[Pentaxtic]

In camera software or desktop app? I know it wouldn't be the same FOV but you would adjust to what you get after the sensor crop...but get the desired shallow depth of field via software simulation app?
Just a question since there is the AA removal simulation in K3.

[elliott]

The DOF is just a function of distance, focal length and aperture, you can get the same DOF on APS-C as you can on any format at the same distance,focal length and aperture. The field of view changes with format, the DOF is going to be the same if you keep everything else the same.

There are fake shallow DOF effects you can do in post, they are commonly used with cell phone cameras and other small sensor compacts because it is otherwise almost impossible to achieve the effect. It can be easy to pick out the poorly done fakes because there is usually something sharp that shouldn't be or vice versa. Also, you can't always easily mimic the unique bokeh that some glass is known for.

Just a clarification, the anti-aliasing of the K-3 is not a software trick, it is just an alternative method of anti-aliasing. Instead of using a piece of glass to apply a very slight blur to the image, it shakes the sensor.

[Lowell Goudge]

The DOF is just a function of distance, focal length and aperture, ......

Actually it also includes magnification from the sensor to the print / viewing size and viewing distance.

The actual definition is the range of acceptably in focus when viewed by someone looking at an 8" x 10" print, where a point source of light was now a circle no larger than 0.01 inches or 0.25 mm. Work this value backwards and you will get the 30 micron circle of confusion for a 35mm full frame and 20 micron circle of confusion for an APS-C sensor

But otherwise , I agree.


Now, for software that simulates full frame on APS-C, it exists, has existed for almost as long as digital, maybe longer, it is called panoramic splicing of shots.

Take 2 shots in portrait mode(perhaps 3 with overlap) and you have full frame shots.

[KevinR]

...Now, for software that simulates full frame on APS-C, it exists, has existed for almost as long as digital, maybe longer, it is called panoramic splicing of shots....

Thanks Lowell. Great explanation on the DOF and the relationship to sensor size and CoC. I knew most of this at a basic level, but never thought about the impact of this on DoF of the APS-C to resemble DOF of FF. Is my new understanding then correct that the full effect of DoF of FF can be acieved by about a 2 pano stitch of APS-C, and effectively one could exceed DoF of FF with a multi-pano stitch??

[Fogel70]

Thanks Lowell. Great explanation on the DOF and the relationship to sensor size and CoC. I knew most of this at a basic level, but never thought about the impact of this on DoF of the APS-C to resemble DOF of FF. Is my new understanding then correct that the full effect of DoF of FF can be acieved by about a 2 pano stitch of APS-C, and effectively one could exceed DoF of FF with a multi-pano stitch??

It's correct, but if shooting with aperture wide open you probably need a lot of overlapping on the images.
Many fast lenses have noticeable less sharpness on edges wide open, which will look weird if stitching without overlapping.

[KevinR]

It's correct, but if shooting with aperture wide open you probably need a lot of overlapping on the images.
Many fast lenses have noticeable less sharpness on edges wide open, which will look weird if stitching without overlapping.

Excellent. Thanks for that and a very useful take-away for me. I do normaly overlap heavily as a rule, so probably safe there. I suppose the point was that the Field-of-View area of 2 portrat APS-C images is about the same as that of a FF image. So all else being the same in the DoF determination function (FL, A, distance) with CoC as the only difference between APS-C and FF, increasing the Field-of-View to the same with multiple images then makes the CoC impact similar. It now all makes sense.

Thanks Lowell and @Fogel70. Thumbs up.

[Lowell Goudge]

I can already hear the arguments though. It is hard to do a portrait with a stitch

[starjedi]

In camera software or desktop app? I know it wouldn't be the same FOV but you would adjust to what you get after the sensor crop...but get the desired shallow depth of field via software simulation app?
Just a question since there is the AA removal simulation in K3.

You can always simulate the shots in FF with APS-C! Eg, for a FF with 50mm F4.5 ISO 200, you definitely can use 75mm F3.0 ISO 100 in pentax APS-C camera to give almost identical effect (assuming the same pixels) despite that under low light enviorments FF might do better job as the area for each CMOS point is 2x larger comparing to APS-C CMOS.

It is easy for Pentax/Canon/Nikon et al to provide this kind of functionality but it will only create confusion as the FF does not rule the world. I can imagine at one time the high-end camera will shift to medium format as long as the cost for making a medium format CMOS dropped signficiant. The larger CMOS/CCD size under the same pixels always will deliver better photography quality, at least in paper.

[kh1234567890]

Thanks Lowell. Great explanation on the DOF and the relationship to sensor size and CoC. I knew most of this at a basic level, but never thought about the impact of this on DoF of the APS-C to resemble DOF of FF. Is my new understanding then correct that the full effect of DoF of FF can be acieved by about a 2 pano stitch of APS-C, and effectively one could exceed DoF of FF with a multi-pano stitch??

Yes, it is called the Brenizer Method - great fun.

[Na Horuk]

There are also "lens booster" things that compress the image circle, but I don't think they exist for K-mount (yet?)
Or you can use something like the Topaz Lens Effects software, which promises a lot. Never tried it, not sure it can deliver.
You can also just open the photo in Photoshop and add blur (regular blur, Gaussian blur, etc.) selectively.

And the final option is buying a lens with bigger aperture (smaller f-number) - I assume this is what you are after. This is when the OoF area looks very blurred, smudged, and the depth of field is very shallow. So a lens like FA 50mm f1.4 Or on the cheap, one of those Lens Baby combinations (some are just blurry, others allow tilt, which creates oddly shallow DoF), but those will not achieve anywhere near the same quality as the FA 50mm. Just keep in mind that this type of photography is very cliche and gets boring fast. There is no such thing as insta-art

[Mazhe]

There are also "lens booster" things that compress the image circle, but I don't think they exist for K-mount (yet?)

They only work with lenses with really long registration distance compared to the body (eg SLR lenses with mirrorless bodies), so don't expect them for K-mount cameras.

[falconeye]

Yes, it is called the Brenizer Method - great fun.

True, although he didn't invent the method.
When first discussed on a forum, people wondered how to describe the effect numerically.

Not wanting to beat a dead horse, but the simplest take at the method is equivalency: Just consider the total area of the stitched image (when still at the image plane) to be the sensor size, and apply known equivalency formulas. The shallow depth of field is a result of the large physical aperture of the kens used when stitching: it has a longer focal length which at the same fstop, translates to a larger mm-diameter.

For a stitched image, you easily end up with some ridiculous 35mm-equivalent settings for ISO and fstop, like F/0.25

[Paul MaudDib]

There are fake shallow DOF effects you can do in post, they are commonly used with cell phone cameras and other small sensor compacts because it is otherwise almost impossible to achieve the effect. It can be easy to pick out the poorly done fakes because there is usually something sharp that shouldn't be or vice versa. Also, you can't always easily mimic the unique bokeh that some glass is known for.

The problem with this technique is that the algorithm doesn't really have any way to measure how much bokeh it should give any given pixel, other than guessing, but you can construct this data and do a better job. Shoot a stereo frame and combine the two to produce 3D depth data using the parallax, then apply a gaussian blur based on how far from the desired plane of focus any given pixel is. Pixels that are close get little blur, pixels that are in the background get a lot of blur. And because you're applying the bokeh in software, you can make it look however you want, in theory.

You can shoot two stereo half-frames onto a single sensor and avoid the need for a stitch or a stereo camera. Panasonic makes such a lens for the M4/3 system.



Of course it would work better on APS-C or full frame, because half of a tiny sensor is even smaller. Or you can simply use two separate sensors and lens systems (or bodies).

Ideally you would have a wider stereo base so you could get better parallax data, but it would probably suffice, particularly up close.

[Paul MaudDib]

It's correct, but if shooting with aperture wide open you probably need a lot of overlapping on the images.
Many fast lenses have noticeable less sharpness on edges wide open, which will look weird if stitching without overlapping.

Don't forget vignetting as well. You're stitching together the edges, so the areas of least brightness tend to overlap and you'll get seam lines in your sky, etc.

[Paul MaudDib]

They only work with lenses with really long registration distance compared to the body (eg SLR lenses with mirrorless bodies), so don't expect them for K-mount cameras.

You could theoretically go from a lens with an even longer registration distance to a DSLR, as that would give room to stick the optics in there. For example you could go from medium format to K mount, probably. The advantage of doing it this way is when you compress the image circle you gain a couple stops of speed and the image becomes sharper (even oughtweighing the image quality loss of the optics). MF to APS-C would be at least several stops, something like 3-4 I think.

No one makes them for MF though. It'd be a pretty hefty chunk of glass anyway.