Comparing lenses for full frame vs. APS-C sensors

[Manfred M]

In other words, if more "detail" is required: use a longer lens or step closer.

That is not always possible. I've been preparing two of my images for a print show at one of the local galleries. I'm going with two prints that will be from a full-frame camera that will be printed on a printer with a native resolution of 360 dpi to 44" wide by 66" high (the actual image will be 40" x 60" with a 2" border all around). I won't be doing those on my little 17" wide printer

Your method doesn't help there, but careful up-rezzing is required...

[William W]

Interesting sub topic-

. . .
What I find intriguing is the difference between traditional and AI-based upsampling. The traditional methods, which use methods like bucubic and nearest neighbor sampling, can be described by mathematical algorithms. If you dig a bit, you can find the math online. The software is simply doing that math. If I understand correctly, the AI-based upsampling is fundamentally different; the software includes a summary of the patterns the computer (theirs, not yours) found in [examining] a very large number of test images. In that respect, it's similar to AI-based selection methods. And I think for now, one can expect the same result: sometimes they will work better, sometimes they won't work as well, and the only way to be certain is to try.

My (only fundamental) understanding is the same as yours. Having had the opportunity to view the use of AI Digital Enhancement Technology for forensic purposes, it is 'amazing': that point aside, I gleaned that the number of and range of 'test images' to which you refer, is critical to the quality and precision of the outcome. Suffice to say that, (for example) for forensic and/or law enforcement applications, the technology can be tailored to suit specific enhancements within a relatively narrow range of parameters whilst still harvesting a large number of 'test images' - and the TR&D budget would arguably be very big.

Not necessarily so for the AI programs we general photographers might buy, or get for free and use; that is use for a wide range of image types.

WW

[xpatUSA]

Originally Posted by xpatUSA In other words, if more "detail" is required: use a longer lens or step closer.

That is not always possible. I've been preparing two of my images for a print show at one of the local galleries. I'm going with two prints that will be from a full-frame camera that will be printed on a printer with a native resolution of 360 dpi to 44" wide by 66" high (the actual image will be 40" x 60" with a 2" border all around).

And what is the intended viewing distance for The Print? 360dpi implies about 10 inches ...

I won't be doing those on my little 17" wide printer

Your method doesn't help there, but careful up-rezzing is required...

On the other hand, something acceptable on my 24-inch monitor at 96ppi viewed at 18 inches would look equally acceptable if viewed on a 48-inch monitor at 48ppi from 36 inches. So there must a different reason than human visual acuity as to why a huge 40x60" print (which should more properly be viewed from about six feet away) gets a whopping 360dpi? No doubt I will learn why shortly ...

[Manfred M]

And what is the intended viewing distance for The Print? 360dpi implies about 10 inches ...



On the other hand, something acceptable on my 24-inch monitor at 96ppi viewed at 18 inches would look equally acceptable if viewed on a 48-inch monitor at 48ppi from 36 inches. So there must a different reason than human visual acuity as to why a huge 40x60" print (which should more properly be viewed from about six feet away) gets a whopping 360dpi? No doubt I will learn why shortly ...

We still have not met with the curator as this will be a 6 person show, so I'm not sure which space I am going to get. Some of the other photographers are still working out the details on what they will be showing and how large their prints will be. The wall I think I might get will give me up to a 12 ft view. As you say, 6ft would be the minimum.

The files were supposed to go to the printer yesterday, but with the 16" snowfall overnight, I suspect little was done.

[DanK]

The 360 dpi is determined by the print head, not the intended viewing distance. That’s the Epson standard. Canon printers use 300 dpi. You can get noticeable degradation if the printing software doesn’t match the printer’s native resolution. I never change this, regardless of the size of the print or the viewing distance.


Sent from my iPhone using Tapatalk

[xpatUSA]

The 360 dpi is determined by the print head, not the intended viewing distance. That’s the Epson standard. Canon printers use 300 dpi. You can get noticeable degradation if the printing software doesn’t match the printer’s native resolution. I never change this, regardless of the size of the print or the viewing distance.

To a non-printer that is most interesting. It means that you or your print driver "re-whatevers" your image size such that 360dpi gives you the desired printed image size. Better use Lanczos3 for that (just kidding).

But I do no different when I allow "fit to window" as opposed to viewing at 100% zoom ...

[DanK]

To a non-printer that is most interesting. It means that you or your print driver "re-whatevers" your image size such that 360dpi gives you the desired printed image size. Better use Lanczos3 for that (just kidding).

But I do no different when I allow "fit to window" as opposed to viewing at 100% zoom ...

Imagine trying to send an image with 243 dpi (an arbitrary number) to a printer that has a head with 360 dpi. You can see the problem.

I did try using 600 dpi on my printer, and it caused a slight but noticeable degradation, so the problem is more than just having the number of nozzles being an even mutliple of the image pixel count.

[xpatUSA]

Imagine trying to send an image with 243 dpi (an arbitrary number) to a printer that has a head with 360 dpi. You can see the problem.

I can indeed! No different to viewing something "fit to window" on my monitor with some unknown algorithm applied by my browser or viewer. A sad and sorry restriction, IMHO.

But still, for the 6 ft viewing distance agreed by Manfred, the ppi need be no greater than 1/(72" x 0.29i*10^-3rad) = about 48ppi or, for eagle-eyes, 96ppi which is conveniently close to 90ppi, an integer ratio in the
Epson.

Assuming that 90ppi is possible, would that cause a "noticeable degradation", assuming that the Viewer is not a foot away with magnifying glass in hand ...

I realize that printing at 90ppi is anathema to "most of us" - so please consider the comment as theoretical!

I did try using 600 dpi on my printer, and it caused a slight but noticeable degradation, so the problem is more than just having the number of nozzles being an even mutliple of the image pixel count.

Perhaps "even multiple" says it right there, 90 being one-eighth of 720 ...

[xpatUSA]

II did try using 600 dpi on my printer, and it caused a slight but noticeable degradation, so the problem is more than just having the number of nozzles being an even mutliple of the image pixel count.

Sorry, I'm struggling with "more than just having the number of nozzles being an even mutliple of the image pixel count".

Why is printing at other than 720dpi, 360, 180, etc., not a problem? Is the number of nozzles for 600dpi an even multiple of the Canon-referred pixel count?

[DanK]

Sorry, I'm struggling with "more than just having the number of nozzles being an even mutliple of the image pixel count".

Why is printing at other than 720dpi, 360, 180, etc., not a problem? Is the number of nozzles for 600dpi an even multiple of the Canon-referred pixel count?

Canon's native pixel count is 300 dpi, not 360. Epson's is 360.

[xpatUSA]

Canon's native pixel count is 300 dpi, not 360. Epson's is 360.

I fold. I was wondering which you use - wasn't clear to me.

[Round Tuit]

Just to throw my 2 cents worth into this conversation, I think that you are mixing up dpi, ppi and "native resolution". To be clear: pixel per inch (ppi) is a characteristic of the image, dot per inch is a characteristic of the printer and the "native resolution" of a printer is expressed in dpi.

The native resolution of my Canon printer is 4800 x 2400 DPI. I don't know which one is the horizontal and which one is the vertical dimension. I presume that it takes, as a minimum, 8 dots to make a pixel. So the maximum pixel density that my Canon printer can output is 300 ppi.

I see no reason why a Canon printer could not print high quality prints at some lower pixel density than 300 ppi but not at any arbitrary number. Mind you, that would create bigger pixels which would give the same result as up-sampling using the next neighbor algorithm.

[ajax]

Looks like discussion has turned to printing which is of great interest to me. As a result of discussion to date I went and looked up the specifications for my Canon Pixma Pro 100 printer (specs a couple of clicks away) and found the following which might be relevant:

Number of Nozzles
6,144

Print Resolution (Up to)
Color: Up to 4800 x 2400 dpi4Black Up to 4800 x 2400 dpi4

Footnote #4. Resolution may vary based on printer driver setting. Color ink droplets can be placed with a horizontal pitch of 1/4800 inch at minimum.

Ink Droplet Size
Picoliter Size 3pl

Ink Capacity
8

From what I found a picoleter is one trillionth of a liter (pretty small amount) and the 8 must refer to the number of ink cartridges of different colors.

I don’t see anything about 300dpi although it is a value commonly used in postings on various websites. Is it possible that the meaning of dpi varies. For example, in above measure of print resolution might it mean droplets per inch. It appears that some uses of dpi mean dots per inch. Is it possible that dots are different than droplets? Might a dot refer to the printer's rendering of an actual color (somewhat like a pixel)?, which certainly requires multiple droplets. My own thought is that based on the range of colors represented by pixels the number of droplets involved in representing the color associated with a pixel must differ a fair amount.

Bottom line is that I now realize I don’t really have a very good grasp for how ink jet printers work. Surprise! Surprise! However, it is also pretty clear the software (printer driver) used to operate the printer has to have lots of image scaling (if NOT sampling) capability in order to handle all of the possibilities for image resolution that it needs to match with paper size.

The question of most interest is what can we as producers of the image file do to limit the amount of transformation that the printer driver does that might have the unintended affect of reducing resulting picture quality. I think the suggestion herein was that by producing a 300ppi (pixel per inch) image file for a Canon printer this might happen. Yes?

[DanK]

The problem with interpreting the specs for printers is that "dot" and "droplet" have several different meanings.

The only number you need is the native print resolution to set in Lightroom, which is 300 dpi. The other parameter that affects this is the print quality setting in the printer's firmware (access by printer properties), but there it's masked by verbal descriptions. The highest quality will be obtained by setting that to "highest quality". I read somewhere that this uses the nozzle pitch of the print head, which is 600 dpi in my case, but you don't need to worry about it--just set it to the highest quality setting.

Some time ago, out of curiosity, I printed two copies of a few prints with identical settings save one. I set the firmware to highest quality and then printed setting LR first at 300dpi and then at 600 dpi. The 300 dpi prints were sharper. The differences were small, but on close examination, they were apparent.

So since then, I have simply set all canon printers to highest quality in printer properties and to 300 dpi in Lightroom. I really think that's all you need to know. the droplet size and number of nozzles doesn't enter into any user decisions.

[xpatUSA]

Looks like discussion has turned to printing which is of great interest to me. As a result of discussion to date I went and looked up the specifications for my Canon Pixma Pro 100 printer (specs a couple of clicks away) and found the following which might be relevant:

Number of Nozzles
6,144

Print Resolution (Up to)
Color: Up to 4800 x 2400 dpi #4 Black Up to 4800 x 2400 dpi #4

Footnote #4. Resolution may vary based on printer driver setting. Color ink droplets can be placed with a horizontal pitch of 1/4800 inch at minimum.

Ink Droplet Size
Picoliter Size 3pl

Ink Capacity
8

From what I found a [picoliter] is one trillionth of a liter (pretty small amount) and the 8 must refer to the number of ink cartridges of different colors.

I don’t see anything about 300dpi although it is a value commonly used in postings on various websites. Is it possible that the meaning of dpi varies. For example, in above measure of print resolution might it mean droplets per inch. It appears that some uses of dpi mean dots per inch. Is it possible that dots are different than droplets? Might a dot refer to the printer's rendering of an actual color (somewhat like a pixel)?, which certainly requires multiple droplets. My own thought is that based on the range of colors represented by pixels the number of droplets involved in representing the color associated with a pixel must differ a fair amount.

Equally, nowhere in my Canon MG8120 manual does it mention a "native resolution" or a recommended image ppi setting. So 300 ppi must have been deduced somehow. Not easy when three of the six cartridges effect 512 dpi and three effect 1536 dpi via the print head - and the big number 6144dpi horizontally is all six added together!

Bottom line is that I now realize I don’t really have a very good grasp for how ink jet printers work. Surprise! Surprise! However, it is also pretty clear the software (printer driver) used to operate the printer has to have lots of image scaling (if NOT sampling) capability in order to handle all of the possibilities for image resolution that it needs to match with paper size.

The question of most interest is what can we as producers of the image file do to limit the amount of transformation that the printer driver does that might have the unintended [effect] of reducing resulting picture quality.

Un-intended? The driver does what it does and we can not affect that.

I think the suggestion herein was that by producing a 300ppi (pixel per inch) image file for a Canon printer this might happen. Yes?

No. There was confusion as to which Manufacturer was being referred to by the examples of 300ppi or 600ppi.

[Round Tuit]

The problem with interpreting the specs for printers is that "dot" and "droplet" have several different meanings.

The only number you need is the native print resolution to set in Lightroom, which is 300 dpi. The other parameter that affects this is the print quality setting in the printer's firmware (access by printer properties), but there it's masked by verbal descriptions. The highest quality will be obtained by setting that to "highest quality". I read somewhere that this uses the nozzle pitch of the print head, which is 600 dpi in my case, but you don't need to worry about it--just set it to the highest quality setting.

Some time ago, out of curiosity, I printed two copies of a few prints with identical settings save one. I set the firmware to highest quality and then printed setting LR first at 300dpi and then at 600 dpi. The 300 dpi prints were sharper. The differences were small, but on close examination, they were apparent.

So since then, I have simply set all canon printers to highest quality in printer properties and to 300 dpi in Lightroom. I really think that's all you need to know. the droplet size and number of nozzles doesn't enter into any user decisions.

Dan, I agree with you that the number to use for the print resolution in the Lightroom and Photoshop print modules is 300 for the Pixma Pro printers. This is an educated guess on my part based on the printer resolution of 2400 dpi. However I must point out that both programs specify the resolution in pixel per inch not dpi.

[LenR]

The question of Canon's 300/600 or Epson's 360/720 has been debated on the forum in the past.
David may find the attached by Jeff Schewe to be of interest.

https://www.digitalphotopro.com/tech...-resolution/2/

[Manfred M]

The question of Canon's 300/600 or Epson's 360/720 has been debated on the forum in the past.
David may find the attached by Jeff Schewe to be of interest.

https://www.digitalphotopro.com/tech...-resolution/2/

Unfortunately, the article is over 11 years old and some of the material is relevant (the parts regarding resolution and viewing distance), but others (especially how the printer drivers work) is quite out of date. The sophistication of that aspect of how photo printers work has improved significantly and a lot of the old "rules of thumb" need to be revisited.

[DanK]

Interesting. I didn’t record the details of the test I did—the size of the image or of the print. Schewe’s article suggests that 600 dpi might be better if I used file that’s large relative the print size. Maybe someday I’ll replicate the test that way. However, it’s a very low priority. I get consistently fine results at 300.


Sent from my iPad using Tapatalk

[ajax]

The question of Canon's 300/600 or Epson's 360/720 has been debated on the forum in the past.
David may find the attached by Jeff Schewe to be of interest.

https://www.digitalphotopro.com/tech...-resolution/2/

Yes! Quite helpful for me even if outdated.