Ad
Ad
Ad
Pages: [1]   Bottom of Page
Print
Author Topic: Epson 3800 print quality question  (Read 7800 times)
xrogers
Newbie
*
Offline Offline

Posts: 8


« on: December 05, 2006, 04:51:36 PM »
ReplyReply

Hello all,

I'm replacing my 2200, and am considering the 2400, 3800 and hp 9180 (I got inkjetart samples for these printers, and the 4800 and Canon 5000 for curiosity and reference---all samples are printed with 360 dpi input resolution, except for the 9180 which was 600 dpi, which unfortunately makes its comparison unfair...).

When I shuffle the prints and blindly order them best to worst for how they render fine detail, I consistently rank them 9180, 2400, 4800, 5000, 3800.  The weak results for fine detail on the 3800 surprised me. I'm not saying the print was visibly bad (it's my favorite for color and tonality), but it is visibly less detailed than the other printers.  And no, I'm not using a loupe or anything (although the differences are more obvious under some sort of magnification).

I would not expect the new screening or head to produce less detail than a 2400 or 4800.  Anyone out there with the printer, have you crtically compared its fine detail to another K3 Epson printer?  Do you think there is something off with my test print?  Or am I just used to the very sharp results inkjets provide, and when Epson says the printer is "more photographic" they mean it prints slightly softer edges?

Does the 3800 benefit from a higher resolution file?  My 2200 got a small (but at times worthwhile) boost from 720 dpi input.  Maybe the 3800 needs more dpi input to render a really sharp edge?  I want this sharpness for some prints; it gives an almost 3-d look.

Thanks much for any replys, information on this printer is still scarce!

Clyde Rogers
Logged
madmanchan
Sr. Member
****
Offline Offline

Posts: 2108


« Reply #1 on: December 05, 2006, 05:01:44 PM »
ReplyReply

Clyde, Jack Flesher has started a thread in this forum regarding his new 3800, and he has briefly compared it in image quality (including detail) to his existing 7800. Hopefully you'll find the info you're looking for in there.

Eric
Logged

xrogers
Newbie
*
Offline Offline

Posts: 8


« Reply #2 on: December 05, 2006, 09:17:16 PM »
ReplyReply

Thanks for the pointer, Eric.  I had read that thread as it grew over the past few days, but hadn't remembered Jack's comment about sharpness vs. the 7800 when I looked at my samples.  I'm now thinking that my 3800 sample is a touch off.

I've also printed the same image using my 2200 using both the ImagePrint and Epson drivers.  The 2200 Epson print (with Epson driver) at 720dpi was quite sharp, exceeded only by the hp9180 (its 600dpi input clearly makes detail comparisons with the other 360dpi samples impossible).  All the other printers trounced all the 2200 prints for tonality, color and depth of black.  All the other printers beat the 360dpi 2200 for detail as well.

Most of my favorite inkjet prints have been done with ImagePrint on the 2200, and my 2200 samples came in dead last in my print comparison.  Printers have definitely improved in the past few years.

I'm still interested in knowing if anyone has run a print with 720dpi in and the driver set to "finest detail" (if I remember that setting right...), and compared it to a print with 360dpi in.  But I am getting less concerned about the 3800's fine detail.

Thanks,

Clyde Rogers
Logged
Jack Flesher
Sr. Member
****
Offline Offline

Posts: 2595



WWW
« Reply #3 on: December 05, 2006, 09:30:33 PM »
ReplyReply

Quote
I'm still interested in knowing if anyone has run a print with 720dpi in and the driver set to "finest detail" (if I remember that setting right...), and compared it to a print with 360dpi in. But I am getting less concerned about the 3800's fine detail.

Thanks,

Clyde Rogers
[a href=\"index.php?act=findpost&pid=88915\"][{POST_SNAPBACK}][/a]

Clyde, there is something wrong with your 3800 sample.  I have sent 360PPI images to the printer at 1440 DPI and you can put a loupe to the print and see additional detail!  (Same thing on the 7800 FWIW.)  

Also, on the 7800 I see essentially no advantage printing at 2880 over 1440, so as a matter of course never bother doing it. (But to be clear I have not specifically tested this yet on the 3800.)

Lastly, "finest detail" is a setting in the Epson Pro series print drivers for enhancing edge detail on text and is *not* advised for images!  So perhaps that was the problem with the print image?
« Last Edit: December 05, 2006, 09:33:12 PM by Jack Flesher » Logged

madmanchan
Sr. Member
****
Offline Offline

Posts: 2108


« Reply #4 on: December 06, 2006, 06:28:00 AM »
ReplyReply

Hi Jack, a question regarding your comment on 1440 vs 2880 dpi printing on your 7800. Do you find the difference essentially to be zero even for glossy papers, too? On my old 2200 I can see a difference with the naked eye, with the 2800 dpi producing smoother tonal gradations, but that could just be specific to the 2200.

Eric
Logged

Jack Flesher
Sr. Member
****
Offline Offline

Posts: 2595



WWW
« Reply #5 on: December 06, 2006, 08:48:54 AM »
ReplyReply

Quote
Hi Jack, a question regarding your comment on 1440 vs 2880 dpi printing on your 7800. Do you find the difference essentially to be zero even for glossy papers, too? On my old 2200 I can see a difference with the naked eye, with the 2800 dpi producing smoother tonal gradations, but that could just be specific to the 2200.

Eric
[a href=\"index.php?act=findpost&pid=88966\"][{POST_SNAPBACK}][/a]

Hi Eric:

It's difficult to quantify these differences in words, but I'll try -- with the caveat that each user should run the test for themselves since mileage may vary. That said, on the 7800 I find the difference so small as to be virtually nill, even on gloss papers. There is a difference, but it is tiny to the point of insignificance IMO.

On my 9600 however, I could see the difference -- it was still fairly small, but definitely notable on close inspection.  It also was not so much a resoluton difference as it was a smoothness of tonality difference.  My choice to not use it very often on that printer was simply due to the fact it used about 2x as much ink for the same image (and took 2x as long to print) and IMO the end result did not justify the additional ink useage or print time, especially if the print was going to be mounted behind glass.

Cheers,
« Last Edit: December 06, 2006, 08:54:44 AM by Jack Flesher » Logged

John Moody
Newbie
*
Offline Offline

Posts: 20


« Reply #6 on: December 06, 2006, 09:50:54 AM »
ReplyReply

Quote
Also, on the 7800 I see essentially no advantage printing at 2880 over 1440, so as a matter of course never bother doing it. (But to be clear I have not specifically tested this yet on the 3800.)

Jack,
I believe that Bill found the same thing on the 9800 as well.
I do see a big difference between 1440 and 2880 on the 3800 with epson premium luster (250).  Blue sky has an obviously smoother look.
I would be interested in your findings.

john
Logged
madmanchan
Sr. Member
****
Offline Offline

Posts: 2108


« Reply #7 on: December 06, 2006, 10:03:42 AM »
ReplyReply

Quote
Jack,
I believe that Bill found the same thing on the 9800 as well.
I do see a big difference between 1440 and 2880 on the 3800 with epson premium luster (250).  Blue sky has an obviously smoother look.
I would be interested in your findings.

john
[a href=\"index.php?act=findpost&pid=88988\"][{POST_SNAPBACK}][/a]


I'm assuming "Bill" in this context refers to Bill Atkinson. One thing I noticed according to Bill's color profile notes is that he himself usually uses 2880 dpi unidirectional on his 9800. That said, in analyzing his color profiles for Premium Luster recently, I noticed that the color gamut obtained using 1440 dpi is slightly (but consistently) bigger than 2880 dpi (measured by volume).

Eric
Logged

John Moody
Newbie
*
Offline Offline

Posts: 20


« Reply #8 on: December 06, 2006, 12:06:02 PM »
ReplyReply

Quote
I noticed that the color gamut obtained using 1440 dpi is slightly (but consistently) bigger than 2880 dpi (measured by volume).

Eric
[a href=\"index.php?act=findpost&pid=88997\"][{POST_SNAPBACK}][/a]

Yes, there is probably a very subtle trade-off between gamut, "lack of dots", and tonality smoothness e.g., a grainy gradient can average to more colors by small variations within the grain.  The spectro averages the dots in the patch, and so does our eye if far enough away.  
At 8" viewing distance, I find the 1440 grain objectionble.  At 2 feet away, I can't tell the difference.

john
Logged
Jack Flesher
Sr. Member
****
Offline Offline

Posts: 2595



WWW
« Reply #9 on: December 06, 2006, 01:24:14 PM »
ReplyReply

Quote
Jack,
I believe that Bill found the same thing on the 9800 as well.
I do see a big difference between 1440 and 2880 on the 3800 with epson premium luster (250). Blue sky has an obviously smoother look.
I would be interested in your findings.

john
[a href=\"index.php?act=findpost&pid=88988\"][{POST_SNAPBACK}][/a]

Funny you should mention smooth blue sky   One of my initial test prints is one that has a large, gradual-gradient clear blue sky with a hazy distant horizon on one side and a high thin cloud cover on the opposite side.  In this image the sky was perfectly smooth without banding at 1440 on both the 7800 and 3800 using my custom paper profile.  For the record, I always use unidirectional mode when printing art images.

Re the larger color gamut of 1440 over 2880:  While the smaller color gamut of 2880 is measurable, I have yet to see a negative gamut result in the few 2880 test prints I've done with it compared to 1440.  The fact also remains I see essentially no significant difference in smoothness or detail between 1440 and 2880 on my 7800.  

However I do believe we are now reaching into very subtle areas of imaging performance -- like wine snobs debating the nuances of one chateaux 1962 Gran Cru Bordeaux compared to an other's...  IMO this level of subtlety requires individual users perform their own tests regarding final image quality.  So my best recommendation is for each individual to run a comparison test for themselves and form their own opinion about the output as it relates to their specific printer and their personal artistic vision.

Cheers,
« Last Edit: December 06, 2006, 01:42:59 PM by Jack Flesher » Logged

jjlphoto
Sr. Member
****
Offline Offline

Posts: 467


« Reply #10 on: December 06, 2006, 01:57:27 PM »
ReplyReply

Quote
....Funny you should mention smooth blue sky......

I think that is an excellent test of a printers drive mechanism and handling of profiles. I have an architectural shot with a dusk sky with subtle gradients and it will bring anything less than a perfect printer to its knees. (Back in the day, I sent back three 1280's 'till I found a good one. Yes, I know the prosumer-level machines are built like toasters, but they still should perform.)
« Last Edit: December 06, 2006, 01:58:18 PM by jjlphoto » Logged

Thanks, John Luke

Member-ASMP
xrogers
Newbie
*
Offline Offline

Posts: 8


« Reply #11 on: December 06, 2006, 03:28:01 PM »
ReplyReply

Thanks to everyone responding here, it is really helpful.  I'm now on my local store's waiting list for a 3800.

Clyde Rogers
Logged
John Moody
Newbie
*
Offline Offline

Posts: 20


« Reply #12 on: December 06, 2006, 04:43:16 PM »
ReplyReply

Thanks Jack, sound advice as usual; I ran a bunch more trials.  
I also ran a colorbase calibration; no clear opinion about that yet.

I did conclude something as a result of printing a long 16-bit Lab color gradient, and additional real image tests.  
These K3 printers are capable of such subtle tonal variations, that the color space you work in can very slightly posterize the image when converted to 8-bit, and while I can't see it on the monitor, I can see it in the print!  Amazing.

I know that may sound like BS, but I'm convinced these printers are ready and able to use >8-bit drivers right now, for real images.

Till then, when it appears, I'm finding that tricks like adding 0.1% gausian noise, and critically, having good profiles, substantially eliminates the issue with only a minor increase in grain.  Obviously, you need a really good image to see the issue in the first place; I'm sure that you have many.

john
Logged
John Hollenberg
Sr. Member
****
Offline Offline

Posts: 763


« Reply #13 on: December 06, 2006, 04:55:31 PM »
ReplyReply

Quote
I did conclude something as a result of printing a long 16-bit Lab color gradient, and additional real image tests.  These K3 printers are capable of such subtle tonal variations, that the color space you work in can very slightly posterize the image when converted to 8-bit, and while I can't see it on the monitor, I can see it in the print!  Amazing.

I know that may sound like BS, but I'm convinced these printers are ready and able to use >8-bit drivers right now, for real images.

If this is true, perhaps the Canon IPF5000 Photoshop plugin is already there as it sends 16 bits (12 real bits from DSLR) of data to the printer.

--John
Logged
Jack Flesher
Sr. Member
****
Offline Offline

Posts: 2595



WWW
« Reply #14 on: December 06, 2006, 05:01:07 PM »
ReplyReply

Quote

These K3 printers are capable of such subtle tonal variations, that the color space you work in can very slightly posterize the image when converted to 8-bit, and while I can't see it on the monitor, I can see it in the print! Amazing.

I know that may sound like BS, but I'm convinced these printers are ready and able to use >8-bit drivers right now, for real images.

[a href=\"index.php?act=findpost&pid=89088\"][{POST_SNAPBACK}][/a]

John, you are absolutely correct!  Getting a bit more advanced here, but I work in a large color space and noticed this first with the 7800 and of course now with the 3800.  It is particularly notable in the high yellows through oranges.  The printer has a larger space than monitors, even a larger space than Adobe RGB if you have the $5K Eizo, so you are correct that you can't see it until you print.  You can kind of detect it on a soft proof if you know what to look for.  

FWIW, one other thing that helps in addition to your trick of adding noise, is converting to Adobe RGB, (RC with BPC) before printing.  In most cases the only visible colors trimmed are these high yellows. This does not totally eliminte the posterization but significantly reduces it.

Cheers,
« Last Edit: December 06, 2006, 05:03:59 PM by Jack Flesher » Logged

Pages: [1]   Top of Page
Print
Jump to:  

Ad
Ad
Ad