of my purposes, and also perhaps for at least a few others here on Lula. Even if not, I personally find it interesting to read in detail about improved processes.
My thanks to Mark Segal, and also to Kevin.
Dan
P.S. Hey, I love those posting verification questions. But arithmetic is just so last-century. How about some questions involving calculus, or quantum mechanics. I need something to challenge my aging and increasingly lazy brain :-)
-
...................
While the Lumjet process may not interest the vast majority of those here, I am glad the article was published, and I may give Lumjet a try. While it doesn't seem like Lumjet would fulfill all my needs (restriction on sizes, mounting options), at least it may be useful for some of my purposes, and also perhaps for at least a few others here on Lula. Even if not, I personally find it interesting to read in detail about improved processes.
My thanks to Mark Segal, and also to Kevin.
Dan
..................
You are welcome. Glad you found it to be of interest.
-
photodan, if you're in North America, it might be worth noting that you can just use mpix, which as far as I know uses Durst Theta gear. Miller's the professional side of the house, is rather more clear about what they use. Although they're shy about printers, they do make clear that the papers are RA-4 papers, so they're using someone's led/laser system.
If you do try out both Lumejet and a less, um, hyperbolic option, I think there are plenty of people who'd like to hear your impressions!
-
photodan, if you're in North America, it might be worth noting that you can just use mpix, which as far as I know uses Durst Theta gear.
Somewhat bad news, they demand sRGB:
https://www.mpixpro.com/help/help.aspx?id=21
-
Somewhat bad news, they demand sRGB:
https://www.mpixpro.com/help/help.aspx?id=21
Is that really a problem for digital c-types?
-
The pro side of the house might do better (Miller's) but I bet not. They're aimed at professionals in the sense of wedding photographers, who (like me) tend to have a pretty blunt instrument take on color management: "does the skin look like skin? does it look like HER skin? we're good to go!"
But mpix is a good cheap way to get an absolute baseline for what these kinds of tech can do for you. It's basically a little more expensive, and (maybe) a little better
than a drugstore print, but it *is* a C-print off a laser printer, not an inkjet. At least, as far as I can tell.
Are all print houses as bloody annoyingly vague about their equipment and methods? They're all vague, blabbering on about their amazing feel and the rich quality of the prints and the vibrancy of the color and jolly little about "We use Durst Thetas and the gamut looks like this and there ya go" or whatever.
-
Is that really a problem for digital c-types?
Yeah, if you're hoping to use the color gamut of the output device. I mean, it's not like sending sRGB to an output device guarantees it will suck, but you're throwing the baby out with the bathwater in the process. Easy to test to (maybe something like this example below will be tested here....):
The benefits of wide gamut working spaces on printed output:
This three part, 32 minute video covers why a wide gamut RGB working space like ProPhoto RGB can produce superior quality output to print.
Part 1 discusses how the supplied Gamut Test File was created and shows two prints output to an Epson 3880 using ProPhoto RGB and sRGB, how the deficiencies of sRGB gamut affects final output quality. Part 1 discusses what to look for on your own prints in terms of better color output. It also covers Photoshop’s Assign Profile command and how wide gamut spaces mishandled produce dull or over saturated colors due to user error.
Part 2 goes into detail about how to print two versions of the properly converted Gamut Test File file in Photoshop using Photoshop’s Print command to correctly setup the test files for output. It covers the Convert to Profile command for preparing test files for output to a lab.
Part 3 goes into color theory and illustrates why a wide gamut space produces not only move vibrant and saturated color but detail and color separation compared to a small gamut working space like sRGB.
High Resolution Video: http://digitaldog.net/files/WideGamutPrintVideo.mov
Low Resolution (YouTube): https://www.youtube.com/watch?v=vLlr7wpAZKs&feature=youtu.be
-
Yeah, if you're hoping to use the color gamut of the output device.
Do these devices (digital c-type printers) have a gamut that is significantly bigger than sRGB?
-
Do these devices (digital c-type printers) have a gamut that is significantly bigger than sRGB?
Case in point, three similar and much older printers who's color gamut is much smaller than a modern ink jet (and you're only seeing one 'view' in 3D here).
The red plot is sRGB. Colors are what clip sending sRGB to such devices (if the image contains such colors). Not pretty:
(http://digitaldog.net/files/sRGB_vs_SilverPrinters.jpg)
-
Thanks.
I normally work in Adobe RGB, and when I make digital c-types it's normally the reds that get knocked back the most. The same thing happens when I convert from Adobe RGB to sRGB, so I wondered if sRGB had a similar gamut to the digital c-type. But you've shown that digital c-types can represent blues, greens and yellows that get chopped off by sRGB. Not good for landscape photographers.
(My lab supplies profiles, and asks its customers to make the conversion, so I know what to expect.)
-
Even if you've got a crappy little space on output, starting somewhere big and loose gives you a lot more room for squeezing into the little space in the best possible way. I guess.
-
Thanks.
I normally work in Adobe RGB, and when I make digital c-types it's normally the reds that get knocked back the most. The same thing happens when I convert from Adobe RGB to sRGB, so I wondered if sRGB had a similar gamut to the digital c-type. But you've shown that digital c-types can represent blues, greens and yellows that get chopped off by sRGB. Not good for landscape photographers.
(My lab supplies profiles, and asks its customers to make the conversion, so I know what to expect.)
The big differences is in the green primary. But yeah, Adobe RGB (1998) has a larger color gamut and as importantly, something called Gamut Efficiency. What gets interesting is to examine the Lab Gamut Efficiency of differing color spaces. You can see a list on Bruce Lindbloom's excellent site Information About RGB Working Spaces.
Information About RGB Working Spaces (http://www.brucelindbloom.com/WorkingSpaceInfo.html)
Examine sRGB who's Lab gamut efficiency is a mere 35%, then Adobe RGB (1998) at 50.6% while ProPhoto RGB has a Lab gamut efficiency of 91.2% despite having two primaries (and thus device values) that do NOT define colors (we can't see them).
-
I've taken a cursory look at their glossy profiles and they seem to be virtually identical to LightJets, Chromiras, AND Lambdas.
Hmm. I wish them luck.
-
I've taken a cursory look at their glossy profiles and they seem to be virtually identical to LightJets, Chromiras, AND Lambdas.
Good to know! With the same/similar papers, kind of what I'd expect.
-
Elliot, to answer your questions in Replies 24 and 27 directly using the largest gamut Lumejet profile (a bit larger than Lumejet Luster), I've included here several gamut diagrams prepared in Colorthink Pro. They show:
#(3) the Lumejet profile has a different shape from sRGB and in some places exceeds sRGB, therefore files limited to sRGB do not use the full possible gamut of the Lumejet process.
#(4) the Lumejet profile fits comfortably within ARGB(98) gamut, explaining why Lumejet asks customers to deliver their files prepared in ARGB(98) space.
By comparison, the largest gamut printer profile I've ever generated (i1Profiler, i1Pro2) was for Red River San Gabriel Gloss Baryta (very similar to Ilford Gold Fibre Silk). It's gamut shape is different from those of ARGB(98) and ProPhoto RGB. #1 shows that it exceeds ARGB(98) in certain parts of the gamut, while #2 shows that ProPhoto RGB fully encompasses it; hence to take full advantage of the San Gabriel Baryta gamut, one is best advised to prepare ones files in ProPhoto RGB colour space. So one is dealing with quite different gamut conditions for C-type versus a very wide gamut inkjet process. This could have differing implications for different kinds of photos as discussed in my Lumejet article, as well as in other paper and printer reviews I've prepared for this site.
-
#(3) the Lumejet profile has a different shape from sRGB and in some places exceeds sRGB, therefore files limited to sRGB do not use the full possible gamut of the Lumejet process.
#(4) the Lumejet profile fits comfortably within ARGB(98) gamut, explaining why Lumejet asks customers to deliver their files prepared in ARGB(98) space.
ALL printers have a vastly different shapes compared to all RGB working spaces. All RGB working space have similar and predictable shapes due on the fact they are based on theoretical emissive display, not printers.
NO Printer can print all of sRGB and thus any RGB working space. Again due to the facts above.
Modern ink jet printers like the Epson shown below have color gamuts, depending on paper of course, that greatly exceed Adobe RGB (1998); they do not comfortably within ARGB(98) Adobe RGB (1998) color gamut.
Red plot: Adobe RGB (1998) vs. my Epson 3880 with luster paper. Again, at least in terms of color gamut and an appropriate working space color gamut, Adobe RGB (1998) doesn't cut it for the Epson and the plots show this Lumejet isn't at all impressive in terms of color gamut compared to that ink jet.
-
Ok, so I am taking another swing at this thing, even though color science makes me all cross-eyed. I'm kind of getting lost in the technicals.
"Every pixel is created as a unique exposure for every one of the 576 RGB LEDs. This is finely controlled by digital circuits that give 2048 grey levels/pixel for Red and Green and 1024 levels for Blue (32bit levels). The individual RGB exposures for each of the 576 LEDs are delivered down the fiber taper and projected in parallel so that they image vertically onto the paper."
Does the print head do 576 pixels in one go, and then go on to the next 576 pixels? Or are all 576 LEDs involved in each pixel? Or, um, is it actually 192 S, 192Gs and 192Bs, doing 192 pixels at a go, or what? The first sentence seems to have been mangled, or maybe I am just persistently not reading it the way it's intended. Either way I cannot make any sense out of what it means.
Slightly later we find:
"4 billion unique colors possible for each printed pixel"
Is this even meaningful? I mean, it sounds sexy, but I'm pretty sure that's orders of magnitude more colors than we can see?
I feel like I could make sense of the analysis of the results, and I understand the bit at the end "these things look good", but the description of Lumejet's process is still pretty opaque.
Also, I have to say that Lumejet's quotation of pixel density in squares rather than lines (160K vs 90K) sounds disingenuous, albeit accurate. When you state it as 400dpi vs 300dpi it doesn't sound like Lumejet has such an advantage. If they stuck with it, I might let it slide, but literally every other reference is to 400dpi. It's only when they want to seem bigger that they go with the square.
Just for humor, imagine this sentence:
"As a result of this, LumeJet claims that its 400dpi print quality is greater than that of multi-colour inkjet printing at over 4000dpi."
re-written as:
"As a result, Lumejet claims that it's 160K pixels/sqinch print quality is greater than that of multi-colour inkjet printing at over 16M pixels/squinch"
which, while it says exactly the same thing, feels a heck of a lot less convincing.
-
"4 billion unique colors possible for each printed pixel"
Good catch!
No, it's not possible and it's marketing BS. But we hear this nonsense all the time from lots of companies. We human's can't even see 16.7 million colors. The number is up to debate but far less (I've heard 12 million most often). So what's being discussed here and being incorrectly called colors? Numbers. Device Values. Case in point is this illustration of two device values (numbers) in lowly sRGB that ARE the same color.
For those that wish to understand the vast differences in color and device values, I offer:
http://digitaldog.net/files/ColorNumbersColorGamut.pdf (http://digitaldog.net/files/ColorNumbersColorGamut.pdf)
(http://www.digitaldog.net/files/ColorNumbersNotColors.jpg)
-
Given that their maximum print width is 12 inches, is it safe to assume they are using some sort of minilab (Fuji Frontier, Noritsu etc)?
-
Does the print head do 576 pixels in one go, and then go on to the next 576 pixels? Or are all 576 LEDs involved in each pixel?
The LED counts only affect the end user if banding is visible when the machine does not expose and advance precisely. Trust that the machine works.
What is more important and easier to understand is, the machine can (or should as many others already do) print pixel-for-pixel from 400 ppi files. Although their website says one cannot see the individual pixels, I'm sure they are mistaken. All these machines easily print discernible individual pixels when the test is provided. Single red, green, blue pixels on white, gray, and black backgrounds are easy to see with a loupe of 8~12 power. Single pixels on white are to be found on ink jets, not so easy to see on gray or black backgrounds. Pixels subsampling is employed also. Lenses may be used also. Common lenses such as Nikon in the case of Chromira LED printers. Thus a digital enlarger exposing typical color photo print material. Awesome, still.
-
Yeah, I get that they're printing dots at 400dpi, but I feel like if you're going to include a paragraph on the 576 LEDs then I ought to
be able to work out what the paragraph means, you know?
-
Yes not a very impressive review (also a tough read) gotta wonder why it even got reviewed...to warn people off?
Doug, you ask why Lumejet got reviewed, and whether it was to "warn people off". I reviewed this one and I've done enough reviews for this site, that I feel I can share some insight that may answer your question. Reviews on LuLa are usually the result of the reviewer believing (and if it's not the publisher himself, the publisher agreeing) that the product or service is worthwhile bringing to the community's attention, normally because it is new, a novel take off on something not so new, or has noteworthy qualities. Speaking for my own work, my interest is primarily to explore the merits, call the shots as I see them and let people make up their own minds about whether they want to buy or not buy. No product scores a 100 and every reader has their own taste about how important any particular feature or issue is to them. Anyone should be hard-put to view the concluding paragraph as "warning off", including due consideration of various objective factors I discuss in the body of the article. Sorry you found it a "tough read". Others have raised several specific points deserving further elaboration. Hang-in, there's more to come.
-
Peer review is a bitch :(
-
I would like to introduce myself. My name is Huw Williams. I am the Chairman and lead investor for LumeJet Print Technologies, which offers the L.Type print service reviewed by Mark Segal last month. I have been a reader and great admirer of this site for a couple of years now and have learnt an enormous amount from the site itself and the forum, too. In fact, I approached Kevin for the first time over a year ago now to see if he would be prepared to take a look at our printing largely because of the hugely in-depth reviews that Mark has carried out on other printers - we wanted to see how we compared. We have also learnt a lot from the approach that Mark has taken to reviewing other printers - as well as the process he put us through, which was extremely rigorous.
I have seen the many comments in the forum relating to the service and wanted to try to respond to as many as possible. I will try to respond to each question or comment in turn, so please forgive me if there is a little bit of repetition. If my answers raise further questions I shall of course be happy to answer them to the best of my, and my team's ability. You can also contact me directly at huw.williams@lumejet.com.
I will just make a few general comments up front, before I answer specific questions.
The most important point is that we do not want to make greatly exaggerated claims for our technology. We print on the same C-Type paper, and develop using the same chemicals, as any other printer can use (mostly Fuji Crystal Archive DPII papers). We are therefore subject to the same limitations on gamut as anyone else using those papers: the paper itself defines the overall gamut volume and limitations, although how you use that gamut volume (through printer profiling) does allow for some variation between machines and operators. Where we are different is that we have developed an entirely new machine, from the ground up, over a period of over 15 years. The machine was originally developed to solve the problem of how to print extremely sharp text on silver halide - and the precision required to do that also allows us to print extremely sharp photographic images. In my answers below I will try to summarise how we did that and why our approach is different to other machines, and why we call the resultant print the L.Type.
The next thing that makes us different is that we are the only print service in the world that designs and builds its own machines. We don't sell them to anyone else. Instead, we own and operate, maintain and upgrade them all ourselves. The engineers who designed and build the machines run them every day - almost like a racing car team.
I make no claims and take no position for C-Type images against inkjet. As has been made clear in Mark's review and many of the forum comments, modern inkjet printers far surpass C-Type prints in terms of gamut. That's a fact of life about which we can do nothing as we don't make the papers ourselves. Within C-Type gamut (which broadly covers sRGB and Adobe 98 at least), there is a valid discussion to be had as to which approach (C-Type, which is continuous tone, or inkjet, which is a half-tone process) gives the better results. My personal feeling is that this is very much a matter for personal taste. I'd go so far as to say that some images look better in one form, and others in the other - but I don't believe there is a 'right' answer. What we do believe, strongly, is that C-Type prints, properly done, are wonderful - and our aim is to stand at the pinnacle of C-Type printing. Only you, the photographers, can judge whether we are close to achieving that. Giclee prints are also wonderful and will no doubt continue to improve: we just don't see any reason why C-Type development should just stop - there is more to come from that medium.
Finally, I want to say that the people on this forum are unusual. The degree of technical knowledge and expertise here is enormous. I am sure that we have a great deal to learn from you. But you are not necessarily our target market. You know how to profile and operate the best inkjet printers and how to get the best results out of them. You have the time and expertise to evaluate different papers and to compare results. But many photographers do not have that time or expertise - or simply don't want the bother of doing so. We want to offer a service that offers reference standard C-Type printing to those photographers - whether they are among the world's leading pros (we number several Phase One and Hasselblad devotees among our user base) or keen amateur photographers. We want people to know that within our limitations (we only go up to 300mm x 1000mm at this stage) we will always strive to do the very best. We won't alter your images in any way, and we will print what we are sent as well as possible and at a reasonable price, and present them as nicely as we can. I very much hope you will give us a try and would be delighted to welcome any questions or comments from you - especially if they help us improve what we offer. We are on a journey and I hope we will continue to improve over time with your help.
I hope the answers below are useful. Thank you for your interest.
-
I confess that I did not perform a DEEP read of this article, but it did not strike me as a particularly new thing at all?
People have been shining colored lights onto silver halide paper to make digital prints for a long time. I don't quite get how Lumejet is any
different from, say, a Durst Theta printer, which has been around for quite a while.
What am I missing?
The essential statement here is correct – and this goes for many other comments made (especially re gamut). We are not doing anything new in terms of the basic principles of shining RGB lights on to silver halide paper, and we are using the same silver halide paper as everyone else. What is different is our focus on the precision with which we do that, both in terms of the physical direction of the light onto the paper, and the control of those light pulses, as well as the way we interact with the paper itself.
There is some (intentionally relatively superficial) description of this at https://www.lumejet.co.uk/technology/. Our technology has been under development since approximately 2000 and many millions of $ (close to $25m) have been spent to date on developing the machine.
Fundamentally, silver halide paper has one basic limitation - the gamut of the paper and the emulsions. This is something we can do relatively little about; we use the best Fuji Crystal Archive Professional DPII papers – their exhibition quality papers – that are available to everyone else (although not all pro labs use these papers, and consumer-oriented labs do not). What we have done ourselves is to re-characterize each media (paper type) using our RGB print head and created bespoke target files for each paper, and then build the best colour profiles we possibly can for our machines.
For the purpose of colour management, a printer must be calibrated to the centre line of the Lab colour space. This calibration produces a neutral Lab grey line from the media white point through to the media black point (DMAX). The calibration provides a repeatable reference point that can be calibrated to before a shift. The reference point is used as the base line foundation upon which the machine’s media colour profile (.icc) is characterised and built.
A media's neutral grey line calibration is specified in the media target (.tgt) file by specifying a number of input R,G,B values and corresponding output C,M,Y colour densities. The target file has a number of incremental steps of input RGB values from white (0xFFFFFF) through grey to black (0x000000). For each step, there is a corresponding output C, M, and Y media colour density that when measured in D50 Lab colour space, produces a Lab neutral colour. When all steps are put together, a Lab neutral grey line is produced by the media. The target file is used by the printer to calibrate the media's DMAX (black). After black calibration, the target file is used to balance the grey to produce the neutral grey line for colour management.
Our target files yield highly accurate neutral LAB gray lines and high DMAX. This is a skilled and ‘black art’ process that few know how to do: Fuji themselves do not do this (or do not share the results if they do), so we had to develop a technique and process in-house to do it ourselves. We believe our target file generation process is very accurate along the whole tone curve and hence we follow the neutral grey line in the 3D gamut. This then gives us the foundation to build our colour profile more precisely through the colour space to the outer edges of our gamut and to determine how we handle out-of-gamut colours.
The key advantage of silver halide is that it is continuous tone. So a single image pixel of any colour is represented on the page by a single pixel of that colour – as opposed to the structured combination of dots of different coloured inks required to create a pixel of a single colour with an inkjet. This does not always matter, but complex effects like subtle split-toning and gentle tonal variation over a page can be very difficult to achieve correctly with inkjet half-tone processes. Halftone is a trick of the eye (side by side dot combinations) that requires a K layer to make up for CMY inks not producing a dense black. Contone is real Newtonian light combination of just CMY dyes as tiny colour filter plates in a vertical stack (dot on dot). And our CMY unit cells are much better aligned, to microns, than the laser and other LED printers, as those were not designed to handle fine text and graphics (requiring 400dpi, the Nyquist minimum for hand held prints). We estimate that inkjet systems would require 12 colours and 10x the pixel density to produce something approaching contone quality .
What really makes the difference with silver halide is how the paper receives light. This is where the majority of our improvement lies. The emulsion layers of the paper contain silver crystals and colour couplers that together, when activated by light, form colour clouds. These clouds are approximately 5 microns in diameter. So the paper is itself capable of resolving extremely small (invisible to the human eye) levels of detail. But this depends on how accurately the paper is imaged. In a traditional darkroom enlarger set-up, light from a small source was beamed through a negative onto paper of the relevant size. The enlargement process caused softening of the image because of the enlargement itself and edge effects as light towards the edge of the paper was not hitting the paper vertically. All silver halide print machines attempt, in various ways, to address this issue – more or less successfully. Some use lasers, some use LEDs, but each of them has imperfections in the accuracy of this imaging process.
We developed the Lumejet printer from scratch over the past 15 years, starting from research projects in Warwick Manufacturing Group’s laboratory in Warwick, UK. The research came about to address the fact that it was then impossible to print sharp text accurately on silver halide. Our printer uses a proprietary print head that incorporates a patented optical fibre taper to reduce an array of 576 individually-addressable RGB LEDs down to a very precise spot. Each LED is 300 microns in diameter, and these are reduced by the optical fibres to 60 microns each. The fibre taper is essentially a pixel projector that takes an array of larger LEDs and produces smaller pixels on the paper to produce ultra-sharp images, text and graphics. The print head is positioned just above the paper and scanned, to micron accuracy, across the paper (like an inkjet with light). As the head passes over the paper, successive RGB LEDs are triggered so that each pixel area on the paper is imaged successively by one LED of each colour, all placed exactly on top of each other. The finished pixel is therefore formed extremely accurately and receives only the light signal from the three R,G,B LEDs forming that pixel. There is minimal cross-talk between pixels. Because the print head is always directly above the paper, light beams always enter the paper exactly vertically, eliminating edge effects. The motion control and signal processing software that allows each and every pixel to be placed precisely where it should be is highly complex and at the core of our proprietary process. The printer prints 12mm swathes at approximately 1.5m/s across the paper (Y axis) to a pixel placement of c. 1 micron. The paper is then indexed forward by the swathe width (X axis), to a precision of c. 5-7 microns, and the next swathe is printed, with a small overlap (which is blended) to eliminate any motion errors.
So the difference between our technology and other silver halide printers boils down to:
(i) our pixel is smaller at 63.5 microns (400dpi) - which is the smallest the human eye can resolve for prints held at 14”.
(ii) our pixels are placed more accurately next door to each other so there is no overlap;
(iii) the paper is always imaged directly from above so cross-talk between pixels is minimized;
(iv) we have made a huge investment in calibrating the printer to the latest papers and profiling our printer to ensure that it gets the most out of those papers without compromise. For instance, some other printers will operate at a higher DMax. This produces deeper blacks, but it also causes yellow flaring on whites due to cross-talk and over-exposure. We currently dial back our DMax slightly to eliminate this problem – we believe it is better to have pure greyscales than to push blacks to the max, although we accept that this will not always be optimal. We continue to work on this area and expect to increase DMax progressively over time.
(v) Our printers are manufactured today by us, and constantly updated – unlike most of the high-end competition that has been discontinued.
(vi) We are the only photo lab in the world that builds its own printers from scratch, so far as we are aware.
In terms of the more detailed technical differences that make our printer stand out:
Principally it is in the design of the print head, which came about from first principles of physics laid out at the initial Warwick R&D stage more than 15 years ago e.g.
1) How the 5:1 fibre taper bundle is made and drawn (so there are roughly 300 fibres in front of each LED) and structured, with interstitial black rods between the fibres used to reduce light scatter;
Attached is an image of the top of a tapered fibre bundle:
2) How the fibres are made (by multiple draws of glass with different indices) and the Numerical Aperture of the fibres themselves, to couple the LED light into the fibre internal core, rather than scatter sideways. The NA of our fibres is c. 0.80, which couples light efficiently down the fibre core with little sideways cross-talk, producing a “tapered light funnel” effect in front of each LED with very sharp edges;
3) How the individual RGB LEDs are robotically placed (to c. 20 microns), wirebonded and aligned on the print head array and exposed through a very accurate and hard edged black mask. This is Gerber plotted at 2400dpi, with precise pitch and interlacing between odd and even rows, to provide c. 10% pixel overlap and avoid micro-banding. We are in fact imaging the mask edges, rather than the LEDs (which vary in size and position);
4) How the fibres are drawn from 20 microns (top) to 4 microns (bottom), so although we print with a 63.5um (400ppi) spot, the edge of the spot is constrained by the 4 micron fibres at the exit end and gives very little halation (flare, light scatter);
5) How 4-6 microns is also the grain size of the AgX emulsions – so we are effectively dropping photons on grains (all part of the design); and
6) How the RGB light at the fibre exit is transferred 1:1 to the paper using a telecentric relay lens (9 stack lens) that has been designed to minimize chromatic aberration (different paths of different wavelengths) and produce parallel light rays into the emulsion.
Sample scans of text and images are shown at https://www.lumejet.co.uk/technology/
Our unit pixel cells are clean with little cross-talk. All of the above design elements were initially addressed and produce really clean, sharp text and graphics: they make for a beautifully sharp photographic reproduction – while older photographic print techniques are inherently less sharp and produce beautiful photos partly through that inherent softness!
To pick up on a few well-known machines:
The Durst Lambda was an RGB laser spinner system, which imaged at 200dpi natively (switchable to 400dpi) and suffered from RGB pixel alignment from middle to edge due to the F-Theta flat field lens (it is also 50” wide). It was aimed at poster making, not small format prints.
The Durst Epsilon and Theta (30”) used multiLED prints heads, imaging at 256dpi, but their LEDs were placed to one side and the light was piped by fibre cables and a large lens onto the paper. Again they were built mainly for images, not text and graphics, and the final beam shaping/delivery was not done as LumeJet.
Other machines are also very good, but none was designed to print text and therefore all suffer from some compromises in sharpness and accuracy relative to our machine.
-
Thanks for the perspective.
I guess as one who has owned and operated LED based silver halide printers for over 20 years (my first one was the Kodak "Pegasus" LED printer which produced pretty remarkable results), I just don't see this as a new process. I think those that have never seen many digital C prints are surprised that the "limited" gamut and "meager" 300 dpi resolution can result in very sharp prints with great saturation, smooth gradations, and visually very competitive with high end inkjet prints.
So my curiosity is piqued ... have they really managed to improve the digital c print over previous technologies. Are they going to manufacture and sell the device to other labs? (seems that's would be their main goal, to market the technology). If you and Kevin can figure out how to test that it would be very enlightening.
I think I have essentially covered this above. What we call the L.Type is generically still a C-Type print. We use the same papers and same chemicals as others. But we use completely different equipment to image that paper compared with what anyone else does.
The easiest way to see how we are different is to look at our text. This is much sharper than text on other silver halide printers and has regularly beaten litho/ Indigo in commercial tests by prospective customers. The pixel control and sharpness that is demonstrated by our text printing carries over into our photo printing – it’s the same thing.
To answer the specific point about manufacturing/ selling the device/ technology: the answer is 'no' - we do not want to sell it at all. To be frank, our machine is slow. And in a world where everyone is obsessed with building faster and faster machines - particularly in silver halide printing - the economics of running a machine slowly to try to make a better print are not attractive to most labs. So we think that a more sustainable business model for us is to operate all the machines in existence ourselves and keep upgrading and developing them as we go. Every time a customer suggests something that could improve - or has gone wrong - we try to learn from it. But if we continue to operate all the machines ourselves, speed is much less of an issue - and that allows us, we think, to produce better prints.
-
Let's not forget process control (consistency in output over time). I'm sure Wayne can tell us how much work it takes from the printers owner's end. But the end user expects (or should expect) the RGB values he output's today and in a year will look identical. And that's easy with modern ink jet printers. What's the dE drift when sending out colors today and in a month? Or among a lab using more than one machine. Again, with a modern ink jet (certainly Epson's and I've got plenty of colorimetric trending data), the differences are not visible.
That sounds like marketing speak! If you send the IDENTICAL set of patches (and 31 is rather tiny) through a Spectrophotometer, you'll NEVER get a dE of 0.00! Impossible. There's noise in the individual readings of the same two sets of patches. Had the Marketing department who made this claim understood such a fact, they would have stated something like 0.04 or something like that (be happy to upload an actual colorimetric report of dE differences measuring the SAME target twice in a row).
This is a very good point. We do everything we can to address this. From profiling our machines; calibration every shift, every new roll; colour process control checking after calibration and monitored during the shift; common chemical reservoirs so all machines are fed identical chemical mixes; precise temperature controls on the processors so all processing takes place in tight tolerance; etc. We use the 31-patch test scan on each order and record those values and will try to provide you with some data over time, though please be aware that that process has been in use for less than a year. The 31 patches, which are the CC24 Lab patches with 6 extra gray scale patches and a media white patch) are all known LAB values and printed absolute colorimetric then measured independently using basICColor software to validate the repeatability and accuracy of the process using a deltaE 2000 measurement.
The accuracy of the prints is being measured against a fixed absolute that is the same for every printer. We believe we hold best-in-class accuracy and it compares well with professional inkjets on each and every job.
I think Mark has addressed the marketing speak point above - we absolutely don't claim a dE(2000) value of 0. That's clearly impossible. As you rightly state, even scanning the same strip twice produces variation in the results. Part of that in fact is because the profiler we use to scan marks the paper, so it can't be scanned again on the identical line, and part of it is because of the natural error in the scanner. But we claim to be broadly in the range of 0.6-0.9 dE(2000) on average across the 31 patches on each print we make and we reject any print with dEs on any individual patch that would provide a visible difference between target and actual output. We believe this is highly competitive against even the best inkjets when they are properly profiled and properly used, particularly impressive in comparison to other silver halide printing, and certainly likely to beat any printer that is not properly profiled/ used/ maintained - which should be one reason why the average user chooses a service like ours.
-
Huw, Thanks for all the detailed information. I also will have a video and report on my experience with the L.Type prints published very soon. I was quite impressed and I think the article and video will explain things more clearly too. Stay tuned.
-
Mark,
What interests me more than the C-prints is their "photobook printing". At present, they ask clients to use In Design instead of providing a template of their own and only create the book block which needs to then be sent to a binder. The description sounds like it may be an improvement over the quality produced by most online photobook producers. From my discussion with them by e-mail, I've learned that Kevin may be using this service. It would be helpful to a lot of us if an evaluation of this product could be made. Even if it is subjective as opposed to time-intense objective testing.
We aren’t desktop publishers – we are set up to print finished files that photographers and commercial designers send us and we typically take in either Indesign files or print-ready pdfs. Having said that, we are working hard to produce a simple bookmaker on our mobile app that will make it easy for the average photographer to produce a good basic layout that will cover 90% of situations. My guess is this is 3 months away but we are working on it.
Kenneth, you and I have exchanged emails as you mentioned - and I'd be delighted to print a test book for you if you would like to send us some images.
-
It's always interesting to hear of new print services. I'll probably give them a spin, although I too am puzzled as to whether they are offering anything radically different from Lightjet or Lambda printing (which are well-catered for here in London). They operate from an industrial estate in Coventry, so I guess it's mail-order only. And their maximum print width seems to be 30cm (up to 100cm long) – so no exhibition prints, just books and (small) portfolios. To their credit, they've garnered some good testimonials. It's rather confusing that they have two websites:
http://www.l-type.com/
https://www.lumejet.com/
It is hard to argue that we are radically different. But people pay good money for incremental changes. Our printer is quite simply the latest and the best developed in the silver halide world and produces the best results. On the other hand, it has clear format limitations that the Lightjet or Lambda don’t – so for larger format prints they will remain an appropriate solution.
We are primarily internet-based, though we are always very happy to see photographers and anyone can visit any time. We often host educational tours for students, too.
Your point about websites is well made. Our first website was www.lumejet.com and this was based on a ROES-type platform that allowed print uploads and a modicum of editing. I would say that this is very much a consumer-oriented approach. We found, however, that the vast majority of our users preferred not to have to upload their prints individually before placing an order - most people seem to prefer to use Wetransfer, or Dropbox or some similar storage mechanism and just to send us a link to their files. So L-Type.com was designed as the site that tells our story, shows what we can do, and essentially allows you to send us the files in any way you want to get them to us. After that, we take your files, follow your instructions, prepare a contact sheet for approval and send a payment link before printing - we know this takes a bit longer, but it seems to give better results in most cases. If you want a fire-and-forget solution we have just launched an iOS app (iOS only, I'm afraid) that is on the App Store under L.TYPE PRO PHOTO PRINTS. This integrates with Dropbox and allows you to order high-res images using your mobile device as a 'remote controller'. The images need to be in a Dropbox folder (we hope to add other integrations soon) but assuming you use Dropbox, an order can be completed in less than 60 seconds if you are familiar with the app and it gives you full preview/ cropping/ re-centering capabilty.
-
Be kind of interesting to have them output actual targets of a lot more color patches over time and measure them independently.
I will try to get you some time-based data, though bear in mind that we have only been undertaking this measurement for a few months.
We could of course add more patches, but X-Rite, FOGRA, etc all use a limited number from across the spectrum and this seems a reasonable approach. Adding more patches just means a larger area of paper will be wasted with every print. I’ve addressed the matter of measurements over time.
-
I am posting this at the risk of revealing myself to be a member of an apparently tiny minority of Lula members who do not own nor plan to own an inkjet printer.
I certainly would love to have the superior color gamut of inkjet pigment printing, the print longevity, the apparent detail; but also the smoothness and seemingly infinite detail of the N-surface paper from large format film, and for it not to cost a fortune, and for the prints not to be damaged by a slip of the finger.
I don't have the room to house a top quality inkjet printer nor the patience to deal with various necessities for getting excellent prints (calibrating the printer, different profiles for each paper, keeping them in sync and up-todate, dealing with machine maintenance and periodic problems, and so on).
Those inkjet labs that I have tried either did not offer printing on papers that interested me for most purposes (e.g. WhiteWall) or others that I tried were too expensive, and the quality of output that varied from mediocre to poor.
What I would like to find is a printing process that will give me the quality I found so attractive with color prints made on N surface paper from large format film, yet to have that with photos now taken with full-frame or small medium format size cameras. I like to look at a print with the proverbial nose to the print. Not a practical way of looking, but that's what floats my last-century boat :-).
I'm sure inkjet printing has improved since I last tried, but I don't know what labs I should consider, nor the exact paper surface (although come to think of it there is probably plenty of info on appropriate papers already here on Lula).
I am interested in the Lumjet printing process and also am a little bit interested with WhiteWall's so-called HD process (although WhiteWall offers their 400 PPI process only on glossy paper which I generally don't like). Lumjet offers various papers types including a type of Matte which might be what I'm looking for.
While the Lumjet process may not interest the vast majority of those here, I am glad the article was published, and I may give Lumjet a try. While it doesn't seem like Lumjet would fulfill all my needs (restriction on sizes, mounting options), at least it may be useful for some of my purposes, and also perhaps for at least a few others here on Lula. Even if not, I personally find it interesting to read in detail about improved processes.
My thanks to Mark Segal, and also to Kevin.
Dan
P.S. Hey, I love those posting verification questions. But arithmetic is just so last-century. How about some questions involving calculus, or quantum mechanics. I need something to challenge my aging and increasingly lazy brain :-)
Dan, thank you for your first few sentences.... you are the customer we're trying to support ;)
Re Whitewall, I don’t know anything about the WhiteWall HD process. I believe it is done on a Polielettronica machine. It claims a 600dpi print density (but we believe this is really a 300dpi machine interlacing up to 610dpi – older LaserLabs had resolutions of 254 and 305 dpi - but this may be wrong), but there are a couple of important caveats to that:
(i) pixel placement without overlap is key – and that is core to our approach and hard to achieve with lasers from a central source;
(ii) there are good technical reasons why making your resolution too fine may not work: the real issue is that AgX photopapers start to fall off in resolving power above c. 450dpi (the so called Modulation Transfer Function roll off point), so there is no real benefit to imaging at great resolution. It will depend also on the beam shape and how the RGB pixels hit the paper one on top of the other. With laser spinners they suffer from jigger (gets worse as they age) and the FTheta lenses (flat field correction lens that takes the spinner's RGB laser parabolas and tries to create a ‘flat field’ across the paper width) are not precisely flat field, particularly at the edges, and suffer from chromatic aberration – different paths taken by different wavelengths. This has been confirmed to us by a major manufacturer, particularly for widths greater than 12";
(iii) 600dpi is way smaller than the eye can see. At 600 pixels per inch, you could fit an entire 70MP file on a piece of paper approx. 16”x12” pixel-for-pixel so this is possibly of interest for printing small images, but irrelevant for large images as you would have to ‘invent’ additional image data to get to any larger format at all.
Having said that, other elements of the Whitewall process accord very closely with our own approach - thinking about how you store paper, for example, and deep attention to all the details of chemistry and every other process step so as to get the best possible result. Whitewall are a superb outfit with a great reputation.
-
Ok, so I am taking another swing at this thing, even though color science makes me all cross-eyed. I'm kind of getting lost in the technicals.
"Every pixel is created as a unique exposure for every one of the 576 RGB LEDs. This is finely controlled by digital circuits that give 2048 grey levels/pixel for Red and Green and 1024 levels for Blue (32bit levels). The individual RGB exposures for each of the 576 LEDs are delivered down the fiber taper and projected in parallel so that they image vertically onto the paper."
Does the print head do 576 pixels in one go, and then go on to the next 576 pixels? Or are all 576 LEDs involved in each pixel? Or, um, is it actually 192 S, 192Gs and 192Bs, doing 192 pixels at a go, or what? The first sentence seems to have been mangled, or maybe I am just persistently not reading it the way it's intended. Either way I cannot make any sense out of what it means.
Slightly later we find:
"4 billion unique colors possible for each printed pixel"
Is this even meaningful? I mean, it sounds sexy, but I'm pretty sure that's orders of magnitude more colors than we can see?
I feel like I could make sense of the analysis of the results, and I understand the bit at the end "these things look good", but the description of Lumejet's process is still pretty opaque.
Also, I have to say that Lumejet's quotation of pixel density in squares rather than lines (160K vs 90K) sounds disingenuous, albeit accurate. When you state it as 400dpi vs 300dpi it doesn't sound like Lumejet has such an advantage. If they stuck with it, I might let it slide, but literally every other reference is to 400dpi. It's only when they want to seem bigger that they go with the square.
Just for humor, imagine this sentence:
"As a result of this, LumeJet claims that its 400dpi print quality is greater than that of multi-colour inkjet printing at over 4000dpi."
re-written as:
"As a result, Lumejet claims that it's 160K pixels/sqinch print quality is greater than that of multi-colour inkjet printing at over 16M pixels/squinch"
which, while it says exactly the same thing, feels a heck of a lot less convincing.
576 LEDs in the print head is actually 192 each of 3 colours (R, G, B). Each printed pixel will contain light only from 1R, 1G, 1B LEDs. We use the terms pixel and dot interchangeably because ideally we will print an image at a native 400 image pixels per inch – each image pixel will be represented by one printed pixel, and that in turn will have received 3 light pulses. The scanning head lays down 192 red pulses at a time, and passes on over the page. Behind it comes a row of green LEDs, and so on. So each pixel is imaged 3 times, very accurately, in turn.
There is a perfectly valid question about the use of the figure of 4 billion colours. There is a wide range of estimates of how many colours people can see – anywhere from 1 million to 14 million – though I’m not sure we know whether the ones I see are identical to the same number you see. So it is absolutely questionable whether printing above 8-bit levels (which gives 16.8 million combinations) is relevant to any printer, though everyone is trying to beat that. The main purpose of such fine distinction between pixels is to maintain a very smooth, gentle tonal gradation across a page without any banding or granularity or other effects. So long as the differences between pixels are not visible to the eye, then the resultant image will appear incredibly smooth. Vignettes and sweeps require this tonal range to avoid density breaks, also smooth transitions into catch-light and catch-dark – shine on metallics, filigree in dresses, et al. A huge proportion of the detail and modelling lies in the very light and very dark areas, which halftone naturally drops out – and we increase the amount of tone data in these regions to try to capture this. We were reviewed by Joel Tjintjelaar of bwvision.com, who is an incredibly exacting and demanding photographer and he said he had never seen printing as good as ours for split tones on B&W photography.
One of the other points related to our use of the maths of 400dpi = 160k pixels/ square inch vs 300dpi = 90k pixels per square inch. Again, we need to be clear. We use ‘dpi’ and ‘ppi’ interchangeably. Because, as stated above, one input pixel from the RIP becomes one pixel on the page. In an ideal world, one image pixel from the original camera sensor becomes one pixel on the page (ie there is no rescaling/ resampling anywhere). Then what we print is exactly what was captured. Taking this comparison and applying it to the 4000dpi of an inkjet (and then squaring it up for area) is unfair - because our 400x400 pixels can represent 400x400 image pixels. Whereas an inkjet printer at 4000dpi would use 16 million dots of ink to represent the same 160k image pixels.
We are trying to produce prints that have all of the detail that the young human eye can resolve at arm’s length. Other printers are good for greater distances and larger sizes. We want our prints to be held close and examined closely. So, as mentioned earlier, we chose a resolution that is as small (c 60 microns) as the human eye can see. This is the 400dpi/ ppi figure. This in turn, as a simple statement of maths, implies that we take 160k image pixels and put them on 1 square inch of paper. So a 16x12” paper holds approximately 31MP
-
Given that their maximum print width is 12 inches, is it safe to assume they are using some sort of minilab (Fuji Frontier, Noritsu etc)?
No. Our machines are completely different to anyone else’s. They are built by us from the ground up. It happens that today we can only do 12 inches. This is, frankly, because our paper handling needs to be so incredibly accurate that we can only currently manage it at 12” width. We want to introduce larger machines but they will probably work on different principles (fixed paper, scanning head) to avoid twisting/ movement of the paper in any way.
-
The LED counts only affect the end user if banding is visible when the machine does not expose and advance precisely. Trust that the machine works.
What is more important and easier to understand is, the machine can (or should as many others already do) print pixel-for-pixel from 400 ppi files. Although their website says one cannot see the individual pixels, I'm sure they are mistaken. All these machines easily print discernible individual pixels when the test is provided. Single red, green, blue pixels on white, gray, and black backgrounds are easy to see with a loupe of 8~12 power. Single pixels on white are to be found on ink jets, not so easy to see on gray or black backgrounds. Pixels subsampling is employed also. Lenses may be used also. Common lenses such as Nikon in the case of Chromira LED printers. Thus a digital enlarger exposing typical color photo print material. Awesome, still.
I am not sure I quite understand this point. If you were to print a single pixel in the middle of a piece of paper you could indeed see it – certainly with a loupe - because it would stand out from the background. Just as you can see a line printed on our machine that is only a single pixel wide. In fact, I’d go further. We can print a single pixel of ANY colour (in our gamut) as a single dot on the page. That dot will be made up of 3 layers of dots in the emulsion, but even under magnification you will only see one dot. Contrast this with an inkjet that must, by definition, place multiple dots of ink of different colours to create one pixel unless that pixel happens to be exactly the colour of one of the inks.
But if you consider an image made up of many pixels, then you cannot see the pixel structure in an L.Type print in any way with the naked eye although you CAN see all of the detail in the image with great sharpness. All you can see under a Loupe is a very faint regular grid pattern – which speaks to the accuracy of the positioning of all the pixels. Under a loupe, an adult will probably find that new detail is revealed because their eyes aren’t quite good enough to resolve it. A younger pair of eyes should be able to see it all, but 400ppi/ 64 microns is pretty much the limit and we see no point in going finer.
-
Yeah, I get that they're printing dots at 400dpi, but I feel like if you're going to include a paragraph on the 576 LEDs then I ought to
be able to work out what the paragraph means, you know?
In a sense the 576 LEDs is a red herring. The figure could be 30,000 LEDs and all that would tell you is the width of the swathe we print at one time. So 576 LEDs = 192 pixels wide (3 colours for each) = 12.2mm wide swathe printed each time. We print, move the paper, print, move the paper, and so on…just like an inkjet.
-
There is a wide range of estimates of how many colours people can see – anywhere from 1 million to 14 million – though I’m not sure we know whether the ones I see are identical to the same number you see. So it is absolutely questionable whether printing above 8-bit levels (which gives 16.8 million combinations) is relevant to any printer, though everyone is trying to beat that.
I've generally heard about 12 million but the point is, it's not anything like 16.7 or billions: IF you can't see it, it's not a color. So the bit (no pun intended) about such massive numbers of device values is, they are all not all colors (assuming we agree upon 12 million and marketing states billions, there's more we can't see than we can by a massive margin). Yet marketing from many, many companies keep trying to get customers to believe more is better. As for billions of device values, I'd be love anyone to demonstrate on a print that has more than 8-bits per color encoding of image data AFTER editing versus the high bit version produces any visual difference on the print. Bit depth is about editing overhead and again, I'd love to see someone demonstrate on a print, even with fine gradations that more than 8-bits per color of a final edited image is inferior to sending a higher bit cousin.
-
Thanks, Huw. That was very informative, I appreciate you taking the time.
-
Indeed. I look forward to trying the service.
-
Quote from: Stephen Ray on February 26, 2018, 08:45:56 PM
The LED counts only affect the end user if banding is visible when the machine does not expose and advance precisely. Trust that the machine works.
What is more important and easier to understand is, the machine can (or should as many others already do) print pixel-for-pixel from 400 ppi files. Although their website says one cannot see the individual pixels, I'm sure they are mistaken. All these machines easily print discernible individual pixels when the test is provided. Single red, green, blue pixels on white, gray, and black backgrounds are easy to see with a loupe of 8~12 power. Single pixels on white are to be found on ink jets, not so easy to see on gray or black backgrounds. Pixels subsampling is employed also. Lenses may be used also. Common lenses such as Nikon in the case of Chromira LED printers. Thus a digital enlarger exposing typical color photo print material. Awesome, still.
I am not sure I quite understand this point. If you were to print a single pixel in the middle of a piece of paper you could indeed see it – certainly with a loupe - because it would stand out from the background.
Yes Huw, my point being exactly as you have described.
Individual pixels are easily discernible because the machines can be that accurate. The small test patches as I have described can verify that and were always a common test for any digital enlarger upon a machine install or after service. Practically all machines could satisfactorily pass the test from edge-to-edge. Lambda, LightJet, Chromira, Politechnica, Fuji, Noritsu, Kodak, are the brands I was familiar with. The machine I have used were high volume work horses that were extremely stable in regards to beam focus. (Years.)
-
I've generally heard about 12 million but the point is, it's not anything like 16.7 or billions: IF you can't see it, it's not a color. So the bit (no pun intended) about such massive numbers of device values is, they are all not all colors (assuming we agree upon 12 million and marketing states billions, there's more we can't see than we can by a massive margin). Yet marketing from many, many companies keep trying to get customers to believe more is better. As for billions of device values, I'd be love anyone to demonstrate on a print that has more than 8-bits per color encoding of image data AFTER editing versus the high bit version produces any visual difference on the print. Bit depth is about editing overhead and again, I'd love to see someone demonstrate on a print, even with fine gradations that more than 8-bits per color of a final edited image is inferior to sending higher a bit cousin.
I completely agree with you in almost all respects. There was a typo in my original response where I said '1million to 14 million' - I mean to type 12 million, so apologies. Let's agree on 12 million discernible colours by a good human eye as a starting point. That is different to saying that the 4 billion theoretical combinations of R,G and B values that you can achieve using 32 bits of data re not all colours. Every single one of those combinations, fed into RGB LEDs and thus into pixels on the paper, produces a visible dot on the page. So all 4 billion combinations produce 4 billion dots. The point, as you say, is that many of those dots will look the same as each other or will not be discernible by the human eye from each other (though I'm still not convinced that the 12 million different colours you see are necessarily the same as the 12 million colours I see).
But imagine a colour whose values in some imaginary, simplified, visible RGB colour space are 1,1,1. Then imagine that the next discernible colour (varying one channel only) is 4,1,1. You and I might both agree that 2,1,1 and 3,1,1 are not discernible from 1,1,1 but they are still colours, and if I replace my two points of 1,1,1 and 4,1,1 with different points of 2,1,1 and 5,1,1 I will produce two perfectly visible (and probably discernible) dots. So one reason for having as many data points as possible is to achieve smooth gradations as you move along a tonal curve, and eliminate any stepping.
Now you are completely correct in saying that 8-bit colour is perfectly adequate in print. In fact, that's all we actually use in terms of our print engine - each of the LEDs is fed with 8-bit colour information from the RIP and that produces 16.8 million visible output combinations (some of which may still be indiscernible to the human eye). But each of our LEDs is controllable to either 11 bits or 10 bits (giving us 32bits in total). So the critical extra step that we undertake is in mapping the 24-bit data we get from an 8-bit RGB colour space to the 32-bit range of allowable inputs to our R,G and B LEDs. The thing is that silver halide paper is non-linear: a 1% increase in energy in, say, the red spectrum will typically not produce a 1% increase in density - and the relationship between energy increase and density increase varies all the way across the range in each of the three channels. It also differs for each paper we use. So having the 2/3 extra bits in each channel allows us to make very fine adjustments to the energy inputs of each channel at critical points in the energy curve in order to get the best possible calibration and colour profile for our printer. So, for example, about 18 months ago we were sent a test file to print by a very prominent camera manufacturer. That file consisted largely of a very pale grey sky that shaded from essentially paper-white to a very slightly darker grey, and had great subtlety throughout the mix of clouds shown. Our early attempts to print that produced a certain lack of smoothness in tonal gradation as we approached the media white point, and an inability to show all the subtlety in the clouds. We spent a good deal of time re-characterising our LED calibration in order to reproduce the correct amount of detail in those low-energy areas of the image (and we have repeated the effort at different points along the curve). I guess one of the key differentiators we would claim is that we are in a position to do this - just like (presumably) the R&D departments of any major printer manufacturer, but unlike the average print service company that has to take the printer it has purchased in the form in which it was provided - whether or not the calibration basis is as good as it should be.
So in summary I think I would say that the point about 4 billion colours could be much better phrased. We aim to produce 16.7 million different colour values on the page (2^24 combinations from 8-bit per channel colour). But to produce these accurately and smoothly, particularly in difficult areas of the gamut, we use the extra 'detail' available from the 32 bits of input data we have available for our LEDs, which gives us 4 billion potential colour combinations. And the mapping of the 24-bit space to the 32-bit space is one of the critical steps in producing the most accurate possible printer.
-
Huw, I'm puzzled by the motivation for your research into refining the digital c-type process. You say that you were driven by a desire to render text better on a c-type, but who wants to print text on a c-type? There are countless other printing options for nice crisp text. I can see that a photographer might occasionally want to put an image caption on the border of a print – but fifteen years research to optimise that? Just curious :)
-
That is different to saying that the 4 billion theoretical combinations of R,G and B values that you can achieve using 32 bits of data re not all colours.
Most, the same colors. :D
So all 4 billion combinations produce 4 billion dots.
That be true if the image had 4 billion pixels all R0/B0/G0 or any one triplet. So what?
I'm still not convinced that the 12 million different colours you see are necessarily the same as the 12 million colours I see).
If you can't see it, it's not a color. Color is a perceptual property.
Now you are completely correct in saying that 8-bit colour is perfectly adequate in print.
Then all that talk of billions of colors device values is marketing. If I can't see the difference, so what?
each of the LEDs is fed with 8-bit colour information from the RIP and that produces 16.8 million visible output combinations
Visible output combinations; you want to consider that again based on previous text?
IF we can't see 16.7 million colors, how can we see 16.7 million visible output combinations? If I make an document for you to print that is 1000x1000 pixels, and every value is R34/B90/B123 is that one color or 1,000,000 visible output combinations let alone that many colors? I would suggest that document contains ONE color; R34/B90/B123. And if that color were instead R0/G255/B0 in ProPhoto RGB, it wouldn't even be a color (it would map to a color at some point after conversion to the print color space).
We aim to produce 16.7 million different colour values on the page (2^24 combinations from 8-bit per channel colour).
That may be your aim, but I suspect that's simply not the case. But I'm open to hear how we can see (an agreed upon value of 12 million colors for this discussion) AND 16.8 million visible output combinations. Further I challenge anyone here to output a document of any kind, after all edits in both 8-bit per color and high bit (what some may call 16-bit encoding even when it usually isn't) to any printer and illustrate the higher bit depth was visible superior. Seems that may not be possible with your device as it sounds like your front end (what you're calling a RIP) fed the LED's 8-bit per color data. If that understanding is correct, then all this talk of high bit encoding producing billions of colors is moot (and wrong), no?
-
And the mapping of the 24-bit space to the 32-bit space is one of the critical steps in producing the most accurate possible printer.
Accuracy in terms of color? I think not. Some other kind of accuracy (placement of dots), perhaps. For lurkers:
Delta-E and color accuracy
In this 7 minute video I'll cover: What is Delta-E and how we use it to evaluate color differences. Color Accuracy: what it really means, how we measure it using ColorThink Pro and BableColor CT&A. This is an edited subset of a video covering RGB working spaces from raw data (sRGB urban legend Part 1).
Low Rez: https://www.youtube.com/watch?v=Jy0BD5aRV9s&feature=youtu.be (https://www.youtube.com/watch?v=Jy0BD5aRV9s&feature=youtu.be)
High Rez: http://digitaldog.net/files/Delta-E%20and%20Color%20Accuracy%20Video.mp4 (http://digitaldog.net/files/Delta-E%20and%20Color%20Accuracy%20Video.mp4)
-
Most, the same colors. :D That be true if the image had 4 billion pixels all R0/B0/G0 or any one triplet. So what? If you can't see it, it's not a color. Color is a perceptual property.
Forgive me, but I think we are talking about cross purposes on the question of whether you can see a colour. Every single colour that can be generated by our printer is, by definition, visible. It may be that 500 near-adjacent RGB combinations all generate pixels that you (or I) cannot tell apart, but they are still all valid visbible colours - just not separable by your or my eyes (I’m not sure it’s been proven that someone else couldn’t be more sensitive in one part of the gamut and less sensitive in another). Obviously there are vast tracts of the LAB colour space that are completely invisible/ theoretical, but so long as we are using 8-bit inputs on 3 channels and printing those on paper, I think we can agree that every possible combination of those RGB values will be visible on the page.
I completely concede the point about 4 billion colours being marketing speak - it is indeed 4 billion input combinations into our LEDs (11 bits into two of the channels and 10 into the other). I will revisit how we present this. I hope you will forgive me, though, on the grounds that every other manufacturer makes at least the equivalent claim (I believe I read a figure into the trillions in the past couple of days from one of them).
-
Further I challenge anyone here to output a document of any kind, after all edits in both 8-bit per color and high bit (what some may call 16-bit encoding even when it usually isn't) to any printer and illustrate the higher bit depth was visible superior. Seems that may not be possible with your device as it sounds like your front end (what you're calling a RIP) fed the LED's 8-bit per color data. If that understanding is correct, then all this talk of high bit encoding producing billions of colors is moot (and wrong), no?
Again, I concede the moot point. Marketing speak re 4 billion colours is just that. 4 billion possible LED input values are fed / mapped to 16 million output values from the RIP. In other words, we take 8 bit data from Photoshop et al, and map those 8 bits to a small portion of the 4 billion input values to the LEDs to try to make the LEDs generate the most accurate possible colours. We do not believe that 16-bit/ channel inputs add anything to the accuracy of printing and agree that no one could tell the difference. On the other hand, having the full 32bits to play with in the LED inputs definitely gives us extra flexibility and accuracy in the calibration of the machine.
-
Accuracy in terms of color? I think not. Some other kind of accuracy (placement of dots), perhaps. For lurkers:
Delta-E and color accuracy
In this 7 minute video I'll cover: What is Delta-E and how we use it to evaluate color differences. Color Accuracy: what it really means, how we measure it using ColorThink Pro and BableColor CT&A. This is an edited subset of a video covering RGB working spaces from raw data (sRGB urban legend Part 1).
Low Rez: https://www.youtube.com/watch?v=Jy0BD5aRV9s&feature=youtu.be (https://www.youtube.com/watch?v=Jy0BD5aRV9s&feature=youtu.be)
High Rez: http://digitaldog.net/files/Delta-E%20and%20Color%20Accuracy%20Video.mp4 (http://digitaldog.net/files/Delta-E%20and%20Color%20Accuracy%20Video.mp4)
I’m sorry but this is one point on which I must disagree with you. This is nothing to do with placement of dots, and everything to do with colour accuracy. The thing is this - the most difficult thing for any RGB printer (ie one with no K channel) to achieve is gray line neutrality. At every point along the gray line, a balance between the three input channels must be achieved to achieve colour neturality. The thing is the paper behaves entirely non-linearly in terms of the relationship between input energy and colour on the page for each channel along the curve. So at some points, a 1% increase in input energy into the relevant LED produces a 1% increase in intensity of the image. At other points, the 1% increase in energy input might produce a much greater (or lesser) increase in intensity of the printed image. If we only had 8 bits of input into each LED channel then we would find it much harder to achieve grey balance at different points on the density curve. Having 2 or 3 more bits on each channel effectively gives us more sensitivity where we most need it to achieve the balance as well as we can. We take a number of points all along that gray line to calibrate the machine, but they are not evenly spaced - in some areas they are very close together; in other areas much farther apart. The 32 bits of input data to the LEDs are therefore crucial.
-
Forgive me, but I think we are talking about cross purposes on the question of whether you can see a colour. Every single colour that can be generated by our printer is, by definition, visible.
Then it can't possibly print more than 12 million colors by your own agreement as to what we can see!
Let's examine your own text, two examples below:
I mean to type 12 million, so apologies. Let's agree on 12 million discernible colours by a good human eye as a starting point.
We agree and now the important bits <g>:
Then imagine that the next discernible colour (varying one channel only) is 4,1,1.
It then is the next discernible color, period.
You and I might both agree that 2,1,1 and 3,1,1 are not discernible from 1,1,1 but they are still colours, (AR:the same colors, not disernabile as two different colors) and if I replace my two points of 1,1,1 and 4,1,1 with different points of 2,1,1 and 5,1,1 I will produce two perfectly visible (and probably discernible) dots.
Visible as the same color dots. They are the same color if they are not visibly discernible. 1/1/1 is either discernible from 2/1/1 or it isn't. Dividing this into 2.23/1.16/1.19 (finer encoding) doesn't make that now more or less discernible; it either is or it isn't discernible; perceptually by the human observer.
So one reason for having as many data points as possible is to achieve smooth gradations as you move along a tonal curve, and eliminate any stepping.
IF it can be seen, it can be seen! Two non discernible numbers are one color and no, it's not smoother IF they are not discernible from each other. Below are two differing sRGB device values. 1/255/240 and 2/255/240. They are the same color because they appear identical (as the deltaE values show the differences are invisible by a huge margin). Doesn't matter if this sRGB document with two values is encoded in 8-bits per color, 16-bits per color or a finer encoding (1.01/255.0.0/240.1.2 vs 2.01/255.00/240.1.2); the two numbers ARE the same color. Print them anyway you wish, they will appear as the same color no matter the encoding precision. Place each next to each other on a print, within a gradient; they appear as the same color. Two sets of device values are either visibly different (on your device or any other) or they are not. With the sRGB values below, no device on this planet will produce two differing visible colors without a bug; the two values are a dE of 0.01! NO encoding will change that!
Analogy: I have a 3lb apple pie baked into a nine inch dish. I can divided that pie into 8 pieces or 16 pieces or a billion pieces. It's still a 3lb nine inch pie. No matter how I divide it up, this doesn't alter it's size and weight. Same with color numbers. More numbers divided up doesn't produce more colors. Two values, two hundred values can all be the same color perceptually. You make a print with those numbers and they all appear the same to the observer.
(http://www.digitaldog.net/files/ColorNumbersNotColors.jpg)
-
Huw, I'm puzzled by the motivation for your research into refining the digital c-type process. You say that you were driven by a desire to render text better on a c-type, but who wants to print text on a c-type? There are countless other printing options for nice crisp text. I can see that a photographer might occasionally want to put an image caption on the border of a print – but fifteen years research to optimise that? Just curious :)
The honest answer is that what looked like a brilliant idea almost 20 years ago might look a little different now (and I have only been involved for two of them!). There was a time when digital images (or scans) were still being projected onto negatives which were then developed as normal (our founder had one such business) and when inkjet first came in it was a tremendously expensive process in many ways, when silver halide was a well-known and relatively cheap medium. But silver halide produced soft text for the simple reason that it was mostly imaged inexactly (e.g. often through an enlarger lens) - so people running silver halide based businesses could see the threat from inkjet, etc, but couldn’t work out how to address the really obvious disadvantage - the need to print sharp text and graphics like inkjet could. Over the years, obviously, inkjet has made amazing advances, particularly in photo printing. Silver halide printing has not advanced nearly as much and most machine manufacturers have given up for good - or concentrate entirely on high-volume/ moderate accuracy machines. We still think there is a place for really accurately imaged silver halide and we believe that the technology developed over the past 15+ years probably takes things almost as far as it is possible to do (though we do have a wide format high-speed print concept in the very early stages of development). The thing is that silver halide is still a really beautiful medium and it has a feel all of its own. Some people love it; others prefer inkjet - I think there’s a place for both in the world, even for the most demanding photographers. Incidentally, I saw the most stunning inkjet print today of really complex architects drawings. I am trying to find what the printer was. I would say that they beat us on the overall quality of line, but our text (even so small that it was completely unreadable without a loupe) was much better - more even, less blotchy, and in the really small details (e.g where multiple lines intersect) our technology resolved the lines at least as well and in some cases better. I will try to post scans over the next few days so you can see what I mean, if that’s of interest.
-
Andrew, you should send a few images in. I sent a ton in and was really surprised. I'll have a video and article on it as soon as I can get it written. They make it easy to order so give it a try and then you can share your findings.
-
I’m sorry but this is one point on which I must disagree with you. This is nothing to do with placement of dots, and everything to do with colour accuracy.
I didn't know if it had anything to do with placement of dot or not because you didn't say color accuracy, only accuracy.
Now that we are on the same page as speaking of actual color accuracy, and considering you've told us the RIP sends 8-bits per color data to the printer, how can a higher encoding have anything to do with color accuracy?
The thing is this - the most difficult thing for any RGB printer (ie one with no K channel) to achieve is gray line neutrality.
And higher bit depth does nothing to address that.
If we only had 8 bits of input into each LED channel then we would find it much harder to achieve grey balance at different points on the density curve.
Now why would that be so? Further, you've stated this to us:
each of the LEDs is fed with 8-bit colour information from the RIP and that produces 16.8 million visible output combinations
So in one breath, you're telling us for better color accuracy you need more bits (makes zero sense to me) and then you tell us the LED is feed 8-bit per color. I send you a 8-bit per color JPEG or TIFF, it is what it is; you're not getting more bits right? AND you're telling us when the rubber meets the road, the LED only gets 8-bits per color anyway.
But you may be able to explain how high bit data and 8-bit per data in any way produces more accurate color. What's the dE difference IF your device dealt with 8-bits vs. more than 8-bits and why would it make any difference?
-
Andrew, you should send a few images in. I sent a ton in and was really surprised.
Moot. I'm surprised by a lack of technical understanding conveyed here about this print process.
Your suggestion is akin to someone saying "We only accept sRGB for our printers and the prints look really good". They may indeed look good, but that tells me nothing about what I may get from the printer if I sent a larger color gamut to the printer and COMPARED it to the sRGB version.
So, does the higher bit depth produce more accurate color as we're told and IF so how?
Does the printer produce billions of colors when some here are under the agreement we can only see 12 million?
-
My guess is that they've got 11 bits on 2 of the LEDs and 10 on the other, and the device values they present to the device produce specific electrical values at the LED and that's that. We could imagine that maybe it's linear or whatever. Point is, there's some sort of mapping from 32 bit device values to the characteristics of the dot on paper, and I bet that mapping is fixed. I assume that 0,0,0 is a damn good imitation of the paper's DMax, and actually I bet there's a whole lot of device value space that smushes down to the paper Dmax. Ditto 2048,2048,1024 being paper white.
Since they're got an enormous amount of room in that device's space, even if it is a fixed mapping and maybe not even a very nice mapping, they can still fit whatever mapping you like onto it.
So, among other things, they can manage a perfectly neutral middle grey.
Who knows, maybe they can custom map each of the 576 LEDs to compensate for variability in the LEDs themselves. With that much room to play with, why not?
-
My guess is that they've got 11 bits on 2 of the LEDs and 10 on the other, and the device values they present to the device produce specific electrical values at the LED and that's that.
Bits from what/where? I send them a document to print that's 24-bit color. Now what; interpolating more bits? I do the same in Photoshop (convert 24-bit data to 48-bits); what did I gain?
-
From anywhere!
I am speculating, of course. But I assume the machine takes 32 bits of stuff and turns it into a little teeny dot, in a pretty fixed way. Further back up the pipeline,
though, they have some choices as to which 32 bits they want to send.
My knowledge of the exact terminology of color science and printing is very very close to zero, but as a former mathematician, engineer, and all around not-quite-idiot, I can imagine that you have, I dunno, a 24 bit value in your picture file which, along with your color space indications and... some other metadata?... means some color. That 24 bit value gets mapped around in the print pipeline to produce some sort of value or set of values that corresponds to physical blotches of ink, or in this case, a 32 bit "MAKE UM LEDS SO BRIGHT OK THX" value.
By having a huge numerical space with massive amounts of redundancy in the machine, at that last stage, they give themselves room to work, and (presumably) rarely find themselves in a situation where the input specifies a certain color, the paper can actually RENDER that color, but precision has been lost and bits thrown away and, bugger it, it's still not going to happen.
So, anyways. I get the value in having loads of "digital space" at the print head. If it costs nothing to build it in, it's going to keep options open down the road.
-
.................and in the really small details (e.g where multiple lines intersect) our technology resolved the lines at least as well and in some cases better. I will try to post scans over the next few days so you can see what I mean, if that’s of interest.
Yes Huw, it is of interest, and thanks ever so much for hanging in with this discussion. I think we are all learning from it.
-
Andrew, you should send a few images in. I sent a ton in and was really surprised. I'll have a video and article on it as soon as I can get it written. They make it easy to order so give it a try and then you can share your findings.
Really good idea. As my article shows, this is a case where the numbers and the math can take you only so far - you need to examine real photographic results side by side to get a feel for what's going on. I would recommend particularly to Andrew - sending them several photos that have nicely graduated skies. Print them as you would conventionally in your Epson 3880 or whatever printer you are now using. Then reprep the files softproofing with their profiles to emulate your Epson softproof as closely as the gamut difference allows, get them to 400 PPI and send them in. It could also be of interest to send them a couple of B&Ws with nice tonal gradations. Remember there is a size limitation of about 12" width.
-
Huw, while we have you here, can you talk a little about the books?
It looks to me like you're printing panoramic prints on Fuji paper, and then folding and gluing them with Imaging Solutions equipment, correct?
This, essentially, take a wide C print and folds it in half, with printed surface facing printed surface. In my experience, photographic prints are
not terribly happy face-to-face, especially if even a small amount of dampness is involved. I actually hand built some layflat books using actual prints,
and the pages stuck together like the dickens. I tipped in tissue to cope with the problem (I was using two prints, one verso and one recto
so this was feasible).
Do you use a coating or something to allay these problems? Or are the specific papers you've specced for the process particularly well suited to this
face-to-face application?
-
Really good idea.
How so? How would it answer a single question I have asked about color numbers and color output of this device?
I have both a P800 and a 3880. AFAIK, and please correct me if I'm wrong, both printers produce a wider color gamut than Lumejet. Both provide far more options for papers. Both provide a print that's far more archival. Both can produce a much larger print. What would I gain by having output from the process? How would it answer my question about higher bit depth producing more color accuracy? Would having them produce prints clear up what in my mind is text which seems at odds:
hrwilliams wrote:
"...each of the LEDs is fed with 8-bit colour information from the RIP and that produces 16.8 million visible output combinations"
"I mean to type 12 million, so apologies. Let's agree on 12 million discernible colours by a good human eye as a starting point".
Which is it (which value is visibly discernible)?
Somewhere, 4.8 MILLION colors (I have stated, device values, NOT colors) have been gone a-missing if we agree to agree on that 2nd sentence just above.
We can skip some nitpickng that this isn't a function of the eye per se!
Peer review is a bitch. But it's necessary on a site which asks us to pay for content. Even if the payment is tiny and well worthwhile mostly. Let's not forget the very first post here which I happen to be finding to be accurate thus far:
http://forum.luminous-landscape.com/index.php?topic=123382.msg1028733#msg1028733 (http://forum.luminous-landscape.com/index.php?topic=123382.msg1028733#msg1028733)
But I am open minded and want to separate the facts from the marketing hype of which the later is a possibility thus far.....
-
Moot. I'm surprised by a lack of technical understanding conveyed here about this print process.
Your suggestion is akin to someone saying "We only accept sRGB for our printers and the prints look really good". They may indeed look good, but that tells me nothing about what I may get from the printer if I sent a larger color gamut to the printer and COMPARED it to the sRGB version.
So, does the higher bit depth produce more accurate color as we're told and IF so how?
Does the printer produce billions of colors when some here are under the agreement we can only see 12 million?
Andrew,
Huw was clear about the gamut constraints of their process - as was I in the review - that compared say to the Epson SC-P5000 I'm using, they are indeed gamut limited. I published a number of gamut comparisons for output options, but not working spaces. Attached is one I had prepared comparing Adobe RGB(1998) with Lumejet Luster paper. Clearly, for photos with colours that would fill or even exceed ARGB(98), there's going to be a lot of gamut compression printing those photos through the Lumejet process. So one is in the hands of Rendering Intents and good profiles to see how all that works. This is where having a range of comparison prints made really helps appreciate relative image acceptability/quality.
Colour spaces are a continuum. In a manner of speaking they get populated with encoded colour values to go to the printer. No matter what the bit depth, any values that exceed the printer/paper gamut are going to be compressed. Higher bit depth means slicing and dicing the colours into smaller differences between them (your three pound apple pie with more slices). At what point do you stop seeing differences and therefore at what point does this matter to the quality of tonal gradations? Again, need to see results.
-
Andrew,
Huw was clear about the gamut constraints of their process - as was I in the review - that compared say to the Epson SC-P5000 I'm using, they are indeed gamut limited. I published a number of gamut comparisons for output options, but not working spaces. Attached is one I had prepared comparing Adobe RGB(1998) with Lumejet Luster paper. Clearly, ifor photos with colours that would fill or even exceed ARGB(98), there's going to be a lot of gamut compression printing those photos through the Lumejet process. So one is in the hands of Rendering Intents and good profiles to see how all that works. This is where having a range of comparison prints made really helps appreciate relative image acceptability/quality.
What isn't yet clear to me is why I'd send images to him and how it would answer my questions.
Colour spaces are a continuum.
Meaning what?
In a manner of speaking they get populated with encoded colour values to go to the printer. No matter what the bit depth, any values that exceed the printer/paper gamut are going to be compressed. Higher bit depth means slicing and dicing the colours into smaller differences between them (your three pound apple pie with more slices). At what point do you stop seeing differences and therefore at what point does this matter to the quality of tonal gradations? Again, need to see results.
I agree and have stated this already. How would that address the discrepancy in his text about describable colors? Is my post (last one #79) answered by sending files to him for output?
-
Huw and you, in post 67 have already agreed on a number of 12 million for discernible colours. So that one is for now settled. The remaining question about how many colours are needed for smooth tonal gradations in whatever gamut size is still on the table. I wonder if there is a mathematical answer to it, or is it primarily empirical and image-dependent.
What you get having some prints made is giving a tangible perspective to all the math issues.
I think Huw was clear that this process is for people who don't want to spend their time making their own prints with inkjet printers on their desks, but prefer to outsource to a high quality service provider. I think it's pretty obvious that people who need prints larger than what they provide, and/or assured greater longevity than they can provide, and/or uncompressed colours beyond their gamut potential would send their printing elsewhere, but those for whom those constraints don't matter would be very pleased with the quality of the output. I don't think any one is pretending that this is be-all and end-all of printing processes. It is a high quality option for a certain clientele - it's a wide world full of people with different requirements.
-
Honestly, if they can build a genuine book block with real C prints that don't stick together, that would be something worth knowing about.
It's a big world and there are many things in it, but books and photographic prints are not, as far as I know, two great tastes that have been
put together particularly well in the past.
-
Huw and you, in post 67 have already agreed on a number of 12 million for discernible colours. So that one is for now settled. The remaining question about how many colours are needed for smooth tonal gradations in whatever gamut size is still on the table. I wonder if there is a mathematical answer to it, or is it primarily empirical and image-dependent.
Consider this set of images; I believe indeed it's image dependent and Huw has mentioned smooth gradients:
Huw was indeed clear that this process isn't for someone like me with two Epson printers which makes me wonder why the suggestion in post #70 and then #77 was recommend I do so.
-
I didn't know if it had anything to do with placement of dot or not because you didn't say color accuracy, only accuracy.
Now that we are on the same page as speaking of actual color accuracy, and considering you've told us the RIP sends 8-bits per color data to the printer, how can a higher encoding have anything to do with color accuracy? And higher bit depth does nothing to address that. Now why would that be so? Further, you've stated this to us:So in one breath, you're telling us for better color accuracy you need more bits (makes zero sense to me) and then you tell us the LED is feed 8-bit per color. I send you a 8-bit per color JPEG or TIFF, it is what it is; you're not getting more bits right? AND you're telling us when the rubber meets the road, the LED only gets 8-bits per color anyway.
But you may be able to explain how high bit data and 8-bit per data in any way produces more accurate color. What's the dE difference IF your device dealt with 8-bits vs. more than 8-bits and why would it make any difference?
This is cut and pasted from my old Kodak LVT manual...
<<Each pixel in an image is represented by three 8-bit bytes or 24-bits of information. Each 8-bit byte represents a color in each pixel. When an image is printed, the 8-bit code is used to “look up” a corresponding 12-bit code stored in a calibrated look-up-table. There are 4,096 12-bit codes available, but only 256 are used at a time for each color input. The 12-bit code is then fed into a Digital-to-Analog Converter which converts the 12-bit information to a corresponding voltage. This voltage controls the modulators that regulate the exposure of the photographic material on the drum.>>
<<For calibration purposes, the exposure time is fixed and density is a function of the light available to the photographic material (represented by the 8-bit code). The aim values that are used are based on the concept of “equal lightness” for Ektachrome and color paper material. This concept allows for a uniform sampling of the 3D color space of the photographic material. For all other materials, the aims are set up to produce a facsimile of the Ektachrome output of the same file.>>
Maybe this helps the conversation.
-
They are the same color because they appear identical (as the deltaE values show the differences are invisible by a huge margin). Doesn't matter if this sRGB document with two values is encoded in 8-bits per color, 16-bits per color or a finer encoding (1.01/255.0.0/240.1.2 vs 2.01/255.00/240.1.2); the two numbers ARE the same color. Print them anyway you wish, they will appear as the same color no matter the encoding precision. Place each next to each other on a print, within a gradient; they appear as the same color. Two sets of device values are either visibly different (on your device or any other) or they are not. With the sRGB values below, no device on this planet will produce two differing visible colors without a bug; the two values are a dE of 0.01! NO encoding will change that!
(http://www.digitaldog.net/files/ColorNumbersNotColors.jpg)
I’m sorry - owing to time differences I have been absent when all the interesting stuff was going on.
I made a comment earlier in this thread agreeing to agree on 12 million discernible colours. Please don’t hold me to that in the sense that none of us knows whether the figure is 12 or 15 or 16 million and my guess is that it varies between individuals. A master of wine, for instance, can discern many more scents than I can. Can we agree that it is in the low double-digits but not fight over whether it’s 12 or 16 million for a few minutes? I absolutely agree that the concept of 4 billion colours is moot - that’s over 2 orders of magnitude greater than the range of discernible colours that a human can see.
However in the interests of allowing me to learn from this, please let me post a question (genuinely) to try to clarify my understanding.
Imagine I paint a gradient by varying only one channel all the way from 0 - 256, while not varying the other two input channels. Without having done the maths, presumably any two adjacent points will have Lab differences that are so small that it is very hard to tell the difference between them. Or at least it must be possible to paint gradients that have this property at some points on the gradient. So if we imagine a section of that gradient where there are say n points. Point 1 has a given value. Point 2 has one channel changed by 1 value and the other two remain the same, and so on up to point n which is the first point that is discernibly different to the eye from point 1. Then my question (and this is a genuine one) is ‘what colour is point n-1’? It is much closer to point n than it is to point 1. So Lab should tell me that it is indistinguishable from point n, which is completely adjacent to it. And yet it is also apparently indistinguishable from point 1, which itself IS distinguishable from point n. How does that work? To me the fact that I can’t tell the difference between two adjacent points doesn’t mean they aren’t different colours - it just means that they aren’t distinguishable by my eyes. Is it not possible that actually someone else might see the differences between points slightly differently? To take a slightly facetious analogy: try alternating sips of whiskey and water. Pretty soon you can’t tell the difference. You can see it; you know it’s there; but you can’t taste it because your brain desensitises. Does the same thing happen with colour?
At this point I’m sure it’s obvious I’m not a colour scientist - for which I apologise. I’m a mechanical/ manufacturing engineer, so I have some ability to understand technical arguments, but my career was in finance. I’m trying hard to catch up!
-
Lets start at square one so I can put this all into perspective in one post (much of this is redundant but it seems necessary).
Ground rule #1
Color, is a perceptual property. So if you can't see it it's not a color. Color is not a particular wavelength of light. It is a cognitive perception, the excitation of photoreceptors followed by retinal processing and ending in the our visual cortex, within our brains. As such, colors are defined based on perceptual experiments. I've placed text from experts to back this up:
Fairchild's "Color Appearance Models". Page 1!
Like beauty, color is in the eye of the beholder. For as long as human scientific inquiry has been recorded, the nature of color perception has been a topic of great interest. Despite tremendous evolution of technology,fundamental issues of color perception remain unanswered. Many scientific attempts to explain color rely purely on the physical nature of light and objects. However, without the human observer, there is no color.
Further on the same page:
It is common to say that certain wavelengths of light, or certain objects are a give color. This is an attempt to relegate color to the purely physical domain. It is more correct to state those stimuli are perceived to be a certain color when viewed under specific conditions.
Page 1 paragraph 2 of Digital Color Management by Giorgianni and Madden:
But color itself is a perception and perceptions only exist in the mind.
Page 11 of The GATF Practical guide to Color Management:
Although extensive research has been conducted, we still not completely understand what happens in the brain when we "see" color. The visual sensation known as color occurs when light excites photoreceptors in the eye called cone cells.
Page 75 of Understanding Color Management by Sharma:
Color is an impression that we form in our brains.
You write: "To me the fact that I can’t tell the difference between two adjacent points doesn’t mean they aren’t different colours - it just means that they aren’t distinguishable by my eyes".
No it measns they are the same color! Repeat: IF they are not distinguishable by your eyes, they ARE the same color. Unless you do not wish to believe what color is, based on the color scientsts above, based on the colorimetry your company uses (deltaE) and which Mark provided in his article. I believe those color scientists. And when I view the two sRGB values on-screen, on a very good reference display, they appear the same color. Because again, do not confuse a color number with a color you cannot see or in this case, two numbers that appear to be the same color and who's dE PROVE it.
Ground rule #2. Color numbers, device values are not necessarily and often not colors! I provided one such example with R0/G255/B0 in ProPhoto RGB. It isn't a color, we can't see it, it falls outside human vision. We can give numbers to wavelengths in the infrared spectrum and we still can't see them; they are in visible to us. They are not colors. In my PDF (URL posted already), the last and critical sentence I wrote was this: Don’t confuse a color number, a device value, as a color you can see!
As to the number of colors we can see (not numbers based on encoding of digital values), we seem to agree that 12 million is a good one. It could be a bit more or less but for this discussion 12 millon on the nose is fine. What it isn't is 16.7million let alone billons of colors! Billions of numbers perhaps but not billions of colors. That's marketing speak from many companies that wish to convince their customers that more is better. You even stated this was a goal and a difficult one but really, it's not at all important in terms of color (what we can see). It's just dividing up that pie into finer slices. The 16.7 million number we hear so often is just math based on computer encoding of numbers (3 to the 8th) so why it has no direct relationship to colors (again, what we humans can see) should be obvious now. Just as the number of miles between Santa Fe and New York has no relationship to how many miles per gallon my car would get driving there!
Huw writes: "Imagine I paint a gradient by varying only one channel all the way from 0 - 256, while not varying the other two input channels. Without having done the maths, presumably any two adjacent points will have Lab differences that are so small that it is very hard to tell the difference between them".
The adjacent points can and DO vary over color space! I provided an example of two triplets in sRGB where the only difference was one value in one channel and the color difference (distance) was a tiny dE 2000 of 0.01; an invisible difference. But I can take other values in sRGB that vary by only 1 value in one channel and the dE is larger. More than 1? That would require a lot of comparisons of triplets in the software I've used to show you the two sRGB values already. If you consider the 12 million visible colors versus the 16.7 million numbers, there's 4.7 million numbers missing here which should tell you a lot about numbers that are not colors.
So this varies. But what is important are these facts:
1. Your device, no device, produces 16.7 million colors. Your device and many devices can produce 16.7 million of the same colors.Your output device, everyone's output device will produce ONE color from TWO of the sRGB triplets I provided. No matter the encoding. Because they ARE the same color.
2. Those who state their devices produce that billions of colors let alone 16.7 million colors are simply using marketing speak in an attempt to make others believe more is better.
3. When the differences in two numbers is less than a dE of one, they are the same color. Doesn't matter how you divide up those two sets of triplets. Finer encoding doesn't change the dE values when the two are perceptually the same color.
Yes, you and I may see colors differently and the 12 million value can differ but ALL this colorimetry is based on the Standard Observer (a theoretical model of human vision). And even accounting for our differences in color perception, it's moot and we simply should not associate the number of colors we (or the standard observer) can see with the encoding of computer numbers! Just like distance in miles and miles per gallon if I can use that quick analogy.
Now 98% of your customers and 99% of other companies customers may believe that these devices can produce 16.7 million or billions of colors. That's simply untrue and wrong! And I don't expect marketing folks to correct this. They didn't when providing spec's for camera resolution by including pixels that were not used to create an image. Or when they provided Dynamic Range of scanners without telling us when they start measuring past a level that's noise. What I expect and hope for, are educated readers who can see through the marketing hype because they actually understand (in this case) the color science behind the claims. I'm not a color scientist, I'm a photographer by training and degree. But nothing I've written requires any formal training in color science. It does help to know some actual color scientists and learn from them and provide them with articles (like my PDF on color numbers) for peer review. So, the above and below from me is in direct reply to post #37 where this question came up:
"4 billion unique colors possible for each printed pixel-Is this even meaningful"?
No, and it isn't possible. Now the reasons why have been outlined.
Now if you wish, we can go into this idea around color accuracy. Let me suggest before hand, we need reference and measured values and we need more than just grays although I totally agree gray's for this task are absolutely necessary to view.
-
My guess is that they've got 11 bits on 2 of the LEDs and 10 on the other, and the device values they present to the device produce specific electrical values at the LED and that's that. We could imagine that maybe it's linear or whatever. Point is, there's some sort of mapping from 32 bit device values to the characteristics of the dot on paper, and I bet that mapping is fixed. I assume that 0,0,0 is a damn good imitation of the paper's DMax, and actually I bet there's a whole lot of device value space that smushes down to the paper Dmax. Ditto 2048,2048,1024 being paper white.
Since they're got an enormous amount of room in that device's space, even if it is a fixed mapping and maybe not even a very nice mapping, they can still fit whatever mapping you like onto it.
So, among other things, they can manage a perfectly neutral middle grey.
Who knows, maybe they can custom map each of the 576 LEDs to compensate for variability in the LEDs themselves. With that much room to play with, why not?
I think this is a pretty good summary of what I was trying to convey.
To take the blocks/ process steps:
We take 8-bit (24-bit) colour data, along with the data’s colour space (.icc) and pass it into our RIP (ColorGATE).The RIP resamples the data within the file and converts the input 24-bit colour with its colour profile into the CIELAB profile connection space (PCS). The output device’s profile (the media specific.icc for our printer) is then used along with the selected rendering intent is used to convert from the PCS to the device specific RGB values in 24-bit colour. The result is a 24-bit colour (.bmp) at 400dpi.
The embedded printer driver convert from (.bmp) to print-specific data for driving the LEDs. The 24-bit pixel colour is channelised into its 8-bit composite data. On a per channel basis (R,G,B), the media specific calibration data is applied. In simplistic terms, this calibration data is a lookup table that has 0-255 on its input, that selects a 10/11 bit calibrated output value per input value. E.g. Input R=250 could lookup an output value of 967(10bit) or 1897(11bit). These 10/11 bits per channel are then re-combined to form the 32-bit time pulse exposure value for that pixel.
Why do we need all these bits? Well, the answer is that you, as a photographer, want a linear CIELAB response in your print. If 0,0,0 is paper white, and 256, 256, 256 is DMax, then 128,128,128 should be half-way there, and 64,64,64 should be 25% density and so on.
As I said in an earlier post, one of the problems that we have to contend with is that AgX photo paper is non-linear. Sometimes, a 1% increase in energy from the LED will generate a 1% increase in density of the colour on the paper. At other times, the ratio is completely different. It varies all along the curve, with some areas being particularly sensitive and others particularly insensitive. This variation is also subtly different between the three channels.
Finally, achieving a neutral gray at different points involves mixing different ratios of C, M and Y dyes on the paper. As I said, as a user, you want a linear increase in the RGB values in your image file to produce a linear response (As measured in CIELAB with a D50 illuminant) on the paper - ie if input values imply a doubling of density, that’s what you want to see on the paper. So we need to map linear data from the image file to non-linear inputs to the LEDs in order to achieve the correct linear response on the paper (as far as possible).
Having the extra bits in the LED control levels allows us to achieve more precise control of the LEDs to achieve the best possible grey. This is not about using the extra bits to achieve a different colour; it’s about using that extra sensitivity to fine tune the relationship between the 3 channels at each point in the curve to achieve the most neutral possible gray. Understanding how each different paper reacts all along the gray line is what enables us to build our target files for each paper which is (as I said in one of my first posts) something that not even Fuji appears to have done. Our target files are designed to deliver a linear neutral grey when measured in CIELAB with a D50 illuminant.
Once we have the correct mapping to produce a neutral grey line that generates linear outputs from linear RGB 8-bit inputs we can extrapolate that to the rest of the colour space. So you have a look-up table that takes 8-bit input values and generates outputs in 10-bit or 11-bit for the LEDs for the whole of the paper gamut. Again, the key is to turn linear 8-bit inputs into linear 8-bit outputs on the printed page, but to do this you have to have non-linear mapping to 10/11-bits for the LEDs and the relationship between the three channels will not always be what you expect.
There are a number of factors that complicate this still further:
First, as many will know, where is even a point on the paper where if you pump too much energy in, it will actually become lighter (‘solarisation’). The response of the paper at this point is unpredictable. Fuji Professional paper’s limit point is, I think, a DMax of 2.5. Above this, you start to get solarisation. So understanding the DMax for the paper is fundamental to the calibration.
Solarisation info for anyone interested:
https://books.google.co.uk/books?id=WEkzDwAAQBAJ&pg=PA126&lpg=PA126&dq=solarisation+silver+halide&source=bl&ots=qvKPEdlg6-&sig=X3i9EifRJOOi6RvC8aF4Tw1Xzmg&hl=en&sa=X&ved=0ahUKEwjBisro09fZAhWHOsAKHZTeBWEQ6AEIUDAH#v=onepage&q=solarisation%20silver%20halide&f=false
Next, there is the question of cross-talk. In order to activate yellow dyes in the paper, you pump in blue light. However the magenta dye is also very slightly sensitive to this blue light and a tiny amount is also activated. At first, this is unnoticeable, but as you approach DMax in the blue/ yellow channel, the rate of yellow density increase begins to roll off. At the same time, the magenta dye has been activated and starts to become much more sensitive in its response to the blue light and more of this magenta leaks out. This is a known ‘feature’ of Fuji papers and results in a slightly orangey tone to very deep yellows. We have spent a good deal of time trying to minimise this.
Then there is the matter of the LEDs themselves. They are all slightly different and they all have slightly different power outputs. We need to account for that - for a given time pulse, each LED will give out different energy. They each also vary slightly in output wavelength and we have to deal with that, too. So we do control every LED individually and this is a crucial part of the calibration process.
So to summarise, the fact that we have 32 bits to play with in terms of LED input values is not aimed at producing more than 2^24 output combinations (ie 8-bit). It is all about mapping 2^24 input RGB values to 2^32 LED input values to allow those LEDs to generate 2^24 output values so that what you see on the paper is what you expect to see from the screen. We are using the fact that we have finer grain control in two areas: (i) in the gray balance; and (ii) in the LED balance to ensure that each LED is balanced relative to its peers. We need the extra bits to do this accurately.
-
Honestly, if they can build a genuine book block with real C prints that don't stick together, that would be something worth knowing about.
It's a big world and there are many things in it, but books and photographic prints are not, as far as I know, two great tastes that have been
put together particularly well in the past.
The answer to that is that we can build you a genuine book block with real C prints that won’t stick together provided you are careful with them. They WILL stick together if they get wet, including a splash. If that happens, you need to dab it off and leave the book to dry, open on that page. It shouldn’t be a problem in normal use. Fuji does have some papers with fingerprint-resistant coatings (the HDX Premium range) which we use. These are also less prone to sticking, I believe, but the coating does definitely reduce colour vibrancy and gamut so better to use where print quality is not at an absolute premium.
-
(I don’t know if the following should be the start of a new thread or not, so I took the easy path and posted as a continuation).
I placed a couple of orders for L-Type single prints with Lumejet and received my orders in California via the post office with each one arriving in just about 12 calendar days (pretty darn good considering the international distance). They were for color prints on Fuji DPII matte and luster papers (they have other papers; and the sample softcover true-photo book looked excellent). I had some of the prints done via their recommended no-extra-cost mounting on acid-free board and they are really great.
I am very pleased with the quality of the prints and the very reasonable cost. The color balance and contrast of the prints are close to viewing soft-proofing in Lightroom via my calibrated NEC monitor, and the presentation of fine details is superior. I was also very pleased with the prompt, friendly, and helpful personal responses to my questions.
With my best quality high-res files printed on the matte paper (at 400 PPI) I was astounded at the quality in the following sense. I can now finally obtain a digitally produced print showing the finest details but yet looking smooth and natural, no matter how closely I look at the print. I have not yet tweaked my sharpening to perhaps get the most out of this particular process, but even so the results are superior to what I’ve been able to obtain elsewhere.
The L-Type prints on matte paper give me a look that is very similar to what I used to get from analog contact and low-enlargement prints on N-surface paper from large format (4x5” and 8x10”) color negs - seemingly almost infinite detail presented in a beautifully natural / smooth manner. I can now get prints that are better than ever before, thanks to digital file post-processing techniques combined with the L-Type printing process.
I did not test color gamut, nor did I directly compare to inkjet prints, as I don’t have such a printer and when I tried a couple commercial inkjet labs at least a couple of years ago I didn’t see a paper (even the smooth ones) that I liked for most of my purposes, and there were other things I didn’t like about the prints although some were very nice in many respects. The one exception was when I had large photos of paintings printed on Hahnemuhle William Turner paper - they came out great and the textured paper provided a nice look.
Time and technology marches on and perhaps there are now papers and inkjet printers available at some commercial labs that would be able to provide my preferred print look. However, even assuming optimistically that is the case my current research indicates that the prints would be much more expensive (by multiples of the L-Type print prices for most sizes).
I don’t have prints made that often, but when I do I will be using Lumejet for a significant percentage of them. Thanks again to Mark Segal and Luminous Landscape for the original article.
Dan
-
The L-Type prints on matte paper give me a look that is very similar to what I used to get from analog contact and low-enlargement prints on N-surface paper from large format (4x5” and 8x10”) color negs - seemingly almost infinite detail presented in a beautifully natural / smooth manner.
Very interesting. Thanks for the review.
-
........... Thanks again to Mark Segal and Luminous Landscape for the original article.
Dan
You are welcome. I'm pleased you are happy with their work. That's the proof of the pudding.
-
Something new:
https://files.constantcontact.com/0d14437c001/964e1921-5941-4425-b106-20c00acafe2f.pdf
LaserLab 127x254 - 50"x100"
World’s biggest integrated digital laboratory
Speed up to 57 square meters/hour
Complete color correction management
Wide range of graphic formats supported
Preset for network printing
Exposes at maximum size in just 60 seconds!
Set up for network printing.
Also available: high resolution option up to 610dpi (optical - true).
-
Looks impressive and reads impressively; would be interesting to link up with a business that invests in one of these machines for a look-see at what it can really do.
-
From the Lumejet web site:
Sorry, we have closed down.
Unfortunately, the journey of L.Type is over.
Please browse the website for its visual content only, not with the intention of ordering. Sorry for any disappointment caused.