1. Forum
  2. Usenet
  3. REC.PHOTO.DIGITAL

  • Topaz Photo AI

    From Alfred Molon@21:1/5 to All on Thu Sep 22 21:10:40 2022
    Some tests with Photo AI (images shot with the Olympus E-M1 Mark II and
    III cameras):

    1. The kingfisher I shot in 2019 with the Olympus 75-300 lens at 300mm /
    ISO 6400.
    Left is my RAW processing result of 2019, right the same RAW just
    re-processed with Photo AI: https://www.molon.de/images/topaz_vs_dxo/kingfisher_photo_ai.png

    2. A red leaf monkey I shot last month in the Danum valley (ISO 6400,
    Olympus 100-400 at 400mm).
    Left is the out-of-camera JPEG, right the output of Photo AI: https://www.molon.de/images/topaz_vs_dxo/red_leaf_monkey_photo_ai.png

    Quite impressive detail recovery, especially in the monkey picture.

    The 75-300 is a relatively inexpensive tele lens, and ISO 6400 is not
    exactly an ISO level you should use with a micro 4/3 camera. But with
    modern software, no problem, you get a sharp and noiseless result.
    --
    Alfred Molon

    Olympus 4/3 and micro 4/3 cameras forum at
    https://groups.io/g/myolympus
    https://myolympus.org/ photo sharing site

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From HunterBD@21:1/5 to Alfred Molon on Thu Sep 22 20:57:59 2022
    On 22/09/2022 20:10, Alfred Molon wrote:
    Some tests with Photo AI (images shot with the Olympus E-M1 Mark II and
    III cameras):

    1. The kingfisher I shot in 2019 with the Olympus 75-300 lens at 300mm /
    ISO 6400.
    Left is my RAW processing result of 2019, right the same RAW just re-processed with Photo AI: https://www.molon.de/images/topaz_vs_dxo/kingfisher_photo_ai.png

    2. A red leaf monkey I shot last month in the Danum valley (ISO 6400,
    Olympus 100-400 at 400mm).
    Left is the out-of-camera JPEG, right the output of Photo AI: https://www.molon.de/images/topaz_vs_dxo/red_leaf_monkey_photo_ai.png

    Quite impressive detail recovery, especially in the monkey picture.

    The 75-300 is a relatively inexpensive tele lens, and ISO 6400 is not
    exactly an ISO level you should use with a micro 4/3 camera. But with
    modern software, no problem, you get a sharp and noiseless result.


    Thanks, Alfred. VERY impressive. 🙂

    I've just had a look here:-

    https://www.topazlabs.com/topaz-photo-ai

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From sobriquet@21:1/5 to Alfred Molon on Thu Sep 22 16:00:10 2022
    On Thursday, September 22, 2022 at 9:10:48 PM UTC+2, Alfred Molon wrote:
    Some tests with Photo AI (images shot with the Olympus E-M1 Mark II and
    III cameras):

    1. The kingfisher I shot in 2019 with the Olympus 75-300 lens at 300mm /
    ISO 6400.
    Left is my RAW processing result of 2019, right the same RAW just re-processed with Photo AI: https://www.molon.de/images/topaz_vs_dxo/kingfisher_photo_ai.png

    2. A red leaf monkey I shot last month in the Danum valley (ISO 6400,
    Olympus 100-400 at 400mm).
    Left is the out-of-camera JPEG, right the output of Photo AI: https://www.molon.de/images/topaz_vs_dxo/red_leaf_monkey_photo_ai.png

    Quite impressive detail recovery, especially in the monkey picture.

    The 75-300 is a relatively inexpensive tele lens, and ISO 6400 is not
    exactly an ISO level you should use with a micro 4/3 camera. But with
    modern software, no problem, you get a sharp and noiseless result.
    --
    Alfred Molon

    Olympus 4/3 and micro 4/3 cameras forum at
    https://groups.io/g/myolympus
    https://myolympus.org/ photo sharing site

    With AI these days you don't even need to take pictures.. AI can just mash up custom images from
    images found online.

    https://thispersondoesnotexist.com/

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Bill W@21:1/5 to Alfred Molon on Thu Sep 22 18:52:00 2022
    On Sep 22, 2022, Alfred Molon wrote
    (in article <Qq2XK.255858$[email protected]>):

    Some tests with Photo AI (images shot with the Olympus E-M1 Mark II and
    III cameras):

    1. The kingfisher I shot in 2019 with the Olympus 75-300 lens at 300mm /
    ISO 6400.
    Left is my RAW processing result of 2019, right the same RAW just re-processed with Photo AI: https://www.molon.de/images/topaz_vs_dxo/kingfisher_photo_ai.png

    2. A red leaf monkey I shot last month in the Danum valley (ISO 6400,
    Olympus 100-400 at 400mm).
    Left is the out-of-camera JPEG, right the output of Photo AI: https://www.molon.de/images/topaz_vs_dxo/red_leaf_monkey_photo_ai.png

    Quite impressive detail recovery, especially in the monkey picture.

    The 75-300 is a relatively inexpensive tele lens, and ISO 6400 is not
    exactly an ISO level you should use with a micro 4/3 camera. But with
    modern software, no problem, you get a sharp and noiseless result.

    I already had the first 3, and they gave Photo AI to everyone who had them already. The good results can be crazy good, and the bad results are just, “oh well”. It’s AI, so anything can happen. It’s just another tool
    but I think it’s a pretty good one overall. My only worry is the company itself. I have no idea how to get any support if I need it, and no idea if they’ll still be in business tomorrow.

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From RJH@21:1/5 to Bill W on Fri Sep 23 04:52:37 2022
    On 23 Sep 2022 at 00:52:00 BST, Bill W wrote:

    But with
    modern software, no problem, you get a sharp and noiseless result.


    Agreed, impressive.

    I already had the first 3, and they gave Photo AI to everyone who had them already. The good results can be crazy good, and the bad results are just, “oh well”. It’s AI, so anything can happen. It’s just another tool but I think it’s a pretty good one overall. My only worry is the company itself. I have no idea how to get any support if I need it, and no idea if they’ll still be in business tomorrow.

    It's the 'free updates for a year' bit, as well as the high (for me) intitial price. Effectively makes it a subsciption model.

    Is there something similar as freeware or relatively inexpensive?

    --
    Cheers, Rob, Sheffield UK

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From David Taylor@21:1/5 to Alfred Molon on Fri Sep 23 06:53:17 2022
    On 22/09/2022 20:10, Alfred Molon wrote:
    Some tests with Photo AI (images shot with the Olympus E-M1 Mark II and
    III cameras):

    1. The kingfisher I shot in 2019 with the Olympus 75-300 lens at 300mm /
    ISO 6400.
    Left is my RAW processing result of 2019, right the same RAW just re-processed with Photo AI: https://www.molon.de/images/topaz_vs_dxo/kingfisher_photo_ai.png

    2. A red leaf monkey I shot last month in the Danum valley (ISO 6400,
    Olympus 100-400 at 400mm).
    Left is the out-of-camera JPEG, right the output of Photo AI: https://www.molon.de/images/topaz_vs_dxo/red_leaf_monkey_photo_ai.png

    Quite impressive detail recovery, especially in the monkey picture.

    The 75-300 is a relatively inexpensive tele lens, and ISO 6400 is not
    exactly an ISO level you should use with a micro 4/3 camera. But with
    modern software, no problem, you get a sharp and noiseless result.

    Thanks for posting! Impressive, yes!

    With the second (Monkey) image, was Topaz processing the OOC RAW data, or the JPEG image shown on the left?

    The Web page seems to give little information on the product - will it process Panasonic RAW format as well as Olympus format?

    Sadly my graphics card is just an NVIDIA GeForce GT 710 with 2 GB, so it's not supported. A pity, but at least it saves me US $199/159!

    --
    Thanks,
    David
    Web: https://www.satsignal.eu

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From HunterBD@21:1/5 to sobriquet on Fri Sep 23 07:54:35 2022
    On 23/09/2022 00:00, sobriquet wrote:
    On Thursday, September 22, 2022 at 9:10:48 PM UTC+2, Alfred Molon wrote:
    Some tests with Photo AI (images shot with the Olympus E-M1 Mark II and
    III cameras):

    1. The kingfisher I shot in 2019 with the Olympus 75-300 lens at 300mm /
    ISO 6400.
    Left is my RAW processing result of 2019, right the same RAW just
    re-processed with Photo AI:
    https://www.molon.de/images/topaz_vs_dxo/kingfisher_photo_ai.png

    2. A red leaf monkey I shot last month in the Danum valley (ISO 6400,
    Olympus 100-400 at 400mm).
    Left is the out-of-camera JPEG, right the output of Photo AI:
    https://www.molon.de/images/topaz_vs_dxo/red_leaf_monkey_photo_ai.png

    Quite impressive detail recovery, especially in the monkey picture.

    The 75-300 is a relatively inexpensive tele lens, and ISO 6400 is not
    exactly an ISO level you should use with a micro 4/3 camera. But with
    modern software, no problem, you get a sharp and noiseless result.
    --
    Alfred Molon

    Olympus 4/3 and micro 4/3 cameras forum at
    https://groups.io/g/myolympus
    https://myolympus.org/ photo sharing site

    With AI these days you don't even need to take pictures.. AI can just mash up custom images from
    images found online.

    https://thispersondoesnotexist.com/

    Fascinating! :-D

    Did you explore this (pop-up)?

    Imagined by a GAN (generative adversarial network)
    StyleGAN2 (Dec 2019) - Karras et al. and Nvidia
    Don't panic. Learn how it works [1] [2] [3]
    Code for training your own [original] [simple] [light]
    Art • Cats • Horses • Chemicals • Contact me
    Another | Sponsor

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From HunterBD@21:1/5 to David Taylor on Fri Sep 23 08:32:56 2022
    On 23/09/2022 06:53, David Taylor wrote:
    On 22/09/2022 20:10, Alfred Molon wrote:
    Some tests with Photo AI (images shot with the Olympus E-M1 Mark II and
    III cameras):

    1. The kingfisher I shot in 2019 with the Olympus 75-300 lens at 300mm /
    ISO 6400.
    Left is my RAW processing result of 2019, right the same RAW just
    re-processed with Photo AI:
    https://www.molon.de/images/topaz_vs_dxo/kingfisher_photo_ai.png

    2. A red leaf monkey I shot last month in the Danum valley (ISO 6400,
    Olympus 100-400 at 400mm).
    Left is the out-of-camera JPEG, right the output of Photo AI:
    https://www.molon.de/images/topaz_vs_dxo/red_leaf_monkey_photo_ai.png

    Quite impressive detail recovery, especially in the monkey picture.

    The 75-300 is a relatively inexpensive tele lens, and ISO 6400 is not
    exactly an ISO level you should use with a micro 4/3 camera. But with
    modern software, no problem, you get a sharp and noiseless result.

    Thanks for posting!  Impressive, yes!

    With the second (Monkey) image, was Topaz processing the OOC RAW data,
    or the JPEG image shown on the left?

    The Web page seems to give little information on the product - will it process Panasonic RAW format as well as Olympus format?

    Sadly my graphics card is just an NVIDIA GeForce GT 710 with 2 GB, so
    it's not supported.  A pity, but at least it saves me US $199/159!


    FYI

    https://commons.wikimedia.org/wiki/File:This_person_doesn%27t_exist_Imagined_by_a_GAN_(generative_adversarial_network)_StyleGAN2_(Dec_2019).jpg

    I was tempted too, but it is FAR too expensive for a home user!

    --
    David

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Alfred Molon@21:1/5 to All on Fri Sep 23 18:56:27 2022
    Am 23.09.2022 um 07:53 schrieb David Taylor:
    With the second (Monkey) image, was Topaz processing the OOC RAW data,
    or the JPEG image shown on the left?

    In both cases Topaz processed the RAW files

    The Web page seems to give little information on the product - will it process Panasonic RAW format as well as Olympus format?

    I don't know which cameras are supported (although Panasonic should be supported). But you could give it a try.

    And yes, you need a modern and powerful PC for this software 8or at
    least a powerful graphics card).
    --
    Alfred Molon

    Olympus 4/3 and micro 4/3 cameras forum at
    https://groups.io/g/myolympus
    https://myolympus.org/ photo sharing site

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From RichA@21:1/5 to Alfred Molon on Fri Sep 23 23:08:20 2022
    On Thursday, 22 September 2022 at 15:10:48 UTC-4, Alfred Molon wrote:
    Some tests with Photo AI (images shot with the Olympus E-M1 Mark II and
    III cameras):

    1. The kingfisher I shot in 2019 with the Olympus 75-300 lens at 300mm /
    ISO 6400.
    Left is my RAW processing result of 2019, right the same RAW just re-processed with Photo AI: https://www.molon.de/images/topaz_vs_dxo/kingfisher_photo_ai.png

    2. A red leaf monkey I shot last month in the Danum valley (ISO 6400,
    Olympus 100-400 at 400mm).
    Left is the out-of-camera JPEG, right the output of Photo AI: https://www.molon.de/images/topaz_vs_dxo/red_leaf_monkey_photo_ai.png

    Quite impressive detail recovery, especially in the monkey picture.

    The 75-300 is a relatively inexpensive tele lens, and ISO 6400 is not
    exactly an ISO level you should use with a micro 4/3 camera. But with
    modern software, no problem, you get a sharp and noiseless result.
    --
    Alfred Molon

    Olympus 4/3 and micro 4/3 cameras forum at
    https://groups.io/g/myolympus
    https://myolympus.org/ photo sharing site

    Yes, they've really improved on this. But I would like to see shots taken in a controlled environment at 200-400 ISO then again at 6400 ISO to see how much of the reclaimed detail matches low ISO detail.

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Alfred Molon@21:1/5 to All on Sat Sep 24 11:11:52 2022
    Am 24.09.2022 um 08:08 schrieb RichA:
    Yes, they've really improved on this. But I would like to see shots taken in a controlled environment at 200-400 ISO then again at 6400 ISO to see how much of the reclaimed detail matches low ISO detail.

    Quite possible that they apply models to regenerate the detail. For
    instance, they detect a monkey and apply the fur model to regenerate the
    fur on the hands.
    Or the AI detects a bird and applies a "feathers model" to regenerate
    the detail.

    Still, the regenerated detail in these images looks quite realistic.
    --
    Alfred Molon

    Olympus 4/3 and micro 4/3 cameras forum at
    https://groups.io/g/myolympus
    https://myolympus.org/ photo sharing site

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Incubus@21:1/5 to Alfred Molon on Mon Sep 26 14:40:22 2022
    On 2022-09-22, Alfred Molon <[email protected]> wrote:
    Some tests with Photo AI (images shot with the Olympus E-M1 Mark II and
    III cameras):

    1. The kingfisher I shot in 2019 with the Olympus 75-300 lens at 300mm /
    ISO 6400.
    Left is my RAW processing result of 2019, right the same RAW just re-processed with Photo AI: https://www.molon.de/images/topaz_vs_dxo/kingfisher_photo_ai.png

    2. A red leaf monkey I shot last month in the Danum valley (ISO 6400,
    Olympus 100-400 at 400mm).
    Left is the out-of-camera JPEG, right the output of Photo AI: https://www.molon.de/images/topaz_vs_dxo/red_leaf_monkey_photo_ai.png

    Quite impressive detail recovery, especially in the monkey picture.

    It doesn't recover detail. It inserts fake detail based on what it
    thinks will look okay to a human. It might be impressive but it's not
    really photography in my book.

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From RichA@21:1/5 to Incubus on Mon Sep 26 17:15:31 2022
    On Monday, 26 September 2022 at 10:40:29 UTC-4, Incubus wrote:
    On 2022-09-22, Alfred Molon <[email protected]> wrote:
    Some tests with Photo AI (images shot with the Olympus E-M1 Mark II and
    III cameras):

    1. The kingfisher I shot in 2019 with the Olympus 75-300 lens at 300mm / ISO 6400.
    Left is my RAW processing result of 2019, right the same RAW just re-processed with Photo AI: https://www.molon.de/images/topaz_vs_dxo/kingfisher_photo_ai.png

    2. A red leaf monkey I shot last month in the Danum valley (ISO 6400, Olympus 100-400 at 400mm).
    Left is the out-of-camera JPEG, right the output of Photo AI: https://www.molon.de/images/topaz_vs_dxo/red_leaf_monkey_photo_ai.png

    Quite impressive detail recovery, especially in the monkey picture.
    It doesn't recover detail. It inserts fake detail based on what it
    thinks will look okay to a human. It might be impressive but it's not
    really photography in my book.

    But at what point is basic "sharpening" in programs like PS "cross the line?"

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Incubus@21:1/5 to Alfred Molon on Mon Oct 3 14:10:06 2022
    On 2022-10-03, Alfred Molon <[email protected]> wrote:
    Am 26.09.2022 um 16:40 schrieb Incubus:
    On 2022-09-22, Alfred Molon <[email protected]> wrote:
    Some tests with Photo AI (images shot with the Olympus E-M1 Mark II and
    III cameras):

    1. The kingfisher I shot in 2019 with the Olympus 75-300 lens at 300mm / >>> ISO 6400.
    Left is my RAW processing result of 2019, right the same RAW just
    re-processed with Photo AI:
    https://www.molon.de/images/topaz_vs_dxo/kingfisher_photo_ai.png

    2. A red leaf monkey I shot last month in the Danum valley (ISO 6400,
    Olympus 100-400 at 400mm).
    Left is the out-of-camera JPEG, right the output of Photo AI:
    https://www.molon.de/images/topaz_vs_dxo/red_leaf_monkey_photo_ai.png

    Quite impressive detail recovery, especially in the monkey picture.

    It doesn't recover detail. It inserts fake detail based on what it
    thinks will look okay to a human. It might be impressive but it's not
    really photography in my book.

    It's not adding any fake detail. If you look closely at the monkey
    picture for instance, you will see that Sharpen AI is not adding a
    single hair which was not in the original image.

    Sharpen AI seems to have some very sophiticated deblurring algorithm,
    meaning that it takes a blurred image and removes the blur.

    You cannot deblur a blurred image. You cannot recover detail that isn't
    there. The application is fooling you. As I said, it might be
    impressive, but the results won't always impress.

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Alfred Molon@21:1/5 to All on Mon Oct 3 16:03:09 2022
    Am 26.09.2022 um 16:40 schrieb Incubus:
    On 2022-09-22, Alfred Molon <[email protected]> wrote:
    Some tests with Photo AI (images shot with the Olympus E-M1 Mark II and
    III cameras):

    1. The kingfisher I shot in 2019 with the Olympus 75-300 lens at 300mm /
    ISO 6400.
    Left is my RAW processing result of 2019, right the same RAW just
    re-processed with Photo AI:
    https://www.molon.de/images/topaz_vs_dxo/kingfisher_photo_ai.png

    2. A red leaf monkey I shot last month in the Danum valley (ISO 6400,
    Olympus 100-400 at 400mm).
    Left is the out-of-camera JPEG, right the output of Photo AI:
    https://www.molon.de/images/topaz_vs_dxo/red_leaf_monkey_photo_ai.png

    Quite impressive detail recovery, especially in the monkey picture.

    It doesn't recover detail. It inserts fake detail based on what it
    thinks will look okay to a human. It might be impressive but it's not
    really photography in my book.

    It's not adding any fake detail. If you look closely at the monkey
    picture for instance, you will see that Sharpen AI is not adding a
    single hair which was not in the original image.

    Sharpen AI seems to have some very sophiticated deblurring algorithm,
    meaning that it takes a blurred image and removes the blur.

    Years ago, some poster hear was using the Richardson–Lucy deconvolution
    to unblur images: https://en.wikipedia.org/wiki/Richardson%E2%80%93Lucy_deconvolution

    Perhaps Topaz have done some research on the matter and found an even
    better was to deblur images.
    --
    Alfred Molon

    Olympus 4/3 and micro 4/3 cameras forum at
    https://groups.io/g/myolympus
    https://myolympus.org/ photo sharing site

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From nospam@21:1/5 to [email protected] on Mon Oct 3 11:00:54 2022
    In article <[email protected]omain>, Incubus <[email protected]> wrote:

    You cannot deblur a blurred image.

    actually, you can.

    more info in the part you snipped: <https://en.wikipedia.org/wiki/Richardson�Lucy_deconvolution>

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Alfred Molon@21:1/5 to All on Mon Oct 3 18:48:56 2022
    Am 03.10.2022 um 16:10 schrieb Incubus:
    On 2022-10-03, Alfred Molon <[email protected]> wrote:
    Am 26.09.2022 um 16:40 schrieb Incubus:
    On 2022-09-22, Alfred Molon <[email protected]> wrote:
    Some tests with Photo AI (images shot with the Olympus E-M1 Mark II and >>>> III cameras):

    1. The kingfisher I shot in 2019 with the Olympus 75-300 lens at 300mm / >>>> ISO 6400.
    Left is my RAW processing result of 2019, right the same RAW just
    re-processed with Photo AI:
    https://www.molon.de/images/topaz_vs_dxo/kingfisher_photo_ai.png

    2. A red leaf monkey I shot last month in the Danum valley (ISO 6400,
    Olympus 100-400 at 400mm).
    Left is the out-of-camera JPEG, right the output of Photo AI:
    https://www.molon.de/images/topaz_vs_dxo/red_leaf_monkey_photo_ai.png

    Quite impressive detail recovery, especially in the monkey picture.

    It doesn't recover detail. It inserts fake detail based on what it
    thinks will look okay to a human. It might be impressive but it's not
    really photography in my book.

    It's not adding any fake detail. If you look closely at the monkey
    picture for instance, you will see that Sharpen AI is not adding a
    single hair which was not in the original image.

    Sharpen AI seems to have some very sophiticated deblurring algorithm,
    meaning that it takes a blurred image and removes the blur.

    You cannot deblur a blurred image.

    Well, it seems you can - compare the images I posted - no fake detail
    has been added.
    And read the article about the Richardson–Lucy deconvolution I posted.
    --
    Alfred Molon

    Olympus 4/3 and micro 4/3 cameras forum at
    https://groups.io/g/myolympus
    https://myolympus.org/ photo sharing site

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Alfred Molon@21:1/5 to All on Mon Oct 3 19:57:53 2022
    Am 03.10.2022 um 18:48 schrieb Alfred Molon:
    Am 03.10.2022 um 16:10 schrieb Incubus:
    On 2022-10-03, Alfred Molon <[email protected]> wrote:
    Am 26.09.2022 um 16:40 schrieb Incubus:
    On 2022-09-22, Alfred Molon <[email protected]> wrote:
    Some tests with Photo AI (images shot with the Olympus E-M1 Mark II
    and
    III cameras):

    1. The kingfisher I shot in 2019 with the Olympus 75-300 lens at
    300mm /
    ISO 6400.
    Left is my RAW processing result of 2019, right the same RAW just
    re-processed with Photo AI:
    https://www.molon.de/images/topaz_vs_dxo/kingfisher_photo_ai.png

    2. A red leaf monkey I shot last month in the Danum valley (ISO 6400, >>>>> Olympus 100-400 at 400mm).
    Left is the out-of-camera JPEG, right the output of Photo AI:
    https://www.molon.de/images/topaz_vs_dxo/red_leaf_monkey_photo_ai.png >>>>>
    Quite impressive detail recovery, especially in the monkey picture.

    It doesn't recover detail.  It inserts fake detail based on what it
    thinks will look okay to a human.  It might be impressive but it's not >>>> really photography in my book.

    It's not adding any fake detail. If you look closely at the monkey
    picture for instance, you will see that Sharpen AI is not adding a
    single hair which was not in the original image.

    Sharpen AI seems to have some very sophiticated deblurring algorithm,
    meaning that it takes a blurred image and removes the blur.

    You cannot deblur a blurred image.

    Well, it seems you can - compare the images I posted - no fake detail
    has been added.
    And read the article about the Richardson–Lucy deconvolution I posted.

    One more article on the matter:
    https://en.wikipedia.org/wiki/Deconvolution

    Combine deconvolution with knowledge how the subject should look (fur,
    feather, tree bark etc.) and you should be to achieve do something.

    My guess is that these algorithms will get more and more sophisticated
    over time.
    --
    Alfred Molon

    Olympus 4/3 and micro 4/3 cameras forum at
    https://groups.io/g/myolympus
    https://myolympus.org/ photo sharing site

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Whisky-dave@21:1/5 to Alfred Molon on Tue Oct 4 05:34:14 2022
    On Monday, 3 October 2022 at 18:58:01 UTC+1, Alfred Molon wrote:
    Am 03.10.2022 um 18:48 schrieb Alfred Molon:


    Combine deconvolution with knowledge how the subject should look (fur, feather, tree bark etc.) and you should be to achieve do something.

    My guess is that these algorithms will get more and more sophisticated
    over time.

    For me that's when the worrying starts is it a photo or a digital representation of what the image should look like.

    An example would be a photo of Yul Brynner . I believe he had hair in the 1940s.
    But if take a picture of him (or rather a picture of a photo in the mid 60s) , would the AI put hair on his head ?
    .

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Alfred Molon@21:1/5 to All on Tue Oct 4 18:54:16 2022
    Am 04.10.2022 um 14:34 schrieb Whisky-dave:
    On Monday, 3 October 2022 at 18:58:01 UTC+1, Alfred Molon wrote:
    Am 03.10.2022 um 18:48 schrieb Alfred Molon:


    Combine deconvolution with knowledge how the subject should look (fur,
    feather, tree bark etc.) and you should be to achieve do something.

    My guess is that these algorithms will get more and more sophisticated
    over time.

    For me that's when the worrying starts is it a photo or a digital representation of what the image should look like.

    An example would be a photo of Yul Brynner . I believe he had hair in the 1940s.
    But if take a picture of him (or rather a picture of a photo in the mid 60s) , would the AI put hair on his head ?

    Topaz Photo AI won't put hair on a bald head.
    --
    Alfred Molon

    Olympus 4/3 and micro 4/3 cameras forum at
    https://groups.io/g/myolympus
    https://myolympus.org/ photo sharing site

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From David Brooks@21:1/5 to Alfred Molon on Tue Oct 4 18:42:06 2022
    On 04/10/2022 17:54, Alfred Molon wrote:
    Am 04.10.2022 um 14:34 schrieb Whisky-dave:
    On Monday, 3 October 2022 at 18:58:01 UTC+1, Alfred Molon wrote:
    Am 03.10.2022 um 18:48 schrieb Alfred Molon:


    Combine deconvolution with knowledge how the subject should look (fur,
    feather, tree bark etc.) and you should be to achieve do something.

    My guess is that these algorithms will get more and more sophisticated
    over time.

    For me that's when the worrying starts is it a photo or a digital
    representation of what the image should look like.

    An example would be a photo of Yul Brynner . I believe he had hair in
    the 1940s.
    But if  take a picture of him (or rather a picture of a photo in the
    mid 60s) , would the AI put hair on his head ?

    Topaz Photo AI won't put hair on a bald head.

    Haha!

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Whisky-dave@21:1/5 to Alfred Molon on Thu Oct 6 05:55:18 2022
    On Tuesday, 4 October 2022 at 17:54:23 UTC+1, Alfred Molon wrote:
    Am 04.10.2022 um 14:34 schrieb Whisky-dave:
    On Monday, 3 October 2022 at 18:58:01 UTC+1, Alfred Molon wrote:
    Am 03.10.2022 um 18:48 schrieb Alfred Molon:


    Combine deconvolution with knowledge how the subject should look (fur,
    feather, tree bark etc.) and you should be to achieve do something.

    My guess is that these algorithms will get more and more sophisticated
    over time.

    For me that's when the worrying starts is it a photo or a digital representation of what the image should look like.

    An example would be a photo of Yul Brynner . I believe he had hair in the 1940s.
    But if take a picture of him (or rather a picture of a photo in the mid 60s) , would the AI put hair on his head ?
    Topaz Photo AI won't put hair on a bald head.

    What about a bald patch.

    How about unbrusing a fruit.
    That was my first cloning session when learning photoshop 3.0.

    I'm sure AI could do that.
    But when should it take over .....
    We teach AI of and machine learning, it;s an interesting concept, and why the term artifical is used.




    --
    Alfred Molon

    Olympus 4/3 and micro 4/3 cameras forum at
    https://groups.io/g/myolympus
    https://myolympus.org/ photo sharing site

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)