Gamuts of Epson 3880 vs Digital-C lab prints?

[Redcrown]

My seldom used Epson 3880 needs about $200 for ink and a maintenance cart, so I'm thinking of giving up home printing and going back to labs. I just don't do enough printing anymore to justify maintaining the Epson.

I've looked at several on-line labs, downloaded their profiles for comparison and testing. I used the on-line profile mapper at http://www.iccview.de/ to compare. A couple of questions come up.

The Epson profile for Epson Premium Luster paper on the Epson 3880 shows a larger gamut than any of the on-line lab profiles show for Fuji Crystal Archive or Kodak Endura. That surprised me. Is it reasonable that injet prints can have a wider gamut than Digital-C prints?

Most labs tell you what kind of paper they use, but some don't say what printer they use (Fuji Frontier, Noritsu, Durst?). Does the printer hardware make any difference if the same paper is used? Comparing profiles from different labs for Fuji Crystal Archive paper I see significant differences in the 3D gamut maps. Wondering why that is?

[Ernst Dinkla]

Yes, inkjet prints can have a wider gamut than C-prints depending on the ink system used, more than 4 inks and up to 11 inks. They can also be more neutral for B&W than C-prints can be. They also last longer as fade resistance is higher, pigment inkjet inks in the first place, modern OEM dye inks following. Price per square foot is higher however, even dry minilab systems (dye inkjet) can not compete with chromogene minilabs on price.

I would not go by the ICC profiles labs show but compare some test print runs of images you know well. Tells more about other conditions like delivery and packaging too.

Met vriendelijke groet, Ernst

http://www.pigment-print.com/spectralplots/spectrumviz_1.htm
December 2014 update, 700+ inkjet media white spectral plots

[digitaldog]

Is it reasonable that injet prints can have a wider gamut than Digital-C prints?

Oh my yes! By a lot.

Most labs tell you what kind of paper they use, but some don't say what printer they use (Fuji Frontier, Noritsu, Durst?). Does the printer hardware make any difference if the same paper is used? Comparing profiles from different labs for Fuji Crystal Archive paper I see significant differences in the 3D gamut maps. Wondering why that is?

Most labs don't have a clue about color management and supply some profile they say you can use for soft proofing but not conversions. It's questionable if the profile they supply represent the print process at the time you are sending your (ugh) sRGB data to them.
It sounds like you care about color gamut, soft proofing, control over how you use an output profile (selection of rendering intent) etc. If so, spend the $200 and output your images as intended for a proper color management workflow.

[BobShaw]

Unless you use a lab that has the same interest in the quality of the prints that you do then you aren't going to get as good a result as you are currently getting.
Check them out carefully.
They should supply custom profiles for the printers and papers they actually use?

If they say send us a 2MB sRGB file then run.

[Wayne Fox]

The challenge of running a chromogenic printer is color consistency is a two step process, first constantly recalibrating the printing system to a consistent state to compensate for the continual drifting inherent in the process.  Once that is set then you can create an ICC profile and use color management, but I’ve always felt there were some limitations because I have no control of that base setting. Another challenge is the layers in chromogenic prints are sort of translucent, where as with pigment inks they are a little more opaque. so a little different “look”. Another challenge with Chromogenic labs is they normally want things forced to sRGB and don’t offer choices as to rendering intent. 

However, it is easy to get caught up in the gamut map comparison and think that because a gamut seems “small” compared to another device the output from that device is poor.  To be honest, even side by side, well done chromogenic prints are rich in color, have good d-max and clean whites and in some cases are even more visually appealing than inkjet prints.  I use Kodak Metallic prints for some of my work, and then face mount to ArtGlass, the end result is stunning.  Side by side the same image using Epson Semi-Gloss and face mounted shows them a little “different”, but both look fantastic.

[Stephen Ray]

But you "might" have control over the base setting...

If you, as a photographer, know the Type C printer is not producing a good grey scale, correct the issue in aim points as the Type C printer owner / operator or make the correction before submitting your file to your Type C print service provider. It's not uncommon for a printer manufacturer to provide aim points (calibration targets) that are not a neutral gray. 

I was trained by Eastman Kodak to use a standard Kodak Color Separation Gray Scale as the aim for their photo printers. Many, if not all, Japanese and German photo printers that I worked with needed to be recalibrated from a cool hue to a more neutral Kodak gray. Lightjet was neutral but contrasty and overly sharpened by default. Chromira needed some special attention too.

Once you have your best B&W print from this color process, generate the ICC profile and keep the B&W as a guide to notice immediate deviations. Now we all know this is not ideal to make B&W prints because there are far better methods nowadays. This particular B&W print is just another calibration guide / tool in your shop.

You might imagine, if you have warmed up from the the ill-advised factory coolness calibrations, skin tone, metallics, landscapes, Pantones, and other images could trend more favorably.

As for chemistry calibration, the processors with the most volume seem to have the easiest chore.

Yes, chromogenic prints can be awesome.


Question; Isn't it interesting that some photographers complain there is not enough color gamut in the Type C process and yet other photographers, (sometimes the SAME  photographer) rant about how some famous guys have over-saturated their landscapes?

[digitaldog]

The challenge of running a chromogenic printer is color consistency is a two step process, first constantly recalibrating the printing system to a consistent state to compensate for the continual drifting inherent in the process.

Based on your experience, would you say this is any more or less a challenge than keeping say an E6 line or C41 process consistent?

[Wayne Fox]

Based on your experience, would you say this is any more or less a challenge than keeping say an E6 line or C41 process consistent?

Compared to film, and even RA-4 prints from negatives, todays printers calibration is a piece of cake. With those processes, especially c-41 and E-6 there just wasn't any real tolerance for time/temperature/replenishments.  but todays printerswill adjust out all the inconsistencies  of the variations of the process, so as long as you calibrate daily or twice daily depending on your volume maintaining a consistent state is a pretty simple.

But as mentioned by Stephen, it's pretty tough to modify the base point made by the manufacturer.  My Chromira seems to have a nice neutral gray, and the Noritsu's I ran in my previous company weren't bad either.  The original digital printers I had from Kodak back in the mid 90's were terrible and maintaining a neutral gray through the entire luminance values, there was always one or two patches with a bias.  The Pegasus printer (the was the code name, we were a test site, I think it ended up being called Kodak LED printer) was much better, but there still no good color management capabilities.  Now most of the printers allow profiling so results can be great as long as the base calibration is solid.

[Wayne Fox]

Question; Isn't it interesting that some photographers complain there is not enough color gamut in the Type C process and yet other photographers, (sometimes the SAME  photographer) rant about how some famous guys have over-saturated their landscapes?

yeah, Peter Lik being the most famous.

[Redcrown]

The OP is back. Been collecting sample prints from 3 labs. Staring, comparing, resting, repeating. Even brought in a couple of "kids" with young eyes to get further analysis.

My conclusion is that lab prints are OK for me, and I've confirmed the retirement of my Epson 3880. Like many things in life, I would describe this exercise as a good example of the "80/20" rule, but actually closer to "90/10".

I can get a 10% overall improvement in image quality with the Epson 3880 over labs, but at a 90% increase in effort and cost. OK, maybe only a 50% increase in effort/cost. But even that lower ratio is beyond my personal threshold of pain.

Two mitigating factors. First, I just don't print enough to amortize the cost of a printer and the effort to maintain ink and paper stock. Second, the audiences for my prints are simply not able to see or appreciate the 10% advantage of inkjet prints.

My test prints included some color challenges. Large areas out of gamut. Both out of sRGB gamut and certainly the printer gamut. But the lab printer conversion engines did a fine job.


My test labs were Mpix, ProDPI, and my local Costco. All prints on Luster paper, no lab color corrections. The loser in all comparisons was Mpix. The Mpix prints were all darker, maybe 1/2 to 1 full stop darker. Shadows blocked up much more than the other labs. Plus the Mpix prints looked more saturated (over saturated). The Mpix prints were $7.50 plus high shipping cost.

ProDPI and Costco were a virtual dead heat. I can see slight differences, but only in side-by-side under very close inspection. ProDPI and Costco prices are the same, about $4.50 for 11x14 luster. ProDPI has free shipping, took 5 business days. I drive 5 miles to my local Costco. There are 3 "pro" labs in town. Didn't use them. They charge $14 to $17 for an 11x14 luster print. Same paper, same machines as Costco/ProDPI.

I chatted up the lab manager at Costco, and was quickly convinced she knows her stuff and takes pride in the work. She seemed happy to talk to a customer who has some knowledge and interest in their process. She knows colorspaces (they accept and convert Adobe98 and Prophoto, even accept tif). She described their maintenance and calibration at length. She was excited because in a couple months they will replace their Noritsu with a Fuji Frontier.

I'm only left with one mystery. I converted a Prophoto image to the Costco profile in three different steps. One was Prophoto direct to printer profile. One was Prophoto to Adobe98 to printer profile. One was Prophoto to Srgb to Printer profile. The Srgb version looked best. It was closer to the original. Both the Prophoto direct and the Adobe98 version showed darker tones in areas that were out of gamut.

So I sent 3 copies to Costco. One in Prophoto, one in Adobe98, and one in Srgb. The prints came out the same as the conversions. Srgb looked best, Prophoto and Adobe98 looked the same. That made me think Costco was not really handling the Prophoto and Adobe98 conversions correctly. But the differences were only in out-of-gamut areas. Other colors were the same in all three prints. Still scratching my head on that one.   

[elliot_n]

Whilst good inkjet prints are technically better than digital c-types in all respects (sharpness, colour gamut, archival stability), there are plenty of people who, with good reason, prefer the look and feel of digital c-types.

[MHMG]

Two mitigating factors. First, I just don't print enough to amortize the cost of a printer and the effort to maintain ink and paper stock. Second, the audiences for my prints are simply not able to see or appreciate the 10% advantage of inkjet prints.   

Fair enough on the first point. As for the second, well, maybe your present day audience can't tell initial image quality differences, but their children will likely observe a bigger discrepancy ten or twenty years down the road as the C-prints are only moderately light fast at best, while the pigment prints from your 3880 are considerably more lightfast (even with Epson K3 yellow's weakness in light fade resistance compared to other OEM pigmented inksets). I know, I know...it's not your problem... and I've already heard all the other counter arguments 

cheers,
Mark
http://www.aardenburg-imaging.com

[Ernst Dinkla]

Question; Isn't it interesting that some photographers complain there is not enough color gamut in the Type C process and yet other photographers, (sometimes the SAME  photographer) rant about how some famous guys have over-saturated their landscapes?

Not a photographer but I would fall in the SAME category with my tastes for prints. Depending on the subject I see prints that could use more saturation and prints with too much saturation. There is also a difference in taste between the US and EU if it is about rich color and that started way before digital prints appeared. Colorants became stronger in color over a century too, for all products not just film and prints. So older generations experienced color use that they only knew of neon signs. When the iron curtain fell that difference became apparent not in time but at that opened boundary, old technology versus new. I wonder what actually caused the over-saturated landscape photography, is it the competition with TV and monitor colors in the living room? Are landscapes too dull as we are used to man made products living in the cities, more and more of us? In painting the colorants for that effect have been around way longer yet realistic landscapes had a more modest use of them, more abstract deviating in both directions though. I have been a silkscreen printer for several decades, no color limitations in that technology, but it did not change my taste much I think.

I do not see it as a contradiction, wide color gamuts being available and a modest use of the saturation tool in Photoshop. I see different tastes in color use and a wider gamut to explore.

Met vriendelijke groet, Ernst

http://www.pigment-print.com/spectralplots/spectrumviz_1.htm
December 2014 update, 700+ inkjet media white spectral plots

[dwswager]

I can get a 10% overall improvement in image quality with the Epson 3880 over labs, but at a 90% increase in effort and cost. OK, maybe only a 50% increase in effort/cost. But even that lower ratio is beyond my personal threshold of pain.

I seriously question this.  Assuming you are sending the same post processed file to the lab versus to the Epson, there is essentially no additional effort.  If we are comparing a cropped, corrected images sent to the Epson versus out of the camera Jpegs, then yes there is more effort, but should be much more than a 10% increase in quality.  Finally, you can send out of camera JPEGs to the Epson. 

As to cost, it depends.  I find the Epson 3880 to be relatively light on ink cost, especially compared to the R2400 that it replaced.  In fact, I justified the 3880 because I basically paid $850 ($300 rebate) for $1100 worth of ink (cost of ink for the R2400) and they threw in the printer for free!  I cut 24 inch rolls to size.  Most of what I print is at least 5x7 or larger (usually much larger).  I do print the odd 4x6, but usually the quality need on those are lower so I let the wife send those out.

[Redcrown]

OP again. I've made several more test prints using Costco (because it only takes 2 hours) and now see some curious results.

The Costco profiles come from drycreekphoto.com. At http://www.drycreekphoto.com/icc/using_printer_profiles.htm I found the following statement:

"...convert the image to the appropriate profile, using your chosen rendering intent. Frontier and Noritsu printers do not read embedded profiles, so the image data must be converted. This changes the data in the file to compensate for how your lab's machine actually prints colors."

That seemed strange, and contrary to many other sources. I asked drycreek about this in an e-mail 3 days ago. No response yet.

So I sent serveral versions of an image to Costco. The image has highly saturated colors, OK in ProPhoto, but lots of out-of-gamut in Srgb and the Costco profile. Plus the Costco profile shows significant differences between Relative Colormetric and Perceptual intents.

First test was do they really handle Adobe98? I sent 3 versions, one in Adobe98, one in Srgb, and one bogus copy where I ASSIGNED Srgb to an Adobe98 file. That last print came out bad, as expected. Dull, dark, desatruated colors. The true Adobe98 version looked much better, did not match the Srgb version. Colors were slightly darker, less saturated. Looked like the Costco conversion of Adobe98 was half way between the Srgb and the bogus Adobe98 version. Conclusion: Don't send Adobe98 to Costco. They do convert it, but in some strange, undesireable way.

Next test was 2 versions converted to the Costco profile (per drycreek advice). One in Relative Colormetric, one in Perceptual. There is significant visible difference in these two versions on screen, but no visible difference in these two prints. Plus, both these two prints match the Srgb version. Conclusion: drycreek is wrong about converting to printer profile. Does not hurt, but does not help either. Drycreek is also wrong about Fuji and Noritsu not reading embedded profiles. 

[digitaldog]

I suspect it's best to contact each individual Costco lab and ask them about converting/embedding.
The Drycreek page discusses converting but not embedding the profile, presumably due to some older front ends barfing when they find an embedded profile. I know that was true for some front ends in the past, I'd hope that's no longer the case this century. But who knows what equipment each shop is using or how old.
I've read a few posts over the last couple months from people saying they generally got very good advise from the Costco people running the equipment, YMMV.

[Redcrown]

Thanks, I plan to talk to the Costco manger next week. First time I went in it was a Wednesday mid-afternoon and I talked to a woman wearing a name tag that said "manager". She seemed to know her stuff and care about it. Last two times I went in it was after 8pm and the two young guys working then clearly did not have a clue.

Makes me wonder how long the machines last in "foolproof" mode between calibrations or maintenance. Might be wise to submit jobs only on weekday mornings, when the manager is there and has had time to clean up the weekend mess.

[bjanes]

OP again. I've made several more test prints using Costco (because it only takes 2 hours) and now see some curious results.

The Costco profiles come from drycreekphoto.com. At http://www.drycreekphoto.com/icc/using_printer_profiles.htm I found the following statement:

"...convert the image to the appropriate profile, using your chosen rendering intent. Frontier and Noritsu printers do not read embedded profiles, so the image data must be converted. This changes the data in the file to compensate for how your lab's machine actually prints colors."

That seemed strange, and contrary to many other sources. I asked drycreek about this in an e-mail 3 days ago. No response yet.

So I sent serveral versions of an image to Costco. The image has highly saturated colors, OK in ProPhoto, but lots of out-of-gamut in Srgb and the Costco profile. Plus the Costco profile shows significant differences between Relative Colormetric and Perceptual intents.

First test was do they really handle Adobe98? I sent 3 versions, one in Adobe98, one in Srgb, and one bogus copy where I ASSIGNED Srgb to an Adobe98 file. That last print came out bad, as expected. Dull, dark, desatruated colors. The true Adobe98 version looked much better, did not match the Srgb version. Colors were slightly darker, less saturated. Looked like the Costco conversion of Adobe98 was half way between the Srgb and the bogus Adobe98 version. Conclusion: Don't send Adobe98 to Costco. They do convert it, but in some strange, undesireable way.

Next test was 2 versions converted to the Costco profile (per drycreek advice). One in Relative Colormetric, one in Perceptual. There is significant visible difference in these two versions on screen, but no visible difference in these two prints. Plus, both these two prints match the Srgb version. Conclusion: drycreek is wrong about converting to printer profile. Does not hurt, but does not help either. Drycreek is also wrong about Fuji and Noritsu not reading embedded profiles.  

You should supply a few more details regarding your workflow and the printers and profiles used by your local Costco. The instructions on Drycreek are rather old and may not have been revised to current Noritsu and Fuji printers. It is possible that older printers did not read the profiles but newer ones do.

If you use the custom Drycreek profiles, the usual and recommended workflow is to convert to the profile using the desired rendering intent and save the image after unchecking the option to embed the profile. Since the profiles are nearly 2 MB, embedding the profile in the file greatly increases the file size. One then instructs the operator not to auto-correct the images. This sends the color data directly to the printer without any modification.

My local Costco Frontier will accept embedded profiles, but I don't know what would happen if you sent a ProPhoto image with auto-correct turned on. If it reads and applies the profile, the image would look OK if there is no serious gamut clipping. If not, the image would likely appear quite unsaturated since the RGB numbers in the wide gamut profile for a given color would be less than for a narrow gamut profile such as sRGB.

Bill