Curvature and WA lenses

I Simonius

(is this the right place to post?)

I notice with my Canon 17-40L @ 17mm WA shots that the horizon is curved slightly in the local landscapes where the horizon is either featureless (marshlands) or it is the sea and where it is in less that the top third of the frame, e.g the higher up the frame it gets the more it curves. (obviously)

I just wondered whether there are any lenses in the 17-12 range ( ideally Canon, but must work with a Canon 5D for me) for FF cameras which don't exhibit this phenomenon

Lastly what is the best (read easiest that actually works) way to straighten the horizon in photoshop (on a Mac) without degrading the image?

thanks

[Stiffkey Norfolk UK]

davidweaver

Try 'Warp" in PS and bring up the corners a tiny bit.

Getting a perfect rectilinear image on a zoom lens is tough. The tiny bit of curvature isn't bad to me at all.

Cheers!

I Simonius

davidweaver wrote:

Try 'Warp" in PS and bring up the corners a tiny bit.

Getting a perfect rectilinear image on a zoom lens is tough. The tiny bit of curvature isn't bad to me at all.

Cheers!

I'll try that thanks

It doesn' look too bad on that pic perhaps because not being local you don't expect it to be perfectly flat, but it looks bad on the Sea which everyone expects to be straight

SloYerRoll

I agree w/ David that the barelling on this specific shot isn't that bad. It can get ALLOT worse that that on lesser glass.

Regarding your Ps corrections. Another simple way to do it is to the warp tool. Choose pucker and make the brush about twice the size of the image (yes, that frickin huge ) then just click once or twice w/ the center of the brush in the middle of the document. You'll then see the barreling on all sides start to go away. Of course, you'll need to play w/ the brush size etc. But once you get the settings down. You can repeat these results every time.

BTW: when dealing w/ wide angle shots in landscape mode. It's pretty common to see some curvature on the horizon. After all, the earth is round. And the far end of your shots can cover hundreds (if not thousands) of miles in width.

I Simonius

SloYerRoll wrote:

I agree w/ David that the barelling on this specific shot isn't that bad. It can get ALLOT worse that that on lesser glass.

Regarding your Ps corrections. Another simple way to do it is to the warp tool. Choose pucker and make the brush about twice the size of the image (yes, that frickin huge ) then just click once or twice w/ the center of the brush in the middle of the document. You'll then see the barreling on all sides start to go away. Of course, you'll need to play w/ the brush size etc. But once you get the settings down. You can repeat these results every time.

BTW: when dealing w/ wide angle shots in landscape mode. It's pretty common to see curvature on the horizon. After all, the earth is round

thanks for the tips

I know it can get worse on lesser glass but it would be nice to find glass that didn't do it at all

silversx80

Correct me if I'm wrong but isn't that just the physics of the wide angle? I mean, you're getting closer to a fisheye the smaller the focal length of the lens.

SloYerRoll

silversx80 wrote:

Correct me if I'm wrong but isn't that just the physics of the wide angle? I mean, you're getting closer to a fisheye the smaller the focal length of the lens.

That's definitely the physics of a wide angle lens. But knowing those physics exist can allow you to make adjustments to counter those effects.

I'm not much of a glass hound, so I'm thinking out loud here:
I think one of the big problems lens makers deal w/ on wide angle is that they have to deal w/ the zoom aspect of it. It a piece of glass was a static 12mm. It would be much easier to correct for any distortion. As soon as you throw different zoom angles in the mix. I'm pretty sure the problem gets exponentially harder to solve.

Anyone know if this is true or total rubbish?

I Simonius

silversx80 wrote:

Correct me if I'm wrong but isn't that just the physics of the wide angle? I mean, you're getting closer to a fisheye the smaller the focal length of the lens.

of course - this one's rectilinear however ( unlike fisheye) - and it may well be that in this day and age of high tech glass that someone has comeup with the (almost) perfect rectilinear lens i.e. one that controls distortions to to a barely perceptable minimum

I Simonius

SloYerRoll wrote:

That's definitely the physics of a wide angle lens. But knowing those physics exist can allow you to make adjustments to counter those effects.

I'm not much of a glass hound, so I'm thinking out loud here:
I think one of the big problems lens makers deal w/ on wide angle is that they have to deal w/ the zoom aspect of it. It a piece of glass was a static 12mm. It would be much easier to correct for any distortion. As soon as you throw different zoom angles in the mix. I'm pretty sure the problem gets exponentially harder to solve.

Anyone know if this is true or total rubbish?

dunno but that was sort of my question i.e. is the 14mm mk2 ( or any other lens) better at this sort of distortion?

claudermilk

Well, you DO realize the world is round, right? wink

Anyway, the UWAs just tend to do this. It's the nature of the beast. I use PTLens (or rather, Bibble's variation of it) to adjust for lens distortion when I need to.

BTW, if you've even been at sea, you would notice the horizon does indeed visibly curve a bit.

SloYerRoll

claudermilk wrote:

BTW, if you've even been at sea, you would notice the horizon does indeed visibly curve a bit.

I thnk this was the reason I mentioned this. I was at sea for over 5 years (not hte entire time. was in the Navy on Subs) and spent allot of it in the middle of the ocean. It's pretty cool to actually see the curve in the earth like you mentioned.

BTW: being out in the ocean like that really puts into perspective how small we really are.

I Simonius

SloYerRoll wrote:

I thnk this was the reason I mentioned this. I was at sea for over 5 years (not hte entire time. was in the Navy on Subs) and spent allot of it in the middle of the ocean. It's pretty cool to actually see the curve in the earth like you mentioned.

BTW: being out in the ocean like that really puts into perspective how small we really are.

I dooo understand what you guys are saying about the world being round.. really.. it's just not quite that small though, i.e. the curvature in my pics would shave enough off the earth were they continued round the shpere to make it about the size of the moon

The problem is that the average viewer will notice it as being not quite normal....unless they from the moon

SloYerRoll

I Simonius wrote:

I dooo understand what you guys are saying about the world being round.. really.. it's just not quite that small though, i.e. the curvature in my pics would shave enough off the earth were they continued round the shpere to make it about the size of the moon

The problem is that the average viewer will notice it as being not quite normal....unless they from the moon

I knew you understood that. I was jsut bringing the point to attention

I was just making a further comment on the whole ocean thing

Art Scott

I Simonius wrote:

of course - this one's rectilinear however ( unlike fisheye) - and it may well be that in this day and age of high tech glass that someone has comeup with the (almost) perfect rectilinear lens i.e. one that controls distortions to to a barely perceptable minimum

Tilt and Shift perhaps................

SloYerRoll

Isn't it DXO optics that corrects all these flaws in different lens'? It references a huge database that reads your exif data and tells the program how to correct any isses that the lens might have.

I've never used it. But have heard rave reveiws about it.

I Simonius

SloYerRoll wrote:

I knew you understood that. I was jsut bringing the point to attention

I was just making a further comment on the whole ocean thing

ahh 'tis the ocean life for me:D

I Simonius

Art Scott wrote:

Tilt and Shift perhaps................

Not sure if that's tongue in cheek Art.....?

T&S will not effect curvature AFAIK:cry

I Simonius

SloYerRoll wrote:

Isn't it DXO optics that corrects all these flaws in different lens'? It references a huge database that reads your exif data and tells the program how to correct any isses that the lens might have.

I've never used it. But have heard rave reveiws about it.

you may be right - I'll have to take another look

I DLd the trial ages ago but it ran out before I got round to testing it properly

I mean where does one get the time?

Robinivich

For software that you probably already have, try Canon's DPP (Digital Photo Professional)

The latest versions (free to update) have a lens aberration correction function that's calibrated exactly to the lens, it takes everything into consideration, even the focus setting if the lens provides it. Also handy for undoing vignetting and CAs, I've tired it with my 10-22 on a 40D and the results are impressive.

:Edit: Forgot to mention, the 17-40L and 5d are compatible and listed in the combos that can work, it seems the 5d doesn't record the focus setting but you can set it manually. All these adjustments work on RAW files only (though it's implied in the manual that canon's working on that)

I Simonius

Robinivich wrote:

For software that you probably already have, try Canon's DPP (Digital Photo Professional)

The latest versions (free to update) have a lens aberration correction function that's calibrated exactly to the lens, it takes everything into consideration, even the focus setting if the lens provides it. Also handy for undoing vignetting and CAs, I've tired it with my 10-22 on a 40D and the results are impressive.

:Edit: Forgot to mention, the 17-40L and 5d are compatible and listed in the combos that can work, it seems the 5d doesn't record the focus setting but you can set it manually. All these adjustments work on RAW files only (though it's implied in the manual that canon's working on that)

thanks - great info - I have never tried DPP but I guess now I will!

ziggy53

ACR and even RAW Therapee also have lens correction in RAW processing which is a great benefit and then PS CS2/CS3 also have lens correction filters. (RAW Therapee can also open common raster image files, not just RAW.)

An important thing to remember is that as you move the horizon away from the image center, the curvilinear distortion tends to become more obvious. While this is sometimes unavoidable, you may have to move the image onto a larger Canvas and off-center in order to achieve the proper correction.

Curvilinear distortion is not corrected in fisheye lenses and that is what distinguishes them, not necessarily short focal length. It is true that very short fisheye lenses do tend to produce the most curvilinear distortion and the most obvious fisheye effect.

Circular fisheye are the most distorted of the bunch and on the appropriate format can produce 180 degrees (or more) of image view, something rectalinear lenses cannot achieve.

You can induce a fisheye effect in software by doing the opposite of curvilinear correction and, especially on short focal length lenses, the effect is quite authentic looking (like you had used a genuine fisheye lens).

I Simonius

ziggy53 wrote:

ACR and even RAW Therapee also have lens correction in RAW processing which is a great benefit and then PS CS2/CS3 also have lens correction filters.

Ill take a decko at ACR then too

thanks

claudermilk

I Simonius wrote:

I dooo understand what you guys are saying about the world being round.. really.. it's just not quite that small though, i.e. the curvature in my pics would shave enough off the earth were they continued round the shpere to make it about the size of the moon

The problem is that the average viewer will notice it as being not quite normal....unless they from the moon

SloYerRoll wrote:

I knew you understood that. I was jsut bringing the point to attention

I was just making a further comment on the whole ocean thing

...and I was at least in part being a smart-aleck.

Anyway, my siuggestion for PTLens stands. It would adjust that effect & from the last time I checked, their lens DB was far more complete than DxO.

I Simonius

claudermilk wrote:

...and I was at least in part being a smart-aleck.

Anyway, my siuggestion for PTLens stands. It would adjust that effect & from the last time I checked, their lens DB was far more complete than DxO.

but I am running a pre-Intel PPC mac on 10.4.11 and...

"Mac Version
PTLens runs on Intel Mac computers under Mac OS 10.5.2 (Leopard) or later with a monitor resolution of at least 1280x800. Currently only the standalone version is available. The plug-in version is under development."

so I don't think it'll work for me:cry

I Simonius

found this from the PTlens link - thanks

http://www.kekus.com/software/plugin.html