Thanks for this information. I can't imagine living without Topaz SharpenAI now.
I recently got a notice from TopazLabs of a new pricing model. The claim was
that the previous model forced them to release new products in order to
generate sufficient revenue. Now the model goes to yearly update model on major
releases. (or something to that affect) Which will allow them to focus more on
improving existing products.
I suspect from the previous pricing model DenoiseAI versus SharpenAI used a
very similar engine as both have denoise and both have sharpen. The denoiseAI
may rely on the sharpenAI technology to determine noise and vice versa.
Releasing more than one product with similar underlying technology to create
revenue.
As for Georgia O'Keefe's reasons, you could also try some of the Topaz Studio 2
abstractions... :-)
Fun with Moose photo https://photos.app.goo.gl/GEoxdJ5dhN9eLooz5
Speculating WayneS
At 2/20/2020 01:19 PM, Moose wrote:
>In my usual way, I've been poking at the uses of various new digital darkrooms
>tools.
>
>I though I'd run into something quite interesting and useful. The Topaz AI
>products are neural networks (?) that are trained by feeding them lots and
>lots of bad and good versions of photos, until they learn to convert "bad" to
>"good".
>
>That means they can be quirky. For example, AI Gigapixel does amazing
>uprezzing on many images. But occasionally it runs into something outside its
>training, and does weird or even awful things to it. Another example, Denoise
>can do an awesome job on noise, except in low detail areas next to high detail
>areas - sometimes.
>
>There can also be an upside. I noticed something happening when I used Denoise
>AI. Roughly, some base ISO images treated to modest "Noise Reduction"
>settings, ZERO "Enhance Sharpness" and "Recover Details" = 5 or 10, had
>enhanced detail, sans ringing or artifacts.
>
>I have tried it on quite a number of files now, and find the effect ranges
>from good to spectacular.
><http://www.moosemystic.net/Gallery/tech/Process/TopazDenoiseAI_sharpen/TzDeNAIshrp.htm>
>
>If you don't have excellent eyesight, you may have to get close to some of
>these.
>
>I've intentionally left EXIF out, to focus on the software. There are three
>different sensor formats and five different cameras involved. All shot at base
>ISO - but one ringer at about 2.5 stops higher ISO. Space-time ranges from
>Bhutan in Fall 2017 to Berkeley a couple of days ago.
>
>Why mostly flowers? In addition to Georgia O'Keefe's reasons . . . They have
>lots of really fine details, texture, hues, and are 3D. Why 3D? You may notice
>in some that it has not only revealed lots of greater detail at the focal
>plane, but also modestly 'sharpened' OoF parts, with the effect diminishing
>with distance from focal plane; it can be a DoF improver.
>
>Where does all that detail come from? Does the app make it up? If so, how is
>it so accurate seeming? Does it do what deconvolution does, only from a
>completely new direction?
>
>I sure don't know. I do know that for a tele nut who often uses 100% crops of
>critters, it's magic!
>
>Sure Shot Moose
--
_________________________________________________________________
Options: http://lists.thomasclausen.net/mailman/listinfo/olympus
Archives: http://lists.thomasclausen.net/mailman/private/olympus/
Themed Olympus Photo Exhibition: http://www.tope.nl/
|