I took this picture of Lower Falls on the Lewis River in fall of 2018. I liked the picture and did post it here back then.
Collage,com had run some specials before at 90% off and I've very good results from them.
Another 90% off sale came and I decided to see how big I could go and still have acceptable results.
I developed the picture in PDCU with some minor adjustments.
Then I used
Topaz Studio 2
Topaz Adjust AI
Topaz Sharpen AI
The result was incredible. A gorgeous 48 x 32 canvas that is so tack sharp and beautiful that it's hard to believe. It printed exactly as it appeared on my screen (ColorMunki adjusted).
No sign of any pixels 32 by 48 inches, even from 12 inches away.
This camera still amazes me and Collage .com has done beautiful work (and who can pass up 90% off). With tax and shipping it cost $60 total.

[K2 to K50]

I took this picture of Lower Falls on the Lewis River in fall of 2018. I liked the picture and did post it here back then.
,,,,,,,,,,,,
The result was incredible. A gorgeous 48 x 32 canvas that is so tack sharp and beautiful that it's hard to believe. It printed exactly as it appeared on my screen (ColorMunki adjusted).

Well, if your big canvas looks even half as good as this compressed downsized version we see here, it must be spectacular. Lush greens, beautiful cascading water curtains, the trees at the back coming right down to the water, and nicely framed with the foreground bushes either side, definitely a wall hanger - and preferably somewhere conspicuous!

Very nice work.

[nocturnal]

Nice one!

There is nothing like a vignetted landscape, I was editing similar leafy green waterfalls yesterday where a medium-heavy vignette really works.

This is how we see, the human eye is vignetted. This is why I have no time for pixel peepers criticising corner sharpness etc.

Who looks at close up corners of a print? No normal human being does this!

[JimmyL]

Thanks for the kind words. Yes, when you popped the vignette on and off there was no question on which way to go.

[JensE]

Congrats to a great result. Always a thrill, I'm nervous when submitting large printing orders. I love many of my K5+18-135mm pictures, just had some printed on this year's A2-sized calendar. I ordered two (a special), an equivalent of 163" x 66" printed area, with only a soft-proof based on my monitor calibration, printed on a new to me printer-paper combination where the softproof profile was all I had. Luckily it worked out well.


My widest canvas print at home is 130cm+sides wide (~54in) and that was taken with just a basic 12MP DSLR and kit lens - fully open. Canvas doesn't require a lot of resolution, smooth surfaces are a lot more demanding.
Did you actually scale up the picture with of those 'intelligent' Topaz products? If so, it would be nice if you could share a comparison of a crop to the native resolution. I could imagine that it can help to get natural-looking results.

[JimmyL]

Thanks; No, I didn't upscale it with Gigapixel to print. I tried upscaling with Gigapixel and the result wasn't as good as the original on my monitor. But, Topaz Sharpen AI worked great. So I just sent the file in and hoped for the best.

[dms]

Actually at f/22 it is likely about 1/2 resolution (due to diffraction), of what was possible at f/11. But it still looks fine, which just shows that large enlargement can be fine, even if much more than suggested. Anyway the DOF likely was the more important consideration than absolute resolution. Also AI sharpen probably helped out, it seems to be quite "smart" even if not actually AI software. If it was AI it would run faster and/or better, if used a second time on the same image, which it does not do.

[JensE]

Also AI sharpen probably helped out, it seems to be quite "smart" even if not actually AI software. If it was AI it would run faster and/or better, if used a second time on the same image, which it does not do.

It could be AI software, if it was trained on a large number of manually edited pictures (before-after) and the resulting trained model shipped with the software.

[dms]

Yes, it is a question of semantics. But the matrix metering is similar in that it is based on thousands of photos to recognize when the pattern is for instance a person backlit at the beach, or the equivalent, and it is not (was not) advertised as artificial intelligence. For me AI means it learns as it goes forward.