Forgot Password
Pentax Camera Forums Home
 

Reply
Show Printable Version Search this Thread
01-11-2022, 06:22 AM   #16
Pentaxian




Join Date: Nov 2010
Location: Pacific Northwest
Photos: Gallery
Posts: 913
I've used the Topaz DeNoise AI and the Sharpen AI here and there, and there have been a few photos that I was able rescue. Not great photos, mind you, but photos that, like normhead's, I considered to be rubbish but now I think they're ok. The DeNoise AI program is agonizingly slow on my machine (like 20 minutes for a single photo) so I use it very rarely... I'll continue to use those plug-ins once in a while, but probably mostly with old photos that I've forgotten about because they sucked (but that I kept anyway!).

01-11-2022, 06:33 AM   #17
Digitiser of Film
Loyal Site Supporter
BigMackCam's Avatar

Join Date: Mar 2010
Location: North East of England
Posts: 20,574
I'm uncertain how I feel about this. On the one hand, I like the full character of any lens I'm using to show through in my images. If it's soft and dreamy wide open, I want the image to be soft and dreamy... If it's sharp in the centre but noticeably softer in the borders and corners, I want to see that. If I want images with shallow depth-of-field that a sharp and contrasty across the frame, I feel I should be using a lens that performs that way...

... and yet, on the other hand, I contradict myself somewhat, because I happily apply sharpening - albeit quite subtle - to my raw images, I remove chromatic aberrations / fringing if they're especially troublesome, I generally add a little local and global contrast, and will sometimes fine-tune the tone curve. In all these adjustments (and more), I'm messing with the true rendering of the lens.

... so I'm not sure why I should be uncomfortable with the idea AI sharpening, but I am. I'm not crystal clear on my reasons just yet...
01-11-2022, 07:15 AM - 2 Likes   #18
Pentaxian
normhead's Avatar

Join Date: Jun 2007
Location: Near Algonquin Park
Photos: Gallery | Albums
Posts: 40,442
Original Poster
QuoteOriginally posted by BigMackCam Quote
.. so I'm not sure why I should be uncomfortable with the idea AI sharpening, but I am. I'm not crystal clear on my reasons just yet...
I remember an older Pentax exec. interview when they talked about softening some of their telephotos a bit so that bird images would look more natural. I'm still aware of that. A few years I stopped sharpening bird images at all and went to micro-contrast instead, which gave a much more natural type image.

Then the last few weeks while applying Topaz AI I noticed I'm getting a lot more likes , here on Flickr and on Facebook. There is distinct social pressure in favour of sharpening.

An un-AI-sharpened Jay


The image is plenty sharp enough....AI Sharpening actually changes the image.


It just looks different, and not as much like the bird. But, based on my limited time with the software, more people prefer the sharpened image than the un-AI-Sharpened one. I'd be amazed if a poll showed anything else. I agree with the Pentax engineers, the second looks overly sharp, the feathers look like wires, not feathers.

I suspect part of what has happened here is the Topaz people have done some research into what people interpret as sharpness, and catered to it. You cannot see detail like the second image sitting in the blind watching the bird hop about. And yet people seem to find it more pleasing than the softer image. The Topaz AI image is as much a filter that give the image a certain look. In my mind it's gone way beyond sharpening. It's crafted micro-contrast where none existed. I believe it has done this largely by taking the soft edges around physical features, and taking the transitions between the hard edges, and making them bigger by turning the transition values into one of the two solid values on either side. This increases both the thickness of the fibres, and the apparent contrast by removing transition values. I have also seen the software take an out of focus lettering on the side of a truck that was unreadable and make it legible. It has to have something to work with, I suspect part of it's work is to remove transition values while emphasizing the more solid colours.

Given that analysis I'm totally on the fence. I believe the Pentax approach is more in keeping with nature, but I think the Topaz people have done a really good job of researching how people perceive sharpness and helping create the images many people like to see. It's a conundrum. And having gone off sharpening for about about 5 years and doing everything possible to avoid what I consider to be an artificially sharpened look, I'm torn. Feathers should like like feathers, not wire constructions, but it would seem, a lot of people prefer the wire construction, what's person to do?

All that being said, I'm thinking most of what I send to the printers will soon be sharpened. The sharpening doesn't do anything for sunsets and pastel scenes. You have to be careful not to create artifacts. But I suspect for images like the above, it will make more compelling prints. Even if it's art, not documentation. For digital images, I'm not sure it would be worth the effort. It generally takes me about 3-5 minutes to apply this "filter." Save as full size tiff, sharpen or d-noise, load the image into preveiw, resize and convert to jpeg. I recently archived my best 52 images of 2021 and sharpened every one that went into that folder,(except for the sunsets). so I've started understanding what I'm doing. Nothing increases understanding like repeating the process on different images. But I'm still a little on the fence about when I should do it. I suspect everything I post from now on will be at least run through the software, even if I use very low values more in keeping with my own philosophy or as in most sunsets, choose not to sharpen at all. Sharpening on water ripples is just wrong. But for wildlife animal , streetscape, etc. I think over -sharpened has become a style. Anyone interested in selling their work is probably going to have to get on the bandwagon or suffer.

Last edited by normhead; 01-11-2022 at 07:51 AM.
01-11-2022, 07:16 AM   #19
Closed Account




Join Date: Feb 2019
Photos: Gallery | Albums
Posts: 819
The results are interesting. There's definitely more definition in the fur, less blotchiness. There's also slightly less contrast overall and a slightly unnatural colour cast, I suspect trying to improve either might negate some of the recovered detail. Like I say, interesting, definitely worth keeping an eye on, not enough to convince me to rush out and buy it but I'm thinking that one day those grainy old negs I'm copying from 30 years ago might just look a bit sharper and more saturated in a future AI version.

01-11-2022, 07:32 AM   #20
Pentaxian
normhead's Avatar

Join Date: Jun 2007
Location: Near Algonquin Park
Photos: Gallery | Albums
Posts: 40,442
Original Poster
QuoteOriginally posted by swanlefitte Quote
A lens good on an optical bench test is independent of the camera. The new lenses often turn out exceptional on the optical bench test.
Often being the word I find distressing. In the age of computer design, I would expect all of them to be excellent. So, what would interest me would not be that some new designs are excellent, it would be compared to the past, what percentage are excellent, at each price point.

QuoteOriginally posted by StiffLegged Quote
My conclusion is you can polish fox droppings all you like, but they’re still fox droppings.
And then there's that too...
A friend of mine shooting a K10D for his portrait studio/weddings accidentally left his camera in MF and shot a whole wedding, with the focus of every image too soft. Software like this absolutely saved his bacon. His clients weren't able to tell, he had botched the images. SO while I understand the sentiment, sometimes you just have to have something, and if you can make the fox turd taste like steak, are you really going got let on it's fox turd?

It's a moral dilemma.
Do you give people what is true to your process. Or do you create what they want?

Last edited by normhead; 01-11-2022 at 07:47 AM.
01-11-2022, 07:47 AM   #21
Pentaxian




Join Date: Feb 2015
Photos: Gallery
Posts: 12,177
QuoteOriginally posted by StiffLegged Quote
No, a crummy microphone will not properly record both strident brass and subtle harp sounds simultaneously and no amount of equaliser is going to reproduce both from your loudspeakers. The dynamic range necessary was beyond the crummy microphone’s ability to record.
Where's the limit that separate a good microphone from a bad one, a good loudspeaker from a bad one, a good lens from a bad lens, a good camera from a bad camera? Is it binary, 0 or 1 , black or white?
01-11-2022, 07:54 AM   #22
Site Supporter
Site Supporter




Join Date: Feb 2009
Location: Cork
Photos: Albums
Posts: 1,882
QuoteOriginally posted by biz-engineer Quote
Where's the limit that separate a good microphone from a bad one, a good loudspeaker from a bad one, a good lens from a bad lens, a good camera from a bad camera? Is it binary, 0 or 1 , black or white?
Taste and experience. If you haven't experienced anything above mediocre then you're happy with mediocre, if you experience better then you recalibrate.

01-11-2022, 07:55 AM   #23
Pentaxian
normhead's Avatar

Join Date: Jun 2007
Location: Near Algonquin Park
Photos: Gallery | Albums
Posts: 40,442
Original Poster
QuoteOriginally posted by 3by2 Quote
The results are interesting. There's definitely more definition in the fur, less blotchiness. There's also slightly less contrast overall and a slightly unnatural colour cast, I suspect trying to improve either might negate some of the recovered detail. Like I say, interesting, definitely worth keeping an eye on, not enough to convince me to rush out and buy it but I'm thinking that one day those grainy old negs I'm copying from 30 years ago might just look a bit sharper and more saturated in a future AI version.
You'd probably like the DeNoise software. It's actually incredible. For noisy images I run them through DeNoise then Sharpening, and that produces a much more pleasing result than using the noise inhibiting function in Sharpening AI. Denoise kills the sharpening somewhat, but it would seem Sharpening is more than capable of bring it back and emphasizing it.
01-11-2022, 07:55 AM   #24
Site Supporter
Site Supporter
luftfluss's Avatar

Join Date: Jun 2011
Location: NJ
Photos: Gallery | Albums
Posts: 11,595
QuoteOriginally posted by normhead Quote
I...There is distinct social pressure in favour of sharpening...
This entire post was great, Norm, but especially this line.


And I feel it, too.
01-11-2022, 08:10 AM   #25
Pentaxian
normhead's Avatar

Join Date: Jun 2007
Location: Near Algonquin Park
Photos: Gallery | Albums
Posts: 40,442
Original Poster
QuoteOriginally posted by StiffLegged Quote
No, a crummy microphone will not properly record both strident brass and subtle harp sounds simultaneously and no amount of equaliser is going to reproduce both from your loudspeakers.
IN this case, I think this software takes features that might not otherwise be visible and emphasizes it in away that is in line with our perceptive abilities. With modern photography and resolutions there is almost always detail to fine to be observed in digital images or prints.

I have some images of this truck or one like it, that were so soft, the lettering was illegible. The sharpening software made it easily legible. The sharpening software actually made something visible not visible in the original file. SO which was the most accurate image? I could read the lettering in real life. I couldn't because of the low light and narrow depth of field, I could after I sharpened the image. Things are never black and white. To me, the topaz sharpening doesn't create, but emphasizes, and organizes, nd take advantage of the probability that in every images, there is detail we can't actually see, where our eyes need a little help to see what's there.
01-11-2022, 08:37 AM   #26
Loyal Site Supporter
Loyal Site Supporter
MossyRocks's Avatar

Join Date: Nov 2017
Location: Minnesota
Photos: Gallery | Albums
Posts: 2,978
From what I've seen of the sharpening abilities the results seem to be able to be replicated in RawTherapee using contrast by level and the various wavelet editing options. For me I like the contrast by levels and wavelet options and will probably stick with them since I have become much more competent in using them and as my abilities have improved I have been able to use them to really sort things out with astro shooting even beyond just sharpening.
01-11-2022, 08:48 AM   #27
Pentaxian
normhead's Avatar

Join Date: Jun 2007
Location: Near Algonquin Park
Photos: Gallery | Albums
Posts: 40,442
Original Poster
QuoteOriginally posted by MossyRocks Quote
From what I've seen of the sharpening abilities the results seem to be able to be replicated in RawTherapee using contrast by level and the various wavelet editing options. For me I like the contrast by levels and wavelet options and will probably stick with them since I have become much more competent in using them and as my abilities have improved I have been able to use them to really sort things out with astro shooting even beyond just sharpening.
If I had Raw Therapy and was any good with it, I'd do a comparison.
01-11-2022, 09:33 AM   #28
Site Supporter
Site Supporter
c.a.m's Avatar

Join Date: Jan 2013
Location: Ottawa, Canada
Photos: Gallery | Albums
Posts: 4,172
QuoteOriginally posted by normhead Quote
It just looks different, and not as much like the bird. But, based on my limited time with the software, more people prefer the sharpened image than the un-AI-Sharpened one. I'd be amazed if a poll showed anything else. I agree with the Pentax engineers, the second looks overly sharp, the feathers look like wires, not feathers.
@normhead, this is an interesting and relevant thread. Thanks for sharing your observations and thoughts.


Not to criticize the second, sharpened image of the Blue Jay, but I noticed some strange artifacts in the highest-res version at your Flickr. Seems to be some funny business going on at the peanut directly below the bird and at the snow in the left side of the image. In the short time you've played with the AI routine, have you noticed much 'collateral damage' in images?

My (limited) understanding of current AI-sharpening functions is that they can work reasonably well on main subjects, but can also introduce secondary flaws into the images.

- Craig
01-11-2022, 09:41 AM   #29
Site Supporter
Site Supporter
acoufap's Avatar

Join Date: Jan 2011
Location: Munich, Germany
Photos: Gallery | Albums
Posts: 4,161
One always have to pay attention the mood and feeling that smoothness can create or the other way around sharpness can destroy.
01-11-2022, 09:47 AM   #30
Pentaxian
normhead's Avatar

Join Date: Jun 2007
Location: Near Algonquin Park
Photos: Gallery | Albums
Posts: 40,442
Original Poster
QuoteOriginally posted by c.a.m Quote
@normhead, this is an interesting and relevant thread. Thanks for sharing your observations and thoughts.


Not to criticize the second, sharpened image of the Blue Jay, but I noticed some strange artifacts in the highest-res version at your Flickr. Seems to be some funny business going on at the peanut directly below the bird and at the snow in the left side of the image. In the short time you've played with the AI routine, have you noticed much 'collateral damage' in images?

My (limited) understanding of current AI-sharpening functions is that they can work reasonably well on main subjects, but can also introduce secondary flaws into the images.

- Craig
You really have to be careful with the artifacts for sure. I just took a shot out my kitchen window of a flock of birds, to get the plane of focus birds sharp, I had to produce artifacts on the edge birds, it's an issue. You won't see the image posted here. I used the Tamron 300 2.8, with the 2x. It's always soft on the K-3, much better on the K-1 so I thought I'd give it ago. I think I might need different strategy.

The idea of masking the main subject, sharpening it, then doing the rest at different settings might be appealing. But right now, I'd rather just reshoot.
Reply

Bookmarks
  • Submit Thread to Facebook Facebook
  • Submit Thread to Twitter Twitter
  • Submit Thread to Digg Digg
Tags - Make this thread easier to find by adding keywords to it!
ai, camera, correction, flickr, head, images, k-mount, lens, lenses, look, noise, norm, pentax, pentax lens, reduction, results, slr lens, software
Thread Tools Search this Thread
Search this Thread:

Advanced Search


Similar Threads
Thread Thread Starter Forum Replies Last Post
Hello again, and again.. and again..... ghigoblin Welcomes and Introductions 9 01-11-2017 02:22 PM
For Sale - Sold: SMC D FA MACRO 1:2.8 100 mm WR -----PRICE DROPPED again- again-again-again watchman323 Sold Items 12 12-09-2013 11:18 AM
Purple fringes, and trees, again, and again, and again. Calmsea Pentax K-5 & K-5 II 39 08-16-2013 02:27 PM
Can Nikon Ai and/or non Ai lenses be adapted to work on the K mount? Vantage-Point Canon, Nikon, Sony, and Other Camera Brands 3 07-28-2013 08:25 PM
Modern Media /Modern Minds seacapt General Talk 24 09-23-2010 03:55 PM



All times are GMT -7. The time now is 01:51 AM. | See also: NikonForums.com, CanonForums.com part of our network of photo forums!
  • Red (Default)
  • Green
  • Gray
  • Dark
  • Dark Yellow
  • Dark Blue
  • Old Red
  • Old Green
  • Old Gray
  • Dial-Up Style
Hello! It's great to see you back on the forum! Have you considered joining the community?
register
Creating a FREE ACCOUNT takes under a minute, removes ads, and lets you post! [Dismiss]
Top