Post processing getting easier (?)
Moderators: Chris S., Pau, Beatsy, rjlittlefield, ChrisR
Re: Post processing getting easier (?)
Sure, I'd love to find easier ways! I'll send you something by PM later today or tomorrow. Remind me if I forget (it has been known to happen).
Re: Post processing getting easier (?)
just played around in Topaz Photo AI
Re: Post processing getting easier (?)
Not a good advertisement for Topaz imho, or AI in general; it made up 'ghost hairs' all over the place which are surely not
there in the orig.
I did try out Topaz SharpenAI v2.x and was enthousiast, did a very good job to all I fed it. Then v3 I found disastrous. No clue about current but this does not look good to me.
When you're the champion of the room, odds are that you're in the wrong room
- Egon

- Egon
Re: Post processing getting easier (?)
It would be helpful to see a normally-sharpened version of the original.
Re: Post processing getting easier (?)
curves, graduation, clarity and unsharp mask sharpening in Affinity
Last edited by lothman on Sat Jun 07, 2025 6:31 am, edited 1 time in total.
Re: Post processing getting easier (?)
@ Lou,
first guestion imho: does it need sharpening at all? (I think not but we do not see the original here online).
For games sake I tried unsharpen in my old PSP9 on the 3rd (end) version... at reach .5, strength 80 (!) and distinguish at lowest (2); as expected I got a bit raised microcontrast but no made up artifacts.
@ lothman
Both at 200%:
1) The version that Beatsy ended up with and liked
2) The Topaz version
We can see a lot of differences:
a) Topaz added about 15 points yellow (did you ask for that?)
b) Topaz lowered contrast significantly
c) Topaz lost a lot of original detail
d) Topaz added ghostlines, not just dark ones but also light ones (I marked a few in red)
Regards, Egon
Edit: have to add the Affinity version is even more horrible imho
When you're the champion of the room, odds are that you're in the wrong room
- Egon

- Egon
Re: Post processing getting easier (?)
3) My version of unsharpened (using PSP9 unsharpen .5 / 150 / 2 - the 150 is already overdone imho, I don't think Beatsy's pic needs any 'improvements'). This is sharpened for internet. If for print the .5 would change to 1.5 or such.... I set the discrimination at lowest, so all and everything gets sharpened.:
When you're the champion of the room, odds are that you're in the wrong room
- Egon

- Egon
Re: Post processing getting easier (?)
Thank you veruy much, Lothman, for posting the non-AI sharpened version. That is very helpful.
Emkraak, most of us have found that deep stacks with high-NA objectives, where diffraction is important, need considerable sharpening. Convolution-based methods hardly introduce any artifacts, unlike Unsharp Mask. I also like to set the Photoshop Masking slider to a middle value to avoid amplifying noise. I am very happy with the "advanced" version of the Photoshop sharpening controls used in this way.
Emkraak, most of us have found that deep stacks with high-NA objectives, where diffraction is important, need considerable sharpening. Convolution-based methods hardly introduce any artifacts, unlike Unsharp Mask. I also like to set the Photoshop Masking slider to a middle value to avoid amplifying noise. I am very happy with the "advanced" version of the Photoshop sharpening controls used in this way.
Re: Post processing getting easier (?)
All very possible, Lou. But my impression is that Beatsy presented what he was already happy with... I don't see any objections or flaws to that. Subsequently I see several sharpenings proposed that make a mess of things imho. - What did I misunderstand?Lou Jost wrote: ↑Sat Jun 07, 2025 11:03 amThank you veruy much, Lothman, for posting the non-AI sharpened version. That is very helpful.
Emkraak, most of us have found that deep stacks with high-NA objectives, where diffraction is important, need considerable sharpening. Convolution-based methods hardly introduce any artifacts, unlike Unsharp Mask. I also like to set the Photoshop Masking slider to a middle value to avoid amplifying noise. I am very happy with the "advanced" version of the Photoshop sharpening controls used in this way.
(btw, I do not use Photoshop or any Adobe software, like I do not use fearbook and the rest of asocial nonsense- point of principle, AND I like the internet the way it was meant to be, like this forum)
When you're the champion of the room, odds are that you're in the wrong room
- Egon

- Egon
Re: Post processing getting easier (?)
It was interesting to see the AI results, I think. But you will need to ask them your question.emkraak wrote: ↑Sat Jun 07, 2025 11:42 amAll very possible, Lou. But my impression is that Beatsy presented what he was already happy with... I don't see any objections or flaws to that. Subsequently I see several sharpenings proposed that make a mess of things imho. - What did I misunderstand?Lou Jost wrote: ↑Sat Jun 07, 2025 11:03 amThank you veruy much, Lothman, for posting the non-AI sharpened version. That is very helpful.
Emkraak, most of us have found that deep stacks with high-NA objectives, where diffraction is important, need considerable sharpening. Convolution-based methods hardly introduce any artifacts, unlike Unsharp Mask. I also like to set the Photoshop Masking slider to a middle value to avoid amplifying noise. I am very happy with the "advanced" version of the Photoshop sharpening controls used in this way.
(btw, I do not use Photoshop or any Adobe software, like I do not use fearbook and the rest of asocial nonsense- point of principle, AND I like the internet the way it was meant to be, like this forum)