Printing large

[Erik Kaffehr]

Well, one would never view a 30x40 at the same distance as an 8x10, for instance. Every print size has an optimum viewing distance.

I have found that, whenever I see my prints on the walls of an exhibition space, that people view from a distance, then move in close, then far away again.

Jim,
What would be your assessment in as far as how large one can print with a 20mp file? I'm still trying to get a handle on it and while the discussion is interesting, it dances around the issue. Given the image that I posted, how large can that be printed that would withstand close inspection at a gallery exhibition?

As I said, that image isn’t that sharp to begin with. An unsharp original can mean that you can print the image as large as you like. I demonstrated that in the blog post referenced in this thread. But that’s not the case here. If that were my image, I don’t think I’d want to print it more than 15 inches high.

Thanks for that explanation. Would you have some feeling for the cause of that unsharpness?

Best regards

Erik

 

[Rob de Loe]

My question relates to interpolation options. If you interpolate the 20mp image, how big can you prinyt it and looj at it from a relatively close distance?

I see where you're going now. I don't know that your question can be answered in anything but a subjective way.

I think I've said this already, but why not just make some prints and see for yourself? Nobody else's opinion on an Internet forum can come close to that as an answer to your question.

Given that it's the direction this is all going, in doing your own testing you probably should try Gigapixel AI and decide how you feel about what it does. Personally, I don't use it because I don't think it's as amazing as boosters suggest, but more importantly, I think we cross a line when we invent information that wasn't there to begin with. I realize what a giant rat hole this is and have seen all the arguments for why my position rests on a shaky foundation. Nonetheless, I personally can't get excited about detail in a print that was made up by an algorithm.

 

[edispics]

If you're asking about interpolation, a good way might be to get a free trial version of Topaz Gigapixel AI (if your hardware will run it effectively) and trying scaling of various amounts and check the results. Free and fast way to evaluate what interpolation might yield.

 

[edispics]

If you're asking about interpolation, a good way might be to get a free trial version of Topaz Gigapixel AI (if your hardware will run it effectively) and trying scaling of various amounts and check the results. Free and fast way to evaluate what interpolation might yield.

I used the sample you provided and scaled by factor of 2 using all auto settings and man made as the major subject type:

View attachment bb7cacff4dc84bfb8195f07fd63fe04d.jpg

Jim is right, of course, the original is soft. But here is the result of a 2 times scaling. You and others can see the results. Using the man made subject mode, as used here, is much more aggressive with hard edges than the natural mode. The result is 11232 by 7488.

 

[JimKasson]

I missed that post where you commented on the image.

See this:

https://www.dpreview.com/forums/post/64277537

Assuming it was sharp enough, how big would you print it?

Maybe 24 inches high, but it would be a little soft unless I used something like GigaPixel AI, which invents detail.

 

[JimKasson]

Well, one would never view a 30x40 at the same distance as an 8x10, for instance. Every print size has an optimum viewing distance.

I have found that, whenever I see my prints on the walls of an exhibition space, that people view from a distance, then move in close, then far away again.

Jim,
What would be your assessment in as far as how large one can print with a 20mp file? I'm still trying to get a handle on it and while the discussion is interesting, it dances around the issue. Given the image that I posted, how large can that be printed that would withstand close inspection at a gallery exhibition?

As I said, that image isn’t that sharp to begin with. An unsharp original can mean that you can print the image as large as you like. I demonstrated that in the blog post referenced in this thread. But that’s not the case here. If that were my image, I don’t think I’d want to print it more than 15 inches high.

Thanks for that explanation. Would you have some feeling for the cause of that unsharpness?

Dunno. I didn't look at the EXIF, but in a scene like that, atmospherics, focus, and diffraction are the usual suspects.

OK, now I've looked at the EXIF. Scratch diffraction, and add camera shake.

--
https://blog.kasson.com

 

[JimKasson]

My question relates to interpolation options. If you interpolate the 20mp image, how big can you prinyt it and looj at it from a relatively close distance?

I see where you're going now. I don't know that your question can be answered in anything but a subjective way.

I think I've said this already, but why not just make some prints and see for yourself? Nobody else's opinion on an Internet forum can come close to that as an answer to your question.

Given that it's the direction this is all going, in doing your own testing you probably should try Gigapixel AI and decide how you feel about what it does.

I've already posted a GigaPixel AI resampling of Macro's test image in this thread.

Personally, I don't use it because I don't think it's as amazing as boosters suggest, but more importantly, I think we cross a line when we invent information that wasn't there to begin with.

It is a line, but any resampler does that to some extent.

I realize what a giant rat hole this is and have seen all the arguments for why my position rests on a shaky foundation. Nonetheless, I personally can't get excited about detail in a print that was made up by an algorithm.

I can get excited about anything that makes the image better. GigaPixel Ai does that sometimes.

 

[Macro guy]

My question relates to interpolation options. If you interpolate the 20mp image, how big can you prinyt it and looj at it from a relatively close distance?

I see where you're going now. I don't know that your question can be answered in anything but a subjective way.

I think I've said this already, but why not just make some prints and see for yourself? Nobody else's opinion on an Internet forum can come close to that as an answer to your question.

Given that it's the direction this is all going, in doing your own testing you probably should try Gigapixel AI and decide how you feel about what it does.

I've already posted a GigaPixel AI resampling of Macro's test image in this thread.

Personally, I don't use it because I don't think it's as amazing as boosters suggest, but more importantly, I think we cross a line when we invent information that wasn't there to begin with.

It is a line, but any resampler does that to some extent.

I realize what a giant rat hole this is and have seen all the arguments for why my position rests on a shaky foundation. Nonetheless, I personally can't get excited about detail in a print that was made up by an algorithm.

I can get excited about anything that makes the image better. GigaPixel Ai does that sometimes.

I had a bit of time to analyze the various files and I have gotten a pretty good idea of what can be done and how large a good file can be printed.

Thanks to all of you guys for your input and thanks Jim for taking the time to make the larger files using various techniques.

This has been invaluable.

 

[Erik Kaffehr]

Thanks!

I looked at the EXIF, too.

Best regards

Erik