Luminous Landscape Forum
Raw & Post Processing, Printing => Digital Image Processing => Topic started by: E Slagle on February 13, 2008, 09:38:37 am
-
On page 335 of Real World Camera Raw for CS3 the author (Schewe) makes this comment:
"Enter 360 pixels per inch in the Resolution field, because you'll almost certainly check your edits by printing to an inkjet printer at 360 ppi."
This comment is in regards to building a soft-proofing/sharpening action.
My question is why? This comment is both unsupported in its reasoning and contradicts what is written earlier in the book.
Thanks, Eric
-
"Enter 360 pixels per inch in the Resolution field, because you'll almost certainly check your edits by printing to an inkjet printer at 360 ppi."
HP and Canon don't like 360.
Epson desktop printers are happy with 720.
Contone printers like different numbers.
360 is not a magic number.
Jacopo
-
HP and Canon don't like 360.
Epson desktop printers are happy with 720.
Contone printers like different numbers.
360 is not a magic number.
Jacopo
[a href=\"index.php?act=findpost&pid=174566\"][{POST_SNAPBACK}][/a]
Jeff discusses this in the Camera to Print video. 360 is the "magic number" so to speak for Epson (and he admits he's only familiar with that brand in this respect). Epson's "base" is 720X1440 making the native resolution of the printer 360dpi. He states (with Michael agreeing) HP and Canon are 600x1200,making their native resolution 300dpi. But the bottom line is to send ALL the pixels you have for the print size you wish assuming the resolution falls within the 180-480ppi range and let the printer do the interpolation to produce the desired size.
-
Okay. This is starting to make some sense. I've done all my homework (Real World CR, C2P, ect); I just thought that Jeff was stating this for some non-printer related reason (i.e. put my nose to print and check out whether I "like" my edits).
Consensus so far seems that this quote is a function of an Epson bias; I use HP so 300 it is for me.
Eric
-
Consensus so far seems that this quote is a function of an Epson bias; I use HP so 300 it is for me.
Eric
[a href=\"index.php?act=findpost&pid=174587\"][{POST_SNAPBACK}][/a]
Again, referencing the Camera to Print video, Michael talks about a conversation with an HP color scientist who confirms what Jeff is suggesting: Send the data you have to the printer as long as it falls within 180-480ppi (don't send it more, driver chokes, if you don't have 180, interpolate). If you interpolate up (use Photoshop) do so in a 180-240 range (more is unnecessary).
-
Just to clarify another thing.
"Native resolution" does not exist.
Some numbers for my Canon printer :
600 PPI for high quality
300 PPI for low quality
619 PPI for high quality, borderless 4x6
615 PPI for high quality, borderless 5x7
609 PPI for high quality, borderless A4
Jacopo
-
Again, referencing the Camera to Print video, Michael talks about a conversation with an HP color scientist who confirms what Jeff is suggesting: Send the data you have to the printer as long as it falls within 180-480ppi (don't send it more, driver chokes, if you don't have 180, interpolate). If you interpolate up (use Photoshop) do so in a 180-240 range (more is unnecessary).
[a href=\"index.php?act=findpost&pid=174593\"][{POST_SNAPBACK}][/a]
OK but I have one problem with this advice. Conventional wisdom is to uprez and then do output sharpening, correct? (You said as much in another thread today).
So if I have a 190ppi file it seems my choices are:
1)Perform output sharpening on 190ppi and send it to printer which will internally uprezz to its native PPI (say 300 or 360, or at least an even multiple thereof).
2) Up-rez to 300 or 360 ppi myself (depending on printer brand) and then do output sharpening and send to print.
Maybe the difference between the two approaches is not huge, but should (2) have an advantage even if only slight? What happens to the capture sharpening I applied to my 190ppi file once the printer interpolates to its native resolution? (using who knows what algorithm)
-
Interestingly, Schewe wrote the following in a thread here the other day:
The actual method of producing the Epson's variable droplet size is an exotic error diffusion algorithm. To the driver, it really doesn't matter whether the resolution is a mathematical divisor.
That used to be the common wisdom, however, Bruce Fraser has written that it's far more useful to use the REAL resolution without resampling and letting the PPI fall where it may for a given image size. So, resize without resampling to get the print size you want and let the actual resolution fall where it may–as long as you are in the range of 180-480PPI.
[a href=\"index.php?act=findpost&pid=174131\"][{POST_SNAPBACK}][/a]
I never have noticed much difference either way, although I make a point to rasterize type to 180 or 360, but that could just be superstition.
-sean
-
OK but I have one problem with this advice. Conventional wisdom is to uprez and then do output sharpening, correct? (You said as much in another thread today).
So if I have a 190ppi file it seems my choices are:
1)Perform output sharpening on 190ppi and send it to printer which will internally uprezz to its native PPI (say 300 or 360, or at least an even multiple thereof).
2) Up-rez to 300 or 360 ppi myself (depending on printer brand) and then do output sharpening and send to print.
Maybe the difference between the two approaches is not huge, but should (2) have an advantage even if only slight? What happens to the capture sharpening I applied to my 190ppi file once the printer interpolates to its native resolution? (using who knows what algorithm)
[a href=\"index.php?act=findpost&pid=174646\"][{POST_SNAPBACK}][/a]
You have to decide what size print you want before you can output sharpen. If you decide to send 190ppi, that's a fixed value at some size of which output sharpening will be decided at this point. You'd know at this point of you only had say 170 for the fixed size print you wished so of course, you'd have to resize up, THEN sharpen. Of course you do this on a layer because you might want a smaller print a week later, and use a different set of pixels going off to the printer for that specific size.
-
On page 335 of Real World Camera Raw for CS3 the author (Schewe) makes this comment:
"Enter 360 pixels per inch in the Resolution field, because you'll almost certainly check your edits by printing to an inkjet printer at 360 ppi."
[a href=\"index.php?act=findpost&pid=174535\"][{POST_SNAPBACK}][/a]
You'll note that the resize is NOT a resample but merely changing the size/resolution factor to one that is a useful starting point.
Yes, I print with Epson printers...the baseline resolution of Epson printers is 180 nozzles/inch (except the 11880 which has 360 nozzles/inch)...but this is the effective base resolution of the printer before the stepper motor is factored in...the "effective" resolution of the Epson's are thought of as 360DPI) that's DOTS PER INCH which should NOT be confused with an image's resolution of Pixels Per Inch which is a different factor.
If you routinely process your images and prefer to print out at 240PPI or even 300PPI, you are free to write your own actions to set the base resolution of your archive file. For me and my output, I tend towards 360 as an archive file. If I want to print out smaller, I resize without resampling to get to 480PPI. If I need a bigger image, I resize without resampling to 240 or 180PPI.
Each final output resolution WILL need it's own separate output sharpening because output sharpening is dependent upon the image's PPI setting. It's not the size that matters (really!) it's the pixel density.
-
It's not the size that matters (really!) it's the pixel density.
[a href=\"index.php?act=findpost&pid=174662\"][{POST_SNAPBACK}][/a]
Now that's something I intend to tell my wife tonight <g>
-
You have to decide what size print you want before you can output sharpen. If you decide to send 190ppi, that's a fixed value at some size of which output sharpening will be decided at this point. You'd know at this point of you only had say 170 for the fixed size print you wished so of course, you'd have to resize up, THEN sharpen. Of course you do this on a layer because you might want a smaller print a week later, and use a different set of pixels going off to the printer for that specific size.
[a href=\"index.php?act=findpost&pid=174656\"][{POST_SNAPBACK}][/a]
Maybe I wasn't entirely clear. When referring to a printer's "native resolution" this has nothing with nozzles or droplets or diffusion algorithms. I'm talking about the PPI which the print driver (ie software) uses. My understanding of the way Windows print drivers work is that if you're printing to an Epson 3800 (non-borderless), your image will get interpolated to 360ppi, the only question is whether you're going to do it in Photoshop or allow the print driver to do the interpolation on the fly before it sends the image data to the printer.
So lets say I have a D2x file that I want to print at 16x24. That gives me 178ppi before interpolation. Now, which approach is likely to be better:
1) Upsize file to 360ppi (in Photoshop or some other tool), and then apply output sharpening.
2) Apply output sharpening to 178ppi file, and then upsize to 360ppi (in print driver).
IMHO option 2 is better. How much difference there is will depend on the particular image and possibly other factors, but it still seems the better way to go if you don't mind a little extra work in order to ensure maximum possible print quality.
Again, this is my understanding of how things work on Windows OS with standard print drivers. I have now idea if there are difference when using Mac's or 3rd party RIP's.
-
Maybe I wasn't entirely clear. When referring to a printer's "native resolution" this has nothing with nozzles or droplets or diffusion algorithms. I'm talking about the PPI which the print driver (ie software) uses. My understanding of the way Windows print drivers work is that if you're printing to an Epson 3800 (non-borderless), your image will get interpolated to 360ppi, the only question is whether you're going to do it in Photoshop or allow the print driver to do the interpolation on the fly before it sends the image data to the printer.
And according to the people in the know, within at least HP and Epson, they say "let the driver do the interpolation". That's discussed in the Camera to Print video. They state that unless the image resolution falls below 180 or above 480, just send that data to the printer and let it have at it.
-
It seems to me that this discussion is being complicated by trying to make PPI and DPI synonymous. In reality they are not. CS3 or whatever is dealing with the pixels that the camera captures. The number per inch is a handy way to judge the size of the image relative to what will be printed.
The printer driver is responsible for converting the pixel information into ink droplets and applying them to paper. That process involves an algorithm that looks at the pixel density per inch and each pixel's color information in RGB and converts it to CMYK DPI data that is used to apply ink to the paper. The dithering, etc. that are also part of the printer driver combine to result in a CMYK dot pattern that looks "smooth" to the human eye.
Based on the Epson, HP, Canon engineer attributions in this thread and the Camera to Print videos it sounds to me like they use approximately 330 plus/minus 30 as the mid-point for the pixel to ink dot conversion. They have added approximately a 50% working range to that mid-point which delivers the 180 to 480 useful span the engineers agree the drivers are capable of handling.
That is good news for us. My 5D files can be printed to a size ranging from 24 inches to 9 inches on the long side without having to up or down res. Just manipulate the PPI setting (or more likely the inch setting) to control the print size.
I then can use PKS output sharpening (using the closest PPI preset) to get the image ready to print.
My experience is that this works well, I just pick the output image size in inches and go.
One Man's View
David
And according to the people in the know, within at least HP and Epson, they say "let the driver do the interpolation". That's discussed in the Camera to Print video. They state that unless the image resolution falls below 180 or above 480, just send that data to the printer and let it have at it.
[a href=\"index.php?act=findpost&pid=174684\"][{POST_SNAPBACK}][/a]
-
Maybe I wasn't entirely clear. When referring to a printer's "native resolution" this has nothing with nozzles or droplets or diffusion algorithms. I'm talking about the PPI which the print driver (ie software) uses.
[a href=\"index.php?act=findpost&pid=174668\"][{POST_SNAPBACK}][/a]
See, that's where ya got it wrong...the driver TAKES the image data at it's actual dimension and then runs what is, in effect, an error diffusion process to determine where droplets will and won't be. The driver (at least from the Epson side and this is coming from a guy WAY smarter than me-Parker Plaisted) takes whatever data it's given and runs it through a sieve (the metaphor for the error diffusion), depending on the resolution settings on the driver, the sieve gets larger or smaller openings to create the stochastic halftone that is then broken down into a droplet map to tell the print head when to and when not to squirt some ink.
In the case of Epson, there are (and I'm remembering this from a while ago so it may have changed with newer printers) at least 3 distinct droplet sizes that the print head can create and they are measured in picoliters which one can not easily translate into a physical size. A picoliter is a trillionth (one millionth of a millionth, or 10 to the -12th power) of a liter but due to differing densities and volumes, that simply does not translate to actual dots per inch.
The best you can do is talk about "relative resolutions", not "absolute resolutions" when it comes to error diffusion type of halftoning that is then printed in droplets.
Yes, there are settings in Windows (and Mac) print drivers that "announce" their "resolution" to the system...but that's more to qualify as low resolution devices vs high resolution devices and should not be construed to be an actual, defined, absolute resolution.
Bruce Fraser, in his Real World Image Sharpening talked about the kind of resolution you need to NOT see any actual dots when printed...but here again, the problem is human vision isn't measured in PPI or DPI...since human vision is measured in minutes of arc (at about 1.5 minute of arc per line pair) that doesn't translate to dots on a page...what Bruce did was to factor out what the human vision was capable of resolving at various distances (since distance has an impact because of the arc). Bruce figured that a person with 20/20/20 vision in good light could resolve about 355 dots/inch at a distance of 12 inches. Note, the 20/20/20 is a Bruce joke, that equates to a 20 year old with 20/20 vision. Close focus gets poorer the older you get.
So, hold a print about 12 inches away...if the print has ~355 DPI, you won't "SEE THE DOTS"...hold it closer and you will. Which is why it's pretty cool that if you are making small prints (where due to viewing distances you NEED more rez) you can resize without resampling and get smaller, higher resolution prints. On the other hand, if making large prints that will be viewed beyond 12 inches, the falloff of the required resolution drops quickly.
Of course, Bruce also has been quoted as saying that the intended viewing distance of ANY print made by a photographer is limited only by the length of their nose...(or the quality of their reading glasses).
What does all this mean?
Well, the bottom line is, where possible, always try to maintain the "native resolution" of your file and resize, without resampling to get the SIZE of the print image you want and let the PPI resolution fall where it will. Once you get the image SIZE figured out (which will then and only then give you the final pixels per inch) you sharpen for that pixel density. This works well when the native resolution is between 180-480 PPI.
Need to make a small print? Resize without resample smaller, then sharpen for the output.
In my case, with the typical MP (mega pixels-yet ANOTHER unrelated measurement scale) that I shoot and the print sizes I generally want, 360PPI is just about optimal...which is a long way of explaining why I chose to set the image resolution to 360PPI (without much concern over the image size).
-
If you routinely process your images and prefer to print out at 240PPI or even 300PPI, you are free to write your own actions to set the base resolution of your archive file. For me and my output, I tend towards 360 as an archive file. If I want to print out smaller, I resize without resampling to get to 480PPI. If I need a bigger image, I resize without resampling to 240 or 180PPI.
[a href=\"index.php?act=findpost&pid=174662\"][{POST_SNAPBACK}][/a]
Thanks for the clarification.
Positioning it as an archiving decision makes far more sense; unfortunately, your book does not convey this well (if I may offer a bit of constructive criticism).
Eric
-
Here is a quick Excel table that illustrates the minute of arc idea relative to distance from the image:
Circum Distance 1 Minute factor LP width
75.3984 12 0.003490667 1.5 0.005236
376.992 60 0.017453333 1.5 0.02618
3769.92 600 0.174533333 1.5 0.2618
That's why a colored light bulb display on a billboard has an acceptable image quality at a stadium or on the expressway.
Printers know nothing of pixels at the print head. The driver software is performing a conversion process from whatever the pixel data is to information that is used to deliver ink from the print head, which as Jeff has said earlier, has a defined mechanical pattern that is determined by the number of nozzles and the stepper motor that controls the head movement.
We should all be grateful to the printer engineers for designing the range tolerance into the drivers so that we can pick a convent pixel per inch setting inside the software that can then be used by the driver to deliver a quality print within the range of the driver.
It sounds like all three major printer manufacturers have drivers that cover essentially the same range of pixels per inch so that we can print images in the most common sizes without having to up or down res.
I have found that importing 5D images from Camera Raw at 240 PPI gives me a good starting point. The image will print about 18 inches on the long side. I can reduce the image size to about 10 inches on the long side and still be within the 180 to 480 tolerance range. or I can go up to about 18 inches on the long side and also be within the tolerance range. On a 13 inch or even a 17 inch printer that means I can get a 2X3 ratio image without having to upres.
The prints look fine, even when I get my nose REALLY close. Don't forget, we are usually running the printer at 1440 or 2880 ink drops per inch. which means there are a lot of ink droplets making up that DPI resolution that the printer driver is using.
David
See, that's where ya got it wrong...the driver TAKES the image data at it's actual dimension and then runs what is, in effect, an error diffusion process to determine where droplets will and won't be. The driver (at least from the Epson side and this is coming from a guy WAY smarter than me-Parker Plaisted) takes whatever data it's given and runs it through a sieve (the metaphor for the error diffusion), depending on the resolution settings on the driver, the sieve gets larger or smaller openings to create the stochastic halftone that is then broken down into a droplet map to tell the print head when to and when not to squirt some ink.
In the case of Epson, there are (and I'm remembering this from a while ago so it may have changed with newer printers) at least 3 distinct droplet sizes that the print head can create and they are measured in picoliters which one can not easily translate into a physical size. A picoliter is a trillionth (one millionth of a millionth, or 10 to the -12th power) of a liter but due to differing densities and volumes, that simply does not translate to actual dots per inch.
The best you can do is talk about "relative resolutions", not "absolute resolutions" when it comes to error diffusion type of halftoning that is then printed in droplets.
Yes, there are settings in Windows (and Mac) print drivers that "announce" their "resolution" to the system...but that's more to qualify as low resolution devices vs high resolution devices and should not be construed to be an actual, defined, absolute resolution.
Bruce Fraser, in his Real World Image Sharpening talked about the kind of resolution you need to NOT see any actual dots when printed...but here again, the problem is human vision isn't measured in PPI or DPI...since human vision is measured in minutes of arc (at about 1.5 minute of arc per line pair) that doesn't translate to dots on a page...what Bruce did was to factor out what the human vision was capable of resolving at various distances (since distance has an impact because of the arc). Bruce figured that a person with 20/20/20 vision in good light could resolve about 355 dots/inch at a distance of 12 inches. Note, the 20/20/20 is a Bruce joke, that equates to a 20 year old with 20/20 vision. Close focus gets poorer the older you get.
So, hold a print about 12 inches away...if the print has ~355 DPI, you won't "SEE THE DOTS"...hold it closer and you will. Which is why it's pretty cool that if you are making small prints (where due to viewing distances you NEED more rez) you can resize without resampling and get smaller, higher resolution prints. On the other hand, if making large prints that will be viewed beyond 12 inches, the falloff of the required resolution drops quickly.
Of course, Bruce also has been quoted as saying that the intended viewing distance of ANY print made by a photographer is limited only by the length of their nose...(or the quality of their reading glasses).
What does all this mean?
Well, the bottom line is, where possible, always try to maintain the "native resolution" of your file and resize, without resampling to get the SIZE of the print image you want and let the PPI resolution fall where it will. Once you get the image SIZE figured out (which will then and only then give you the final pixels per inch) you sharpen for that pixel density. This works well when the native resolution is between 180-480 PPI.
Need to make a small print? Resize without resample smaller, then sharpen for the output.
In my case, with the typical MP (mega pixels-yet ANOTHER unrelated measurement scale) that I shoot and the print sizes I generally want, 360PPI is just about optimal...which is a long way of explaining why I chose to set the image resolution to 360PPI (without much concern over the image size).
[a href=\"index.php?act=findpost&pid=174711\"][{POST_SNAPBACK}][/a]
-
Epson wrote:
All Epson large format printers use 360dpi as the input resolution (this is the resolution data is rasterized at)
As for the Epson desktop products, they rasterize data at 720dpi
http://files.support.epson.com/pdf/pro10a/pro10aps.pdf (http://files.support.epson.com/pdf/pro10a/pro10aps.pdf)
Jacopo
-
the bulletin you posted the link to is dated 2004.
That is ancient history in the technology business.
I would expect that the information that Jeff is relating from the Epson engineers is more up to date than this bulletin.
If they say 180 to 480 that implies that 360 is about the middle of the range.
DMW
Epson wrote:
All Epson large format printers use 360dpi as the input resolution (this is the resolution data is rasterized at)
As for the Epson desktop products, they rasterize data at 720dpi
http://files.support.epson.com/pdf/pro10a/pro10aps.pdf (http://files.support.epson.com/pdf/pro10a/pro10aps.pdf)
Jacopo
[a href=\"index.php?act=findpost&pid=175872\"][{POST_SNAPBACK}][/a]
-
Everyone seems to think that a driver can only accept 1 input value to work out the output.
Some of the Epson printers now have up to 5 droplet sizes of which 3 are used at any given output resolution. There's no reason to think that the driver can only accept data at a certain resolution because there's no constant translation from ppi to dpi - it changes depending on the colours, the output resolution and even the paper type (particularly between matte and glossy) as well as neighouring colours for any given "dot" (and I use "dot" because that output dpi will almost certainly consist of numerous dots fired from the printer).
It's obvious when you consider a case such as an image that is 3600x3600 pixels printed 10"x10" at 360dpi in the driver. The printer can't print it with a simple 1 dot to 1 pixel arrangement as each colour requires more than 1 ink dot, and this is further complicated by the need for it to be related to its neighouring dots to provide accurate colour across the image. Each time you send a different image of the same specs in the above scenario, the actual number of dots laid down by the printer will be different.
So, with no constant relationship between image ppi and printer dpi, there's no need for the driver algorithm to be restricted to a single image ppi source (and thus need to interpolate). The algorithm will have a range of values for which it can provide an answer (ie a dot pattern that works).
As Jeff and Andrew (in particular) have mentioned numerous times, they're not making this stuff up - it's been reported to them by engineers from more than one of the printer companies and it easily stands real world testing.
Whether individuals prefer the results from doing their own interpolation to a given ppi is another matter, of course.
-
dmward
That is ancient history in the technology business
If you are able to understand it, you can download the kit for drivers building of Epson. This is not out of date.
You get the same numbers.
Farmer
Everyone seems to think that a driver can only accept 1 input value to work out the output
This is exactly how any printer driver works, not only Epson.
Dithering is the process that connects PPI to DPI, but input pixels are sampled at a definite rate (depending on driver settings).
Jacopo
-
This is exactly how any printer driver works, not only Epson.
Dithering is the process that connects PPI to DPI, but input pixels are sampled at a definite rate (depending on driver settings).
[a href=\"index.php?act=findpost&pid=176143\"][{POST_SNAPBACK}][/a]
Yes, based on driver settings - there's no single magical number for all print jobs. It's also rather more complicated (particularly with the work in the new LUT for the R1900 which will flow through to other printers, for example) than just sampling back to 360 (or 300 or 600 or 720). How the printer decides to get the data it wants to fill the matrix is device specific, which means it's going to be better than some software just doing the sampling for you that knows nothing about the capabilities of the printer.
That's why the printer company engineers are telling us leave your image between 180 and 480 and send it to the printer.
-
Yes, based on driver settings - there's no single magical number for all print jobs. It's also rather more complicated (particularly with the work in the new LUT for the R1900 which will flow through to other printers, for example) than just sampling back to 360 (or 300 or 600 or 720). How the printer decides to get the data it wants to fill the matrix is device specific, which means it's going to be better than some software just doing the sampling for you that knows nothing about the capabilities of the printer.
That's why the printer company engineers are telling us leave your image between 180 and 480 and send it to the printer.
[{POST_SNAPBACK}][/a] (http://index.php?act=findpost&pid=176154\")
You don't pay attention to my words.
I quote myself in this same thread:
Just to clarify another thing.
"Native resolution" does not exist.
Some numbers for my Canon printer :
600 PPI for high quality
300 PPI for low quality
619 PPI for high quality, borderless 4x6
615 PPI for high quality, borderless 5x7
609 PPI for high quality, borderless A4
And in another thread:
I know that large format Epson printers use a PPI value of 360.
I am not sure if 11880 follows this rule, you can check it.
If 360 is the right value then 180=360/2 may be good enough, 360 is the best.
The 240 value is a good value for Epson desktop printers (240=720/3), but 360 is better and 720 is the best.
Sampling is performed before dithering, so the best thing to do is resampling to the driver PPI value.
Dithering may alleviate artifacts apparence from printer driver resampling more or less.
To check the 11880 PPI go to my web site and download PrinterData (Windows application).
[a href=\"http://www.photoresampling.com/index_eng.php]http://www.photoresampling.com/index_eng.php[/url]
The LUT you are referring to is to compute the conversion from RGB to percentage of inks. This is off topic.
which will flow through to other printers
What does it mean ???
Jacopo
-
Yes, I did read. I understand that you produce software for resampling images for printing so you have a certain agenda to push. That's fine, btw, I'm not criticising you or suggesting anything negative about your software (in fact, it's well reviewed from what I've seen).
I know what the LUT is for, the point is that the entire process is more complex than some people (note - I've made a point of being broad in my comments because they are NOT targetted at you specifically) choose to suggest.
You need to stop taking comments made in a public thread to be solely directed at you, mate. They're not. They're general comments made regarding the entire discussion.
Your utility reports the 11880 at 360 unless "Finest detail" is ticked in the driver in which case it reports 720. No surprises there.
What you're apparently suggesting, having read through your site, is that your resampling gives a better result than sending images to the printer at other ppi or resampling using the likes of Photoshop. I haven't tested your software, so I can't comment.
When I have time, I'll install your software and do some prints and run some blind tests to see if people can pick the differences.
My recommendation, though, is that people send data anywhere from 180 to 480 and they will have excellent results and won't be able to pick the difference except under a loupe. In fact, 120 is fine in some cases, but I would avoid it generally.
FWIW, I'm not guessing about this stuff, but doing the test with your software will be interesting for sure.
-
You need to stop taking comments made in a public thread to be solely directed at you, mate. They're not. They're general comments made regarding the entire discussion.
[a href=\"index.php?act=findpost&pid=176159\"][{POST_SNAPBACK}][/a]
My comments are not directed to me.
I tried to explain how a printer works.
A driver do 3 things:
1. sample the image to pick up pixels color
2. translate the RGB data into ink percentages (the LUT)
3. apply a dithering algorithm to simulate the colors
Mxing this 3 steps is to make anything foggy.
Is the general discussion devoted to clarify how printers work?
I gave a well documented contribute. You can't say the same for others.
I think you should be happy if someone try to explain something without saying : "trust me, do that. It is the best that you can do".
In any case, you and everyone, are free to close eyes and brain and believe any undocumented and obscure explanation.
Jacopo
-
In any case, you and everyone, are free to close eyes and brain and believe any undocumented and obscure explanation.
Yeah, 'cause that's a great way to promote your software, by calling anyone who disagrees with you names.
Perhaps you can provide some of that well documented evidence to back up the claim on your website that your resampling algorithm is better than anyone elses?
-
Perhaps you can provide some of that well documented evidence to back up the claim on your website that your resampling algorithm is better than anyone elses?
[a href=\"index.php?act=findpost&pid=176174\"][{POST_SNAPBACK}][/a]
Out to print please (and on a number of subjects). I've tried all kinds of such software, haven't found any yet that in anything other than a toss up were "better" than using Photoshop and when they are, its nearly impossible to see on print (and hardly visible on screen). Yet I hear all kinds of users of products saying "I blew up this image XXX% and look how good it looks" (never anywhere as good as using real data and never next to the same print made with upsizing and proper subsequent sharpening in Photoshop).
-
Yeah, 'cause that's a great way to promote your software, by calling anyone who disagrees with you names.
Perhaps you can provide some of that well documented evidence to back up the claim on your website that your resampling algorithm is better than anyone elses?
[a href=\"index.php?act=findpost&pid=176174\"][{POST_SNAPBACK}][/a]
I am not promoting anything. I was speaking about printer drivers. You introduced the resampling algorithm that is off topic as was the LUT.
Jacopo
-
On your website (to which you referred me) you state, "GBLS is my own algorithm which, offers better results than all other common algorithms."
You're the one claiming that the driver samples at a fixed rate and that providing data at that rate gives the best results. You also make the above claim.
I'm going to test your software - I'm fortunate enough to have access to a wide array of printers.
The point I have been making (and admittedly not as well as people like Jeff and Andrew who are more used to doing this) is that the drivers are designed to work within a range, not a single value, and providing data anywhere within that range will give you acceptable results - results that are superior to user-resampling to some specific resolution before printing. You claim otherwise.
Since you seem to not want to discuss this further, I'll do some testing.
-
Well I was going to test it here at work, but the demo version does not allow file saving or printing (and splashes the word "demo" dozens of times all over the image).
I understand the need to protect your software, but a demo that can't actually be demonstrated is difficult to evaulate - indeed, there's no way to verify or dispute your claims without first paying you...
So, as far as I can see there's no way to test this without paying the 55 Euro to buy it, which I'm not going to do.
Oh well.
-
Epson drivers do interperlate to 360 ppi.
This can be demonstrated by either having an image that has a repeating pattern of lines such as corigated iron or even paterned suiting material.
Or setting up in photoshop a document with close lines in a pattern.
If these are processed at 300PPI and printed the is a strong liklyhood that there will be an interferance paterns visible, if processed at 360ppi these paterns are not there.
Peter
-
Epson drivers do interperlate to 360 ppi.
This can be demonstrated by either having an image that has a repeating pattern of lines such as corigated iron or even paterned suiting material.
Or setting up in photoshop a document with close lines in a pattern.
If these are processed at 300PPI and printed the is a strong liklyhood that there will be an interferance paterns visible, if processed at 360ppi these paterns are not there.
Peter
[{POST_SNAPBACK}][/a] (http://index.php?act=findpost&pid=176376\")
The 360 ppi figure is often given for Epson printers. [a href=\"http://www.rags-int-inc.com/]Rags Gardner[/url] has done some interesting experiments with the Epson 2200 by printing images with a grid pattern and observing the interference patterns that result. On his site, go to the technology, injket tabs to reach the post. According to his results, the Epson 2200 has a native resolution of 288 ppi. This resolution is determined by the physical spacing of the printer nozzles, which is fixed in the direction perpendicular to the print head, but can be varied by the printer stepper in the direction of movement of the print head. The software interpolation may not match the hardware resolution with some printers, so there is a further complication.
In his Real World Sharpening book, Bruce Fraser refers to the above resolution as the addressable resolution of the printer. He states that some pundits claim to get better results by printing at even multiples of the printer's native resolution, often assumed to be 360 ppi for Epson, and this does apply to targets containing line pairs but in real world photography there is little difference. Indeed, he states that resampling of the image to the "native resolution" may do more harm than good.
Within limits, he recommended printing at the native capture resolution without resampling. Those who resample to the "native" printer resolution or some multiple thereof have to know the native resolution, and it may not be what they think it is.