Focus Mask Tool

[fotagf8]

I have several questions about the focus mask tool.  I am using a Leaf back.  Does it only work on Leaf and Phase One raw files?  When I import a tif file, I don't receive any focus information?

Second, I am a bit puzzled by what it selects.  I would assume everything on the same plane would be in focus, but if I shoot a skyscraper that is parallel to the sensor plane, some of the windows are shown as in focus while others are not.  I would have assumed the entire face of the building would be green.  Third, I am puzzled why water close to me and a building a mile away are both perceived as being in focus.  Even more odd, it was a windy day and I was using a relatively slow shutter speed, so the water is even more puzzling to me.   

Thanks

[Keith Reeder]

Assuming you're using Capture One Pro (and not "DB") it should work with any Raw file you open.

It's a pointless, non-functioning gimmick, IMHO - I've yet to see it agree with what I know to be the sharp/in focus parts of an image: it will frequently try and tell me that the entire (very OOF) background is "in focus" (red = in focus according to Capture One here); and mucking around with the threshold values provides no benefit whatsoever.

It simply does not work, in my experience - and in yours too, by the sound of it. Maybe it behaves better for controlled, predictable studio headshots...

[byarvin]

I find that the focus mask tool works amazingly well with both Olympus and Canon raw files and saves hours of rough editing. It was a major reason for me to try CaptureOne and the main reason I recommend it over other advanced raw processors.

That being said, CaptureOne seems to be filled with buttons that don't work for somebody or other. For example, when I click on "hide focus mask" nothing will happen most of the time. And one of the trial versions I had wouldn't do any sort of manual color corrections - the entire menu was missing.

People often encourage me to push my work to the edge and beyond and that's exactly what the CaptureOne engineers are doing. Strange things happen when you're out there, sometimes you just have to bear with them. For me, the results are worth it.

[fotagf8]

Well I don't know if I would call it a gimmick.  I had to calibrate my helical focus ring.  I found the focus mask helpful there.  In camera, I marked each photo with an additional star in the series, which matched the number of additional markings on the helical that I turned it by.  The portion of the photo that I was using to measure sharpness did get increasingly more green, which meant that I need to add a factor of +3 in using the distance setting.  So far that has proven to be correct.

I think you answered part of my question.  If the tool works on any RAW file, that may explain why it doesn't seem to work on a tiff.  I do agree that parts of Capture One are a bit kludgy, but that may be that I come to it after having devoted significant time and effort to learning and using Photoshop.

[MichaelEzra]

Check out the Focus Mask in RawTherapee:)
See bottom of the page http://50.87.144.65/~rt/w/index.php?title=Preview_Modes
An improvement in the recent version (not illustrated in this article) was made to indicate the degree of focusing - the focus mask is more opaque in better focused / sharper image areas.
it works on any image that can be opened within RawTherapee.

[Rawcoll]

Assuming you're using Capture One Pro (and not "DB") it should work with any Raw file you open.

It should but it doesn't. Like auto-fill in the mask tool. It won't work properly with Fuji raws, or Panasonic RW2 files if I remember correctly.

I agree with byarvin, "CaptureOne seems to be filled with buttons that don't work for somebody or other." That's been my experience.

Ian

[Bart_van_der_Wolf]

Check out the Focus Mask in RawTherapee:)

+1

In general, the focus-mask or focus-peeking implementation in many programs and firmware is easily fooled by high contrast edges.

The better implementations attempt to separate real high spatial frequency detail by using a sort of high-pass filtering, e.g. by comparing the original with a slightly blurred version and using thresholds, but one has to be clever with optimal coding efficiency, because high-pass filtering can take a bit of time (but can also be done very efficiently, using bit-shifts instead of multiplication or division).

The particular focus-mask implementation in RawTherapee is much better than most others at nailing what is real detail and what is just high contrast, although it sometimes remains hard to reliably label a high contrast edge as unsharp. I also like the fact that it analyses the image preview at the current zoom level, because that allows to quickly switch between different levels of detail with the scroll/zoom button of a mousepointer. Zooming in to 200% gives a quite accurate focus indication, and by zooming out the influence of noise is usually reduced. It works also nicely in concert with the sharpening functions, because it recalculates the focus mask pretty fast.

Cheers,
Bart

[Keith Reeder]

In general, the focus-mask or focus-peeking implementation in many programs and firmware is easily fooled by high contrast edges.

But Bart, in the example I post up the page, the entire OOF background is lit up like a battlefield, and there's not a high contrast edge to be seen.

[Keith Reeder]

I find that the focus mask tool works amazingly well with both Olympus and Canon raw files.

I'd be interested to know what kind of subject matter you shoot: I'm a Canon user, and the Focus Mask has long proven to be truly useless for my bird and sport photography - it communicates no useful information whatsoever about the image.

[Bart_van_der_Wolf]

But Bart, in the example I post up the page, the entire OOF background is lit up like a battlefield, and there's not a high contrast edge to be seen.

Hi Keith,

I don't know why Capture One decided that, and I don't know which sensitivity settings you used, but there is a difference in the use of the tool. CO attempts to analyze the image at full resolution (or as it looks at a somewhat smaller size), in order to allow a quick preview of 'focused' versus 'defocused' images in thumbnails and other small versions (e.g. zoomed out in multiple image previews). It's more a tool for comparing between shots in a session.

That doesn't mean that I think it should not be more accurate, it should and it could (as RawTherapee proves).

Cheers,
Bart

[byarvin]

Most of my work is macro - especially food. I love the way the focus mask shows the band of sharp focus and how it fades away. I find it very useful.

[fotagf8]

Thanks all.

I am not familiar with Focus Mask in RawTherapee.  Can this be implemented as a Photoshop plugin?

[Bart_van_der_Wolf]

I am not familiar with Focus Mask in RawTherapee.  Can this be implemented as a Photoshop plugin?

Hi,

No, RawTherapee is a standalone Raw converter. However, you can assign Photoshop as external editor of the Raw conversion result. It can also read TIFF and JPEG files and judge focus on those.

Cheers,
Bart

[MichaelEzra]

I don't really see a practical use case of focus mask in photoshop, it would be best suited for lightroom or bridge. I am actually planing to look into that.

[Jack Varney]

Keith, regarding the "battlefield" of in sharp indications, have you gone to the Preferences settings to increase the sharpness "Threshold"? Edit>Preferences>select the Focus Mask Icon and push the Threshold slider up to suit. Some testing to gain a feel for the effect is in order. I have found the focus mask quite useful.

[MichaelEzra]

I'd be curious to see how RawTherapee shows focus mask over this problem image. When I designed the detection algorithm one of the goals was to make it adjustment free, something that simply works in majority of the cases. Of course tuning UI controls can be provisioned and algorithm allows for that as well, but it does not seem necessary.

[Bart_van_der_Wolf]

I'd be curious to see how RawTherapee shows focus mask over this problem image.

Hi Michael,

Me too!  Your algorithm seems to be quite robust, although maybe a bit insensitive in dark areas (which may be a good thing for most images).

When I designed the detection algorithm one of the goals was to make it adjustment free, something that simply works in majority of the cases. Of course tuning UI controls can be provisioned and algorithm allows for that as well, but it does not seem necessary.

I agree that the thresholds are chosen well on common image content (and makes it easy to use). Of course the control freak in me would not mind even more control over the detection process ...  In the attached example, the region of best focus is accurately selected, even the surface scratch (battle scar ) on my resolution test target gets picked up. However, the sharp edge is missed for some reason, maybe the slant is enough to elude the sample radius from allowing to detect it, or it is not 'noisy' enough. The crop is unsharpened, after sharpening the scratch gets highlighted and the detection area shrinks in the center of the 'star', as it should, but the slanted edges keep escaping.

Cheers,
Bart

[Keith Reeder]

Keith, regarding the "battlefield" of in sharp indications, have you gone to the Preferences settings to increase the sharpness "Threshold"? Edit>Preferences>select the Focus Mask Icon and push the Threshold slider up to suit

Hello Jack,

yeah, as I mention up the page, "and mucking around with the threshold values provides no benefit whatsoever".

It's not an issue for me - I maintain that I can tell better than any algorithm (including Raw Therapee's, which I find equally unhelpful in terms of telling me anything I didn't already know) what is sharp in an image - which is why I consider the whole idea of Focus Mask functionality to be a gimmick - a solution looking for problem.

[Keith Reeder]

In the attached example, the region of best focus is accurately selected, even the surface scratch (battle scar ) on my resolution test target gets picked up.

And is it providing you with any useful - much less new - information in that case, Bart?

That's my stumbling-block with this functionality: I don't need to be told by software what my eyes have already told me (especially given the likelihood of some pretty profound disagreement between the two on that score! ) so what's the point?

[Bart_van_der_Wolf]

And is it providing you with any useful - much less new - information in that case, Bart?

Actually yes it does. That crop is from a series of focus positions to find the best possible focus from a series of slightly different focus distances (on a focus rail), as a prelude to an aperture series test. I used a laborious analytical procedure to derive the radial blur amounts  (see attached charted measurements) required for deconvolution capture sharpening. I also learned that the blur radius becomes elliptic when out of focus, which I didn't notice before, and suggests decentering or a non-circular aperture.

Question is, can a focus mask pick up such small differences, and thus help in finding the best focused image in a series? Well, it does show that the best focus produces the smallest circular region of focus, something that requires laborious measuring of the blur radius to achieve otherwise. And actually, an expansion of the Focus mask could perhaps provide a Figure of Merit (FOM) of the degree of focus it found ...

The Focus-mask is also helpful in quickly picking out the in-focus shots of a series of portraits (see attached crop).

It is also useful in showing the focus plane in a tilted shot with a Tilt and Shift lens.

It is also useful in showing field curvature, or (in case of Capture One) when using tethered shooting, or focus stacking.

That's my stumbling-block with this functionality: I don't need to be told by software what my eyes have already told me (especially given the likelihood of some pretty profound disagreement between the two on that score! ) so what's the point?

The point is speed. I is much easier to spot the plane of best focus when it is highlighted, especially when shooting tethered, or to spot a better focused image in a series when paging through a series. The other day I shot a scene where I needed the focus to be optimal at 50 metres distance with a TS-E 45mm lens with only a fraction of focus throw at that distance, which was easy enough with tethered shooting to my table with an App with Live View tethering. The focus mask was more critical than my eye could spot. Of course the indication must be accurate to be useful, and Capture One apparently needs to do some homework.

Cheers,
Bart