Ad
Ad
Ad
Pages: « 1 2 [3] 4 5 ... 12 »   Bottom of Page
Print
Author Topic: Clarification on Print Resolution  (Read 52550 times)
John R Smith
Sr. Member
****
Offline Offline

Posts: 1357


Still crazy, after all these years


« Reply #40 on: June 07, 2011, 12:26:42 PM »
ReplyReply

I've previously tested a lot in the range of 180 to 480, but I have not tested LR3 at 720. It goes on the list........thanks for the heads-up.

I have tested the difference between 360 and 720 ppi from LR on my Epson R2400. To utilise the 720 file, you have to have "Photo RPM" selected in the 2400 print driver - for other Epsons, it will be called something different, I expect. There is a definite improvement in the rendition of fine detail using 720 ppi, as long as the "native" resolution reported by LR was above 360 in the first place.

John
Logged

Hasselblad 500 C/M, SWC and CFV-39 DB
and a case full of (very old) lenses and other bits
Mark D Segal
Contributor
Sr. Member
*
Offline Offline

Posts: 6821


WWW
« Reply #41 on: June 07, 2011, 12:41:21 PM »
ReplyReply

Thanks for that John; when time permits I'll find a good high frequency image and test it on my 4900 and see what I see!
Logged

Mark D Segal (formerly MarkDS)
Author: "Scanning Workflows with SilverFast 8....." http://www.luminous-landscape.com/reviews/film/scanning_workflows_with_silverfast_8.shtml
John R Smith
Sr. Member
****
Offline Offline

Posts: 1357


Still crazy, after all these years


« Reply #42 on: June 07, 2011, 01:09:33 PM »
ReplyReply

Mark

I'm always working in B/W, of course, so I don't see differences in colour rendition. The file I used to test was the one I attach here, and where I really noticed the difference between the two output resolutions was in the pattern of the cups on the table, the tablecloth, and the patterns in the subject's clothing. Oh yes, I forgot, the writing in the rosette on the van. It's subtle stuff like this which you are aware of as a kind of smoothness, rather than an upfont "Wow that's better" response. But as soon as you put a loupe on it you are convinced.

John
« Last Edit: June 07, 2011, 02:09:57 PM by John R Smith » Logged

Hasselblad 500 C/M, SWC and CFV-39 DB
and a case full of (very old) lenses and other bits
Mark D Segal
Contributor
Sr. Member
*
Offline Offline

Posts: 6821


WWW
« Reply #43 on: June 07, 2011, 01:13:36 PM »
ReplyReply

Understood.

The loupe is fine for pixel-peeping in the positive sense of seeing in fine detail what's going on. For more general purposes looking at the print at normal viewing distance is what really matters; so that is what I shall focus on most when I get to testing it. most likely on 13*19 inch prints.
Logged

Mark D Segal (formerly MarkDS)
Author: "Scanning Workflows with SilverFast 8....." http://www.luminous-landscape.com/reviews/film/scanning_workflows_with_silverfast_8.shtml
John R Smith
Sr. Member
****
Offline Offline

Posts: 1357


Still crazy, after all these years


« Reply #44 on: June 07, 2011, 01:21:20 PM »
ReplyReply

. . . since print drivers don't really interpolate the image data they receive. The drivers simply convert whatever image resolution they receive into a dither for determining what droplets of ink to print and where.

Is this really true? Since this seems to be the crux of the issue. Bart seems to agree with me that whatever you send the printer will be resampled if it is not, say, 360 or 720 ppi (for Epsons). My impression was that ppi (pixels per inch as measured on the paper) and printer dpi (dithered output) are two quite separate animals, and that if you send an Epson 279 ppi it will be resampled by the printer's "black box" before it is passed to the dithering and screening department.

It would be nice to know if I am completely wrong on this, and if so, why.

John
« Last Edit: June 07, 2011, 02:02:20 PM by John R Smith » Logged

Hasselblad 500 C/M, SWC and CFV-39 DB
and a case full of (very old) lenses and other bits
digitaldog
Sr. Member
****
Offline Offline

Posts: 8628



WWW
« Reply #45 on: June 07, 2011, 04:19:02 PM »
ReplyReply

This confused me too for the last few minutes until I realized that you have to uncheck the "Print Resolution" button (which I always leave checked  Smiley) and then the native resolution figure appears next to the print dimension on the image.
Rob P

So it does. Very cool (and useful). I’m going to have to rebuild all my existing print templates with the checkbox off (easy to do, turn off, click Update settings).

On one template, for a 5x7, it even shows 720+ since apparently anything over 720 native isn’t updated to the actual value. 
Logged

Andrew Rodney
Author “Color Management for Photographers”
http://digitaldog.net/
Schewe
Sr. Member
****
Offline Offline

Posts: 5425


WWW
« Reply #46 on: June 07, 2011, 04:27:08 PM »
ReplyReply

My impression was that ppi (pixels per inch as measured on the paper) and printer dpi (dithered output) are two quite separate animals, and that if you send an Epson 279 ppi it will be resampled by the printer's "black box" before it is passed to the dithering and screening department.

I think the term "resampled" is wrong...my understanding of the Epson black box is that the printer's error-diffusion dithering algorithm is more like a fine sieve...it takes whatever image data it is given to it and based on the resolution settings of the driver, drops the image data against the dithering to determine when and where to spirt a droplet of ink.

The resolution settings of 2880/1440 on the highest setting for the 79/9900 is not dots per inch, it's droplets per inch with each droplet being 3.5 pico liters in volume. Now, exactly how Epson's most recent dither works I can't tell you other than it was a joint effort between Epson and the Rochester Institute of Technology's Munsell Color Science Laboratory and supposedly isn't a standard error-diffusion process but some kind of exotic math to arrive at the dot dither.

But in any event, I'm pretty sure the Epson driver does not resample up or down prior to generating the dither.
Logged
Schewe
Sr. Member
****
Offline Offline

Posts: 5425


WWW
« Reply #47 on: June 07, 2011, 04:28:25 PM »
ReplyReply

On one template, for a 5x7, it even shows 720+ since apparently anything over 720 native isn’t updated to the actual value. 

If the native rez is above 720 and shows the plus, it'll end up being down sampled to the current 720PPI max.
Logged
Wayne Fox
Sr. Member
****
Offline Offline

Posts: 2811



WWW
« Reply #48 on: June 07, 2011, 05:07:27 PM »
ReplyReply

Have you tested the LR output sharpening when saving out an image in LR's Print to JPEG? What seems to be lacking? Sharpening strength or sharpening type? There is some traction towards improving the Print to JPEG output sharpening in the future so it would be useful to get users' feedback.
We are using Lightroom to output rendered jpeg files which are then printed on our Imetto printer (basically a chinese version of a Durst Lamda, laser based RA-4).  We use it to print almost exclusively on Kodak Metallic Paper (very glossy).

LR seems to be hitting a sweet spot for us, results are good, better than without, better than the printers built in sharpening, don't see many artifacts (normally setting jpeg to 100 for better quality), normally using standard sharpening for glossy media type.

As far as improving it for jpeg rendering ... gonna have to leave that to experts to see if they can make it better. I'm not one of them.
Logged

Farmer
Sr. Member
****
Offline Offline

Posts: 1630


WWW
« Reply #49 on: June 07, 2011, 05:25:42 PM »
ReplyReply

The resolution settings of 2880/1440 on the highest setting for the 79/9900 is not dots per inch, it's droplets per inch with each droplet being 3.5 pico liters in volume. Now, exactly how Epson's most recent dither works I can't tell you other than it was a joint effort between Epson and the Rochester Institute of Technology's Munsell Color Science Laboratory and supposedly isn't a standard error-diffusion process but some kind of exotic math to arrive at the dot dither.

But in any event, I'm pretty sure the Epson driver does not resample up or down prior to generating the dither.


My only quibble with this, Jeff, is that it's not fixed at 3.5pl - at any given resolution (2880, 1440, 720, 360 etc) there are a number of dot sizes available.  The choice varies depending on the resolution and multiple dot sizes are employed as part of the "exotic math".  the LUT that was developed as part of all that gives choices from around 18,446,774 trillion colours (not a typo).

Otherwise, that's also my understanding.
Logged

Schewe
Sr. Member
****
Offline Offline

Posts: 5425


WWW
« Reply #50 on: June 07, 2011, 06:18:02 PM »
ReplyReply

My only quibble with this, Jeff, is that it's not fixed at 3.5pl - at any given resolution (2880, 1440, 720, 360 etc) there are a number of dot sizes available.

Actually, according to Epson when you select the 2880 rez option it ONLY prints with the smallest 3.5 pico liter droplet not the variable sizes...the variable droplet is only used for 1440/720.
Logged
Farmer
Sr. Member
****
Offline Offline

Posts: 1630


WWW
« Reply #51 on: June 07, 2011, 09:49:59 PM »
ReplyReply

Actually, according to Epson when you select the 2880 rez option it ONLY prints with the smallest 3.5 pico liter droplet not the variable sizes...the variable droplet is only used for 1440/720.

Yes, good point.  2880 mode is fixed, but all other resolutions are variable, and the choice of which variable sizes to use varies depending on other settings (media, microweave and so on).  It's a very complex matrix.
Logged

John R Smith
Sr. Member
****
Offline Offline

Posts: 1357


Still crazy, after all these years


« Reply #52 on: June 08, 2011, 01:40:53 AM »
ReplyReply

On one template, for a 5x7, it even shows 720+ since apparently anything over 720 native isn’t updated to the actual value. 

Yes, that's because the output resolution should be over 720 ppi for that print size, but 720 is the max that LR can output. You see this a lot with big MF images and small print sizes.

I think the term "resampled" is wrong...my understanding of the Epson black box is that the printer's error-diffusion dithering algorithm is more like a fine sieve...it takes whatever image data it is given to it and based on the resolution settings of the driver, drops the image data against the dithering to determine when and where to spirt a droplet of ink.

But in any event, I'm pretty sure the Epson driver does not resample up or down prior to generating the dither.


Thanks for that, Jeff. I am quite happy to be put right on this, but what I still don’t understand is – if the printer does not re-sample the incoming file to its own ppi grid, as it were, then why would you bother (as you yourself do) to specify exactly 360 or 720 ppi on the LR output side? According to your version of events, it wouldn’t make any difference if you sent the Epson 359 or 718 ppi, because the input would just be dithered, not re-sampled.

My apologies if I am just being dim  Wink

John
Logged

Hasselblad 500 C/M, SWC and CFV-39 DB
and a case full of (very old) lenses and other bits
John R Smith
Sr. Member
****
Offline Offline

Posts: 1357


Still crazy, after all these years


« Reply #53 on: June 08, 2011, 02:06:17 AM »
ReplyReply

The loupe is fine for pixel-peeping in the positive sense of seeing in fine detail what's going on. For more general purposes looking at the print at normal viewing distance is what really matters; so that is what I shall focus on most when I get to testing it. most likely on 13*19 inch prints.

Mark

Strangely enough, my own testing seems to show that other people often prefer prints with lesser quality rather than a technically higher one. I did an experiment on this a while back – I got some of my older MF B/W negatives scanned as reasonable quality TIFFs (3000x3000 pixels) and printed them via my Epson to the same size (about 10x8, quite small) as my existing wet darkroom prints of the same subjects. I tried to match the overall balance, contrast etc as closely as possible between the new prints and the old, so that they were a good match. The paper was as close a match as I could manage, too (Ilford Gallerie silver prints and Harman Baryta inkjet).

Almost without exception the people I showed them to preferred the inkjet prints from the scans rather than the darkroom prints. This was despite the fact that under a loupe the silver prints clearly had finer detail and smoother tonal gradations, especially in the highlight areas. It seemed to me that the viewing audience was responding to apparent higher levels of sharpness in the digital prints, a result of (quite subtle) USM. Typical comments were that the digital prints looked more “punchy”, “crisper”, or had “better detail” (when in fact they did not).

All of which was educational, but quite depressing.

John
« Last Edit: June 08, 2011, 07:44:43 AM by John R Smith » Logged

Hasselblad 500 C/M, SWC and CFV-39 DB
and a case full of (very old) lenses and other bits
BartvanderWolf
Sr. Member
****
Online Online

Posts: 3461


« Reply #54 on: June 08, 2011, 03:27:05 AM »
ReplyReply

I think the term "resampled" is wrong...my understanding of the Epson black box is that the printer's error-diffusion dithering algorithm is more like a fine sieve...it takes whatever image data it is given to it and based on the resolution settings of the driver, drops the image data against the dithering to determine when and where to spirt a droplet of ink.

Hi Jeff,

I've never heard an assumption like that before, but there's a first time for everything. We both don't have a solid proof (like an official statement by one of the major printer manufacturers) to support either position, but I think there is more circumstantial evidence to support simple interpolation. Not only is interpolation a logical simplification to speed up the dithering process, there are also moiré effects created that are completely consistent with (poor quality) print driver interpolation/resampling.

The printer resolution feedback from the printer driver itself when interrogated, is IMHO also a clear indicator that the fixed PPI settings are the basis for further dithering, and that interpolation is happening.

As a historical tidbit I found an old contribution by Mike Chaney (developer of Qimage) on an other website, where he also explains in simple terms what happens and why it can help to interpolate to 720/600 PPI prior to sending the data to the printer driver.

I also seem to remember a post somewhere from a relatively credible source that hinted at improvements in some printer drivers using bicubic interpolation instead of bilinear, but I'd have to search and see if I can still find that post.

There is also a faq entry on Eric Chan's website specifically mentioning the resampling issue in connection to the Epson driver settings.

Finally, it's important to remember that in order to avoid accidental downsampling artifacts by lower quality printer driver resampling, setting the printer driver to accept the highest possible input resolution (720/360 or 600/300 for the brand and paper choice) is the safe route to use. Substandard downsampling is a huge source for artifacts. That's why a program like Qimage adopted the use of anti-aliasing prefiltering for downsampling. The current LR version 3 also seems to have improved in that department. It's better to make sure what quality goes into the driver, than leaving it up to that black-box to find a quick and dirty solution.

Of course not all images are that critical that one will always run into resampling trouble, and not all images have enough detail to absolutely need the highest possible PPI setting. Nevertheless, I've seen no clear evidence that using 720/600 PPI input files has a detrimental effect on image quality, but 360/300 can have that effect.

Quote
But in any event, I'm pretty sure the Epson driver does not resample up or down prior to generating the dither.

May I suggest you check with Eric Chan what evidence he found that resampling is what is happening? I know he has exchanged views with Mike Chaney, but presumably also with others (not to be named individuals at Epson?). And perhaps (if he happens to read this) Eric can react here if he feels he can add something to clarify the situation.

Cheers,
Bart
« Last Edit: June 08, 2011, 06:09:29 AM by BartvanderWolf » Logged
Ernst Dinkla
Sr. Member
****
Offline Offline

Posts: 2806


« Reply #55 on: June 08, 2011, 04:21:10 AM »
ReplyReply


Finally, it's important to remember that in order to avoid accidental downsampling artifacts by lower quality printer driver resampling, setting the printer driver to accept the highest possible input resolution (720/360 or 600/300 for the brand and paper choice) is the safe route to use. Substandard downsampling is a huge source for artifacts. That's why a program like Qimage adopted the use of anti-aliasing prefiltering for downsampling. The current LR version 3 also seems to have improved in that department. It's better to make sure what quality goes into the driver, than leaving it up to that black-box to find a quick and dirty solution.

Cheers,
Bart

Bart,

The rest of your message is along the lines of what I would have written. It would be a strange thing that Qimage reports requested input PPI numbers like 300 and 600 PPI for HP and Canon models and 360 and 720 PPI for Epsons including the latest models if the drivers could cope with any input resolution straight away.

I am not completely happy with the part above. The point is that if aliasing can happen with the original image resolution at print size versus a lower optical resolution of the printing process then anti-aliasing should be done. By upsampling no extra aliasing conditions are created but it disguises the fact that aliasing still can happen as the 600 or 720 PPI does not represent the optical resolution of the printing process, the last will be lower. The same low optical print resolution may already be available with 300 or 360 PPI input. The example given of 560 PPI input then could give a better print if a good anti-aliasing filter was used to bring it down to 300 or 360 PPI instead of laying down 600 to 720 PPI print quality on a paper that barely does a 300/360 PPI equivalent in print quality. No magical anti-aliasing happens then with inkjet dots, the cell size still has to represent color and detail at pixel level. Correct, it will be worse with black only halftone screening but it is not dissolved by high resolution droplet distributions, stochastic screening, diluted inks or droplet variation if the paper quality can not support that printer quality setting. Again correct that good downsampling became more an issue with higher MP images, that trend of more MP is still going on. A simple rule to extrapolate everything upwards to the next native printer resolution may not be correct for all image content. It will be correct when the image resolution at print size is worse than the print resolution, it may prove wrong with the opposite.

met vriendelijke groeten, Ernst

Dinkla Gallery Canvas Wrap Actions for Photoshop

http://www.pigment-print.com/dinklacanvaswraps/index.html

Logged
Ernst Dinkla
Sr. Member
****
Offline Offline

Posts: 2806


« Reply #56 on: June 08, 2011, 05:32:15 AM »
ReplyReply

Yes, good point.  2880 mode is fixed, but all other resolutions are variable, and the choice of which variable sizes to use varies depending on other settings (media, microweave and so on).  It's a very complex matrix.

Set against the Canon iPF wide format models that squirt 4 picoliter droplets in all resolution settings, they do not have a variable droplet size but their droplet size is very close to the Epson minimum droplet size of 3.5 picoliter. It is interesting that Epson does not reveal what the maximum droplet size is but a conservative estimation of 3 droplet sizes and a maximum droplet of 11 picoliter should not be far off. Of the dry minilab models (Epson head technology - Fuji, Noritsu Brand) the smallest droplet is not mentioned but should be 2 picoliter or below for 4x6 prints and there are 5 droplet sizes in total.

met vriendelijke groeten, Ernst

New: Spectral plots of +250 inkjet papers:

http://www.pigment-print.com/spectralplots/spectrumviz_1.htm
Logged
BartvanderWolf
Sr. Member
****
Online Online

Posts: 3461


« Reply #57 on: June 08, 2011, 06:06:07 AM »
ReplyReply

I am not completely happy with the part above. The point is that if aliasing can happen with the original image resolution at print size versus a lower optical resolution of the printing process then anti-aliasing should be done. By upsampling no extra aliasing conditions are created but it disguises the fact that aliasing still can happen as the 600 or 720 PPI does not represent the optical resolution of the printing process, the last will be lower. The same low optical print resolution may already be available with 300 or 360 PPI input.

Hoi Ernst,

We agree that downsampling is aliasing artifact prone, unless proper anti-aliasing precautions are taken. So, e.g. letting a sub-optimal resampling routine downsample to 360 PPI is more likely to wreak havoc than upsampling to 720 (substitute 300 and 600 PPI for Canon/HP). The same potential risk is there if the image data exceeds 720 (600) PPI for the intended output size and thus downsampling is a given, so one had better use a very good downsampling algorithm to begin with (almost certainly not the printer driver, at the current state of affairs).

If I understand you well, what remains to be established is whether (upsampled) 720/600 PPI image data still has the potential of aliasing, given the fact that the dithering/droplet placement accuracy is higher. Is my assumption correct that that is what needs to be established, or is it the concern that the (paper) medium itself has a lower physical resolution?

If that is a correct assessment, then I'd say that aliasing primarily manifests itself when sampled in an ordered grid. It's the image detail being slightly out of sync that potentially results in massively lower resolution aliases. Both the (random and at a higher frequency) dithering should suppress aliasing, and the fibres and ink diffusion in the print medium should also help to break the regularity of the pixel grid. Adding to that, with the selection of non-glossy media, most printer drivers switch to a lower maximum PPI (something that e.g. Qimage automatically picks up and deals with).

To avoid opinions, and to get some practical results/facts, it could help to do a simple (but merciless) test with different PPI input file settings on different media. For matters involving aliasing, I find a 'zoneplate' target very helpful. I have one available for download on one of my downsampling example webpages. As detailed on that webpage, one should first convert the image mode from indexed to RGB in Photoshop, and then the target can be tagged (resized but not resampled) at different PPIs and printed as one is accustomed, and also printed after resizing to a different PPI with resampling. Converting to RGB mode is necessary to avoid shortcomings in Photoshop, and possibly other photoeditors. Keeping the image in 8 bit/channel RGB mode should be sufficient for most workflows, although one can also try if switching to 16-b/ch before resampling and/or converting to the output colorspace, helps to avoid unforeseen artifacts.

It is possible that not only downsampling but also upsampling will still produce aliasing artifacts, despite the loss of per pixel resolution, but that's due to sub-optimal resampling. A good example would be the 259/360/361 PPI test on an Epson printer. GIGO (garbage in garbage out) still rules.

Cheers,
Bart
Logged
Schewe
Sr. Member
****
Offline Offline

Posts: 5425


WWW
« Reply #58 on: June 08, 2011, 08:30:02 AM »
ReplyReply

I've never heard an assumption like that before, but there's a first time for everything. We both don't have a solid proof (like an official statement by one of the major printer manufacturers) to support either position, but I think there is more circumstantial evidence to support simple interpolation. Not only is interpolation a logical simplification to speed up the dithering process, there are also moiré effects created that are completely consistent with (poor quality) print driver interpolation/resampling.

Actually, I didn't make this up, I'm just repeating an explanation I got from an Epson engineer who was trying to explain how the Epson dither works...moiré or edge aliasing in high contrast diagonals and circles are a result of the interference between the dither and the image edges or other high frequency textural data. Upsampling to 600/720 tends to eliminate them because at the highest image resolutions, there is far less risk of encountering moiré when the dither occurs and this follows along with the Nyquist theorem of oversampling reducing  errors.

BTW, Eric Chan's 3800 page predates this latest round of testing we did regarding changing the Lightroom 3 max output resolution to 720PPI. I worked with Eric and Kevin Tieskoetter of the LR engineering team to get the resolution changed and worked with Eric (who has refined the output sharpening in LR) to test the output sharpening at those higher resolutions...
« Last Edit: June 08, 2011, 08:43:51 AM by Schewe » Logged
BartvanderWolf
Sr. Member
****
Online Online

Posts: 3461


« Reply #59 on: June 08, 2011, 10:31:50 AM »
ReplyReply

Upsampling to 600/720 tends to eliminate them because at the highest image resolutions, there is far less risk of encountering moiré when the dither occurs and this follows along with the Nyquist theorem of oversampling reducing  errors.

While true, I assume there is also an even simpler underlying reason. When the PPI results in a value above 300/360 for the output size chosen, at the default settings the printer driver will resample (yes, I do believe resampling takes place) down to 300/360 PPI. Sub-optimal interpolation/resampling is one thing, but improper (if any) AA-filtering before downsampling the pixel data can look terrible. I'm not suggesting that the Epson person was lying to you, but maybe he also didn't tell the whole story ...

Frankly, there is only so much one can expect from a printer driver. Image processing is not one of those things (although it would be nice), it's perhaps better left for dedicated output oriented software/RIPs/etc. that delivers the color-managed output data optimally resized to the printer driver. It's the kind of control spoken about earlier that one needs if components in the chain drop the ball.

Cheers,
Bart
Logged
Pages: « 1 2 [3] 4 5 ... 12 »   Top of Page
Print
Jump to:  

Ad
Ad
Ad