Laowa 9mm F2.8 Zero-D (Fujifilm)

[MD]

Hi,

It appears there is no profile for this lens. How do i go about making the *perfect* profile? Because the optimized results are different, sometimes very different, with each panorama. Resulting in different output resolutions at '100% optimum size' as well. (I guess i need to adjust my setup a little better..) 

Also there is a lot of vignetting (like +2EV), and some color cast to deal with. Both of which are corrected fairly accurately with an LCC profile. But I'm not sure what works better (to prevent banding in skies and such). To correct these things in advance, maybe just the color cast (since PTgui probably doesn't), maybe neither, or maybe it shouldn't be an issue either way. The exposure optimizer still detects vignetting even if it seems perfectly corrected already, though.

ps. I also have an 8mm Fisheye, and i was expecting some difference but man, the Laowa may be sharper with more MP in the end, but the Fisheye has so much more capture area to work with. And that often means better consistency. So not sure what i'll keep using. Is a non-fisheye even recommended at all?

Cheers,

[Erik Krause]

Am 28.03.2021 um 17:16 schrieb MD:

> Is a non-fisheye even recommended at all?

Rectilinear ultra-wide lenses offer no advantage over fisheyes with the
same focal length. You just need many more shots to cover the sphere,
but you get the same resolution.

--
Erik Krause
http://www.erik-krause.de

[PTGui Support]

PTGui only includes lens profiles for fisheye lenses. This is because it
needs to know the lens projection before generating control points, and
there are different kinds of fisheye projections. You lens has a perfect
rectilinear projection so this is not needed.

Once you have a succesful stitch, you can save it as a template. Keep in
mind that vignetting depends on the lens aperture.

Kind regards,

Joost Nieuwenhuijse
www.ptgui.com

On 28/03/2021 17:16, MD wrote:
> Hi,
>
> It appears there is no profile for this lens. How do i go about making
> the *perfect* profile? Because the optimized results are different,
> sometimes very different, with each panorama. Resulting in different
> output resolutions at '100% optimum size' as well. (I guess i need to
> adjust my setup a little better..)
>
> Also there is a lot of vignetting (like +2EV), and some color cast to
> deal with. Both of which are corrected fairly accurately with an LCC
> profile. But I'm not sure what works better (to prevent banding in skies
> and such). To correct these things in advance, maybe just the color cast
> (since PTgui probably doesn't), maybe neither, or maybe it shouldn't be
> an issue either way. The exposure optimizer still detects vignetting

> even if it /seems/ perfectly corrected already, though.

>
> ps. I also have an 8mm Fisheye, and i was expecting some difference but
> man, the Laowa may be sharper with more MP in the end, but the Fisheye
> has so much more capture area to work with. And that often means better
> consistency. So not sure what i'll keep using. Is a non-fisheye even
> recommended at all?
>
> Cheers,
>

> --
> You received this message because you are subscribed to the Google
> Groups "PTGui Support" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to ptgui+un...@googlegroups.com
> <mailto:ptgui+un...@googlegroups.com>.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/ptgui/91201667-2381-4309-8473-631eca897873n%40googlegroups.com
> <https://groups.google.com/d/msgid/ptgui/91201667-2381-4309-8473-631eca897873n%40googlegroups.com?utm_medium=email&utm_source=footer>.

[MD]

I see, a fisheye needs a profile for stitching, while a rectilinear lens only needs some small optimizations after. So if i take the average of the best results for example and save that as a profile. That would allow me to safely disable the focal length calculation in the optimizer to avoid it landing on 8mm again, and turning the final image into a 90MP one instead of the expected 120MP? But the other values would then be re-calculated again anyway..

As for the vignetting, yes of course the LCC profiles are aperture dependent. I'll keep using pre-corrected images then.

Op maandag 29 maart 2021 om 10:10:32 UTC+2 schreef PTGui Support:

[MD]

Just one more in vertical orientation, or 2 if very generous with the overlap. A small effort anyway.

The top and bottom images when straight up/down cover the holes exactly with about 0.1% overlap at the narrowest though, so that's no fun.

On 2021-03-28 21:02, Erik Krause wrote:

> 9.5% more linear resolution at the expense of how many more shots?

--
Erik Krause
Herchersgarten 1
79249 Merzhausen 

Am 28.03.2021 um 20:50 schrieb Mdnjou:

> Indeed. But in this case it amounted to about a 20% increase. 100 >
120MP final image.

Op zondag 28 maart 2021 om 20:16:48 UTC+2 schreef Erik Krause:

[Erik Krause]

Am 29.03.21 um 11:24 schrieb MD:

> That would allow me to safely disable the focal length calculation in
> the optimizer to avoid it landing on 8mm again

You should leave the focal length at the value PTGui optimizes it to,
for several reasons. One is that focal length can differ up to 5% from
the written value. The other is that lens distortion might actually
change the field of view which PTGui uses internally. The focal length
display is calculated from that assuming no distortion and hence is for
display only.

--
Erik Krause

[MD]

That's why you create a profile and/or optimize isn't it, to find the true focal length. But if 9 out of 10 times it ends up between 9.2 - 9.3mm, but then it calculates one at 8.2mm, i'm supposed to leave it at that? Compensate the output with a 110% optimum resolution or something. If so then why even have a profile?

Op maandag 29 maart 2021 om 12:00:57 UTC+2 schreef Erik Krause:

[Erik Krause]

Am 29.03.21 um 12:19 schrieb MD:

> But if 9 out of 10 times it ends up between 9.2 - 9.3mm, but
> then it calculates one at 8.2mm, i'm supposed to leave it at that?

Depends. It's the value at which the provided control points fit best.
The reason for that might be parallax, control points clustered in a
small part of the overlap or too few or misplaced control points. So if
the panorama looks good on visual inspection (along the seams) I'd leave
it at that. If not, check for the above points.

If you shoot for a lens profile your panoramic head should be perfectly
adjusted and you should shoot in a large space like a cathedral. And
don't set control points on nearby objects to avoid any possibility of
parallax but set lots of them well distributed throughout the overlap.

--
Erik Krause