Bart,
Do you have any suggestions for starting values of the three sliders and formsharpening, focus and stabilize. I know I am asking for a large amount of advice but cannot seem to find any article with suggested values except what Topaz has as default.
Hi Jed,
There is no best default.
The optimal values differ mostly with one's input image quality. Personally, for certain types of work like architecture, or product photography, I'm mostly shooting from tripod and use magnified Live View for focusing. In such cases, one can assume that "Sharpen" is the best mode to use because there won't be 'defocus' involved, and the chance of subject motion is minimal, and camera shake should be zero.
When shooting action, subject motion and or "Defocus" is likely, even at high shutter speed. If some subject motion adds to the sensation of a dynamic situation, then "Stabilize" would
not be the best starting point, but "Focus" could be. But if the AF nailed the focus, and there is still some residual overall motion blur or camera shake, the "Stabilize" would be worthwhile as a first attempt.
The best amount setting to choose depends on the amount of blur in the original image that we're faced with, and which we need to match with the right model strength.
In my experience, it helps to learn to judge what is required by starting at 0 (which already does a bit of sharpening). Then try 10, 20, 30, etc. Zooming in to 400% helps to see minute changes, and because a smaller Region of Interest is chosen, the calculations will be a bit quicker. Usually, at 20 I already see a significant difference, and if I do, then 30 or 40 might already be too much (not much sharper but more artifacts can start to develop). In other cases, it may take 60 or 70 to see a noticeable impact. You'll gain experience as you try that more often.
By now I can often see (on my own images) approximately where the better starting point lies, without having to slowly hone in on the IMHO optimal value. But when faced with someone else's images, I may need to start searching again, because the different lenses and shooting technique pose a different challenge to solve.
Another thing to keep in mind is that better AI models are still being trained, so the software may behave a bit different with certain images after a new update is released. More functionality is also being added, so when we might have learned to be a bit more conservative in the amounts chosen, in order to avoid artifacts, we may need to go further, or less far, before that happens on a new version. One potential area of needed improvement is with very small text in images. I assume that only a relatively few images with text have been used to train the AI models. That may be a good thing to avoid random structures being replaced by text where there is none...
So again, values of 30 to 60 are more commonly required on my images, but on other images it may be different. However, it is also important to keep in mind that we are applying Capture Sharpening here, and we only should restore Capture process losses back to fine focus. Sharpening for output is another next step, with other tools (like e.g. Topaz Precision Detail).
Cheers,
Bart
