Topaz Labs InFocus, a new deconvolution based sharpening plugin

[ErikKaffehr]

Hi,

I have played around with Smart Sharpen, Focus Magick, Raw Developer and Topaz In Focus. Topaz in Focus can do more aggressive sharpening than the others. The Unknown/Estime function is nice but doesn't work on symmetric effects. So if I approximate a PSF for the left top corner it would make havoc of the bottom left corner.

My experience is that it works, but it's easy to overdose. Like putting a Porsche engine in  Volkswagen, it will go fast but it will be hard to keep it on the road.

Best regards
Erik

Bill Janes on 21. november:
> My main concern is how Topaz compares to other available deconvolution algorithms.

Has anybody compared it to the R-L deconvolution of Raw Developer?

 

[Bart_van_der_Wolf]

Bill Janes on 21. november:
> My main concern is how Topaz compares to other available deconvolution algorithms.

Has anybody compared it to the R-L deconvolution of Raw Developer?

Hi Hening,

Topaz Labs' InFocus does create artifacts easily when pushed beyond the correct settings. It also seems to have a harder time getting an optimal result, at least when comppared to e.g. FocusMagic (which unfortunately doesn't run on all platforms, and hasn't seen any real development for years). A lot of that may have to do with the underlying algorithm(s) of InFocus.

Topaz Labs have chosen to utilize a more modern algorithm than the well tested (but not necessarily optimal) Richardson Lucy algorithm (e.g. used to salvage the early Hubble Space Station imagery). For well behaved, but slightly blurred, images it works fine. However, for the more pathological (reject) cases it (too) easily fails with generating ringing artifacts when pushed too far, compared to its alternatives (which means restoration is possible, but not in all cases with InFocus, yet).

Cheers,
Bart

[deejjjaaaa]

I have played around with Smart Sharpen, Focus Magick, Raw Developer and Topaz In Focus.

you can add Adobe Camera Raw to the mix (can be used on .tiffs in postprocessing too) w/ the "Detail" slider set @ 100 (= pure deconvolution of some form, w/o any mix of USM, according to Eric Chan)

[Lightsmith]

I bought and tried using FocusMagic but it did a very poor job of adjusting OOF images. Part of the problem with OOF or unsharp images is a lack of boundary contrast. I always make sure to up contrast before sharpening as doing this cuts the amount of sharpening in half and does a much better job of preserving image detail.

What is different with InFocus, used properly, is that it allows you to adjust what it calls the "micro contrast" which builds detail in the structure of the area by darkening midtones. It infills independently of any standard sharpening and so it does things that no sharpening software can do. At least it does what cannot be done with the latest versions of Noise Ninja, Dfine 2.0, Neat Image, DeNoise, and CS5's NR, all of which I tested with images two weeks ago as a separate NR evaluation.

[kim]

Interesting to read the previous responses. Like many of them I have Focus Magic which I have found over 5+ years to be invaluable for sharpening some problem images. The lack of ongoing development by the owners of Focus Magic is disappointing and I’ve kept 32bit PS CS3 on my machine so I can continue to use it.

Maybe I was doing something wrong with Topaz InFocus but I found it to be pretty useless and no better than USM. On my images it increased artifacts without improving sharpness. I watched one of the (impressive) videos showing how to use InFocus but despite doing what the instructions said, it didn’t deliver the results. In contrast, Focus Magic did a great job rescuing those same images. I should add that I already have the Topaz DeNoise and Detail plugins, both of which do a superb job.

It seems that InFocus is a product that divides opinion so it’s more true than ever that someone who is interested in it needs to try it for themselves to see if it suits their images and workflow.

[Gordon Buck]

The lack of ongoing development by the owners of Focus Magic is disappointing and I’ve kept 32bit PS CS3 on my machine so I can continue to use it.

Focus Magic works on my 32bit CS5.

[Tim Gray]

Focus Magic left my workflow with the introduction of Topaz Detail.  Interesting that no one has commented on Detail vs In Focus.  So far I haven't seen situations where In Focus achieved results that were observably better than Detail.

Any other experiences?

[Craig Lamson]

Focus Magic left my workflow with the introduction of Topaz Detail.  Interesting that no one has commented on Detail vs In Focus.  So far I haven't seen situations where In Focus achieved results that were observably better than Detail.

Any other experiences?

I love Topaz detail and I too quit using Focus Magic once I got the hang of Detail.

[felix5616]

I downloaded the trial version but never received the authorization code and could not contact their tech support, very disappointed.

[Bart_van_der_Wolf]

I downloaded the trial version but never received the authorization code and could not contact their tech support, very disappointed.

From their Website:
"If you don't receive the trial keys after ten minutes, please check your spam folder. If you still can't find the email, let us know by sending a message to keys@topazlabs.com. Include which Topaz product you want and whether you use Mac or Windows"

Cheers,
Bart

[Rajan Parrikar]

I downloaded the trial version but never received the authorization code and could not contact their tech support, very disappointed.

The same thing happened to me over the past week.  I sent requests for a trial code for ReMask but didn't hear back even after 3 or 4 attempts.  Then I called Topaz and they promptly emailed me a code.  My guess is that their auto code generator is not working.  I don't know if they have fixed it now.

[kim]

I sent requests for a trial code for ReMask but didn't hear back even after 3 or 4 attempts. 

I had a similar problem and believe it was due to having used that same email address in the past for requesting a trial code. Using a "fresh" email address solved it.