Sony 16mm Fisheye and DXO

[Old Hydro]

Does anyone have experience correcting the curvature from the 16mm fisheye with DXO. The software claims it can do it. I like the idea of 180 degree view that can be corrected, but before I spend a $1000, I'd like to know if its all advertising or a correction that really works.

I'd like a lens wider that my 20mm, but it seems like the only choices are the 16mm fisheye, the CZ 16-36 and the Sigma 12-24mm.

I shoot with the a850. Might look hard at the a99 if it becomes a real product.

[Dr. Harout]

How about Samyang 8mm?

[pakodominguez]

How about Samyang 8mm?

That fisheye is APS only. But Samyang have a 14mm that they claim is FF ( http://www.samyang.plproduct,76,category,5,samyang_14mm_f28_if_ed_umc_aspherical ). Adorama sells this lens on "Bower" version, but they claim is "optimized" for APS, even if in the descriptin they say "
The 14mm f/2.8 works perfectly even with latest high resolution full frame sensors." ( http://www.adorama.com/LNB1428SO.htm )

Correcting the fisheye curvature won't get you wider than 20mm because you'll crop a lot on the corners.

if 17mm is wider enough for you, get the KM 17-35 f2.8-4, that is a wonderful lens for about 400 US$

[agorabasta]

Pako, once again, what???

The 14mm Samyang is no fisheye in any way. It's an absolutely fine lens on APS. It's wonderful on FF with only one caveat - moustache type of distortion, though rather mild.

The 8mm APS Samyang is a fisheye, though not an extreme one. If you crop it to perfect rectilinear output, it makes about 12mm cropped equivalent. And guess what - shimming the integrated hood turns it into an FF lens easily. Then if you plan to correct it till perfectly rectilinear output, no shimming is even needed.
But the problem is that it's rather unusable wide open, you may really use it starting f/5.6-f/8; then the focus control calibration is about inexistent - my APS bodies catch infinity just above 0.4m mark, etc, etc. But you may still get stunning results with that lens in good enough light. In a word - it's a landscape/beach thingie.

[pakodominguez]

Pako, once again, what???

I know my English is not the best one, but read again and you'll see we are talking about the same thing.

The 14mm Samyang is no fisheye in any way. It's an absolutely fine lens on APS. It's wonderful on FF with only one caveat - moustache type of distortion, though rather mild.

I never said the 14mm was a fisheye. The OP doesn't need a fisheye but a wider than 20.mm (rectilinear) lens.

The 8mm APS Samyang is a fisheye, though not an extreme one. If you crop it to perfect rectilinear output, it makes about 12mm cropped equivalent. And guess what - shimming the integrated hood turns it into an FF lens easily. Then if you plan to correct it till perfectly rectilinear output, no shimming is even needed.

If you want to use the whole FF, not shimming the lens cause a huge vignetting, so you probably have that 12mm you are talking about, but not at 24MP...

But the problem is that it's rather unusable wide open, you may really use it starting f/5.6-f/8; then the focus control calibration is about inexistent - my APS bodies catch infinity just above 0.4m mark, etc, etc. But you may still get stunning results with that lens in good enough light. In a word - it's a landscape/beach thingie.

You got a bad copy? Some people on this forum love that lens...

In any case, if manual focus is not a problem, the 14mm is a way better solution IMHO

[agorabasta]

Pako, you've said that correcting fisheye would get at 20mm. That would be true for a 14mm fisheye with distortion close to that 8mm Samyang.

All the 8mm Samyangs are 'bad copies'. And they all produce fine results at certain conditions. And then you see those admirably fine pix posted around.

The 14mm Samyangs are truly fine lenses. They even produce some fine bokeh when focused close at f/2.8 which is most improbable with UWA lenses in general. Even the smaller blur in the transition area looks beautiful - a rare quality indeed.

[pakodominguez]

Pako, you've said that correcting fisheye would get at 20mm. That would be true for a 14mm fisheye with distortion close to that 8mm Samyang.

My bad.

I was looking for wider than my KM 17-35 last year that I had a client that hired me for photographing "fancy" interiors decoration and one of the solutions I though was exactly that: de-fishing fisheyes. It just doesn't work that well quality-wise.

I discarded the Sigma 12-24 because the sampling issues with it. and, at that time, the Samyang 14mm was announced, but was not available. so I stick with my KM 17-35...

[agorabasta]

One more thing - a true 180deg fisheye on FF should produce full 180deg coverage both horizontally and vertically, which means that the FF captures a perfectly round image that is ideally covering it's full height, which in turn means that the side edges of the sensor remain unexposed. It would mean that an 8mm 180deg horizontal APS fisheye is exactly a full 180deg FF fisheye. But the 8mm Samyang is a diagonal 180deg APS, which means that it's full image circle is larger than the 35mm height, being exactly the APS diagonal.

So the term 'APS diagonal" in the name of the lens refers to the diameter of its image circle which is greater than that of a 'circular' FF fisheye.

Hope I clarified something here instead of muddying the matters further yet.

[01af]

Pako, you've said that correcting fisheye would get at 20 mm. That would be true for a 14 mm fisheye with distortion close to that 8 mm Samyang.

That's nonsense.

Pako's statement about the Samyang 14 mm wide-angle lens and his statement about correcting a fish-eye lens appeared in two separate paragraphs. Any connection between these two is solely in your head.

De-fishing an image taken with a fish-eye lens will yield an image equivalent to that from a wide-angle lens the same focal length as the fish-eye. So de-fishing an image taken with the Sony SAL 16 mm, for example, will get you an image with the coverage of a 16 mm wide-angle lens ... in theory. However in practice, the farthest corners will be very poor in image quality so it will make sense to cut them away, so you'll arrive at the equivalent of a slightly longer lens. Still I'd guess it's not 20 mm but maybe 17 or 18 mm.

... a true 180° fisheye on FF should produce full 180° coverage both horizontally and vertically, which means that the FF captures a perfectly round image ...

Duh!

... that is ideally covering its full height ...

Typically, a circular-image fish-eye lens will cover slightly less than the frame's full height. For example, the image circle diameter usually is 22.5 - 23.0 mm for 35-mm format, not 24 mm. And by the way—a circular-image fish-eye lens is not any more 'true' than a full-frame fish-eye lens. They simply are two different kinds of fish-eye lenses, both equally 'true.' The full-frame type usually suits pictorial work better; the circular-image type is better for some kinds of scientific or industrial applications, and maybe also for those who want to create 3D spheric panoramas with a minimum number of shots.

It would mean that an 8 mm, 180° horizontal APS fisheye is exactly a full 180° FF fisheye.

This statement it just garbage. It is not even clear what you actually are trying to say because 'horizontal fisheye' lenses don't exist, and the 'FF' abbreviation is ambiguous in this context ... but the statement is nonsense in either interpretation anyway.

But the 8 mm Samyang is a diagonal 180° APS, which means that its full image circle is larger than the 35 mm height, being exactly the APS diagonal. So the term 'APS diagonal' in the name of the lens refers to the diameter of its image circle which is greater than that of a 'circular' FF fisheye.

That's correct but trivial. So what's your point?

[agorabasta]

01af, let me guess - you've got mistreated at the kindergarten and come here venting, right?

[KevinBarrett]

No need to be inflammatory, boys.