Barrel distortion at DVinfo.net

Brian Luce · June 9th, 2009, 10:36 PM

I tested a 5d today and the 17-40mm canon lens got a huge ripple of barrel distortion during panning at all focal lengths. 17 to 40. Is this because of the large sensor?

How much do you have to spend to get a lens that's usable? This was an L series and cost about $900.

Jim Newberry · June 10th, 2009, 01:39 AM

Quote:

Originally Posted by Brian Luce

I tested a 5d today and the 17-40mm canon lens got a huge ripple of barrel distortion during panning at all focal lengths. 17 to 40. Is this because of the large sensor?

How much do you have to spend to get a lens that's usable? This was an L series and cost about $900.

No, I don't think it's because of the large sensor (although with a smaller sensor the distortion might be less obvious). I'm not sure if there are any extremely wide angle lenses (such as 17mm) that have no barrel distortion--especially zoom lenses. I have the Canon 24-70mm which is a great lens, but it too suffers from barrel distortion. I use PTlens to correct for it when it's an issue (often it's not a problem). PTlens is for stills though...

Marcus Marchesseault · June 10th, 2009, 03:25 AM

Are you sure you are describing significant barrel distortion? All the expensive short lenses I have seen do a good job of keeping the lines fairly straight. Are you sure you aren't just seeing the extreme forced perspective of a short lens on a full-frame sensor? Anything but a "normal" lens is going to have effects on the perspective in the scene. This may not be so obvious on a still shot with a wide lens, but panning is where it will be really apparent. Since you say it is during a pan, I don't think you are talking about barrel distortion but rather the forced perspective of a short (wide) lens. If you don't want that effect, you must use something very close to 50mm. If you go to a longer lens, you will experience a flattening of the perspective where the relation of the distance between objects becomes less apparent. In a short lens, the distance between objects in the scene is exaggerated and distorted compared to what is perceived by the human eye. Some people really like this effect.

Richard Hunter · June 10th, 2009, 04:15 AM

If it is only apparent during panning could it be rolling shutter effects?

Richard

Brian Luce · June 10th, 2009, 07:57 AM

Quote:

Originally Posted by Marcus Marchesseault

Are you sure you are describing significant barrel distortion? All the expensive short lenses I have seen do a good job of keeping the lines fairly straight. Are you sure you aren't just seeing the extreme forced perspective of a short lens on a full-frame sensor? Anything but a "normal" lens is going to have effects on the perspective in the scene. This may not be so obvious on a still shot with a wide lens, but panning is where it will be really apparent. Since you say it is during a pan, I don't think you are talking about barrel distortion but rather the forced perspective of a short (wide) lens. If you don't want that effect, you must use something very close to 50mm. If you go to a longer lens, you will experience a flattening of the perspective where the relation of the distance between objects becomes less apparent. In a short lens, the distance between objects in the scene is exaggerated and distorted compared to what is perceived by the human eye. Some people really like this effect.

I really don't know, but for sure you can't pan with that lens. I hope you're right. That 17-40L series doesnt seem suitable for video.

Don Miller · June 10th, 2009, 08:25 AM

No reason to guess when you can post an example. I don't have the 17-40, but plenty of video has been shot with it. The lens is considered professional and a good value. Barrel distortion is more of a problem with still than video.

Brian Luce · June 10th, 2009, 08:28 AM

Quote:

Originally Posted by Don Miller

No reason to guess when you can post an example. I don't have the 17-40, but plenty of video has been shot with it. The lens is considered professional and a good value. Barrel distortion is more of a problem with still than video.

I can't, I was just playing with one in the shop for about 30 min, tried a few lenses and noticed that rolling affect during any pan with that particular lens.

Olof Ekbergh · June 10th, 2009, 08:50 AM

I have shot quite a bit with that lens.

It is very wide, so naturally it has some distortion.

Below is video that has quite a few shots using the 17-40 L.

Most of the shots in the roundhouse are with that lens. Some of them full 17mm frame. Check the tilt up on the engine with steam leaking out.

Some of the outdoor shots are with the 24-105 F4 also with the 70-200 F2.8.

CSRR 5 minutes of Winter Steam

Most of the footage is from EX3's.

Brian Luce · June 10th, 2009, 10:15 AM

Is *any* of this footage with the 5d?
Nice images btw.

Quote:

Originally Posted by Olof Ekbergh

I have shot quite a bit with that lens.

It is very wide, so naturally it has some distortion.

Below is video that has quite a few shots using the 17-40 L.

Most of the shots in the roundhouse are with that lens. Some of them full 17mm frame. Check the tilt up on the engine with steam leaking out.

Some of the outdoor shots are with the 24-105 F4 also with the 70-200 F2.8.

CSRR 5 minutes of Winter Steam

Most of the footage is from EX3's.

Tom Hardwick · June 10th, 2009, 10:23 AM

Just to make it quite plain that the sensor size has no bearing on a lens's barrel distortion. I'm quite surprised to hear you say you have barrel distortion on Canon's 17 mm 'while panning'. Barrel distortion is a bending outwards of all straight lines that don't pass through the centre of the frame, so panning also has nothing to do with barrel distortion.

Canon's L series are usually pretty well corrected, and a 20 mm Canon prime I had zero barrel distortion. At the price I'd expect the zoom to be the same, albeit with a more restricted maximum aperture.

I'm pretty sure you're seeing the effects of CMOS's rolling shutter, Brian.

tom.

Olof Ekbergh · June 10th, 2009, 10:36 AM

Brian,

To answer your question.

"Is *any* of this footage with the 5d?
Nice images btw."

All the footage shot with Canon lenses are with the 5DmkII.

On the EX3 I use the stock lens.

The opening shot is 5D, the next two as well, shot before sunrise.

The shots of the steam engine sitting emitting steam, and the door opening, some of the shots in the cab are with the 5D and 17-40.

Later in the clips there is an interesting comparison EX3 to 5D, the two shots of the train going through the Iron bridge was shot at the same time, the one from the right side of tracks was 5D using 24-105 F4, the one from the left side lower down was the EX3.

I find the 5D footage needs some softening, I do that in Color.

These clips are just sort of a teaser of the final production we are currently working on.

Bill Binder · June 10th, 2009, 10:56 AM

As someone who owns both FF and 1.6x cropped bodies, as well as the 24-70mm and 16-35mm Canon lenses, maybe I'm missing something here, but FF does play a role here doesn't it?

True, the lens distortion is what it is no matter FF or crop, but with a crop you are way more in the center of the lens image and with FF you go way farther out to the sides of the lens image. So, with the crop, you're likely to see less "distortion" than with the FF, but of course it won't be as wide as FF either. My point is, it would be pretty easy for someone to use a particular wide lens on a crop and think it doesn't have much distortion, but when someone uses the exact same lens on a FF they would. No?

Also, I kind of doubt this is rolling shutter related given the description. Rolling shutter during a pan would create diagonals of verticals, not barrels curving on all boundaries. No?

Finally, I'm thinking using the term "barrel" here might be misleading, although I'll let the OP opine on this. I say this because both my 24-70 and 16-35II do a pretty damn good job of keeping barreling to a minimum even at full wide -- meaning lines stay mostly straight. HOWEVER, that doesn't mean there isn't significant distortion at the sides say at full 16mm wide -- meaning a face could look mighty stretched if off in a corner as opposed to the middle of the frame. Also, although this type of distortion can be obvious in a still, and can be mitigated somewhat depending on composition, all bets are off when seen during a panning video because then objects are being panned into and out of the major distortion zones, and it becomes hugely obvious -- to me at least. This is pretty normal from my take on things, but what do I know?

You could always try the new wide Tilt-Shift if you have the coinage, although not sure how that would pan either, heh...

Evan Donn · June 10th, 2009, 02:19 PM

just tried out the 14mm f2.8 L II this weekend (~$2k) and while it has almost no visible barrel distortion it does have quite a bit of perspective distortion which is especially apparent when panning. I think that's simply the nature of extreme wide angle lenses, and usually you use them specifically for that ability to exaggerate perspective.

Jim Newberry · June 10th, 2009, 04:48 PM

Quote:

Originally Posted by Bill Binder

As someone who owns both FF and 1.6x cropped bodies, as well as the 24-70mm and 16-35mm Canon lenses, maybe I'm missing something here, but FF does play a role here doesn't it?

Yes, it does play a role, as you pointed out (it's just that the sensor obviously doesn't cause the distortion). I haven't used the 17-40 in a long time, but my Canon 24-70 definitely has barrel distortion--no question about it. I highly doubt there are any wide angle zooms that are completely devoid of it.

If anyone doubts that wide angle zooms have barrel distortion, try shooting an interior at the widest focal length, and see if you can get the verticals perfectly straight and parallel (without correcting in post).

Tom Hardwick · June 11th, 2009, 01:23 AM

Quote:

Originally Posted by Jim Newberry

If anyone doubts that wide angle zooms have barrel distortion, try shooting an interior at the widest focal length

An even simpler test is to zoom to full wide, then frame up your TV screen so that it fills your viewfinder. Jim's right - most zooms are pretty poor in this respect. And yes, it's the perspective distortion I want from a WIDE wide-angle.

tom.

June 9th, 2009, 10:36 PM	#1
Brian Luce Inner Circle Join Date: Jan 2006 Posts: 2,290	Barrel distortion I tested a 5d today and the 17-40mm canon lens got a huge ripple of barrel distortion during panning at all focal lengths. 17 to 40. Is this because of the large sensor? How much do you have to spend to get a lens that's usable? This was an L series and cost about $900.

June 10th, 2009, 02:19 PM	#13
Evan Donn Major Player Join Date: Feb 2006 Location: Philadelphia Posts: 795	just tried out the 14mm f2.8 L II this weekend (~$2k) and while it has almost no visible barrel distortion it does have quite a bit of perspective distortion which is especially apparent when panning. I think that's simply the nature of extreme wide angle lenses, and usually you use them specifically for that ability to exaggerate perspective. __________________ My latest short documentary: "Four Pauls: Bring the Hat Back!"

June 10th, 2009, 03:25 AM	#3
Marcus Marchesseault Trustee Join Date: Nov 2005 Location: Honolulu, HI Posts: 1,961	Are you sure you are describing significant barrel distortion? All the expensive short lenses I have seen do a good job of keeping the lines fairly straight. Are you sure you aren't just seeing the extreme forced perspective of a short lens on a full-frame sensor? Anything but a "normal" lens is going to have effects on the perspective in the scene. This may not be so obvious on a still shot with a wide lens, but panning is where it will be really apparent. Since you say it is during a pan, I don't think you are talking about barrel distortion but rather the forced perspective of a short (wide) lens. If you don't want that effect, you must use something very close to 50mm. If you go to a longer lens, you will experience a flattening of the perspective where the relation of the distance between objects becomes less apparent. In a short lens, the distance between objects in the scene is exaggerated and distorted compared to what is perceived by the human eye. Some people really like this effect.

June 10th, 2009, 04:15 AM	#4
Richard Hunter Trustee Join Date: Dec 2004 Location: Lipa City Batangas, Philippines Posts: 1,110	If it is only apparent during panning could it be rolling shutter effects? Richard

June 10th, 2009, 08:25 AM	#6
Don Miller Major Player Join Date: Mar 2006 Location: Chicago Posts: 706	No reason to guess when you can post an example. I don't have the 17-40, but plenty of video has been shot with it. The lens is considered professional and a good value. Barrel distortion is more of a problem with still than video.

June 10th, 2009, 08:50 AM	#8
Olof Ekbergh Sponsor: Westside AV Join Date: Dec 2008 Location: Mount Washington Valley, NH, USA Posts: 1,365	I have shot quite a bit with that lens. It is very wide, so naturally it has some distortion. Below is video that has quite a few shots using the 17-40 L. Most of the shots in the roundhouse are with that lens. Some of them full 17mm frame. Check the tilt up on the engine with steam leaking out. Some of the outdoor shots are with the 24-105 F4 also with the 70-200 F2.8. CSRR 5 minutes of Winter Steam Most of the footage is from EX3's.

June 10th, 2009, 10:23 AM	#10
Tom Hardwick Inner Circle Join Date: Mar 2003 Location: Billericay, England UK Posts: 4,711	Just to make it quite plain that the sensor size has no bearing on a lens's barrel distortion. I'm quite surprised to hear you say you have barrel distortion on Canon's 17 mm 'while panning'. Barrel distortion is a bending outwards of all straight lines that don't pass through the centre of the frame, so panning also has nothing to do with barrel distortion. Canon's L series are usually pretty well corrected, and a 20 mm Canon prime I had zero barrel distortion. At the price I'd expect the zoom to be the same, albeit with a more restricted maximum aperture. I'm pretty sure you're seeing the effects of CMOS's rolling shutter, Brian. tom.

June 10th, 2009, 10:36 AM	#11
Olof Ekbergh Sponsor: Westside AV Join Date: Dec 2008 Location: Mount Washington Valley, NH, USA Posts: 1,365	Brian, To answer your question. "Is any of this footage with the 5d? Nice images btw." All the footage shot with Canon lenses are with the 5DmkII. On the EX3 I use the stock lens. The opening shot is 5D, the next two as well, shot before sunrise. The shots of the steam engine sitting emitting steam, and the door opening, some of the shots in the cab are with the 5D and 17-40. Later in the clips there is an interesting comparison EX3 to 5D, the two shots of the train going through the Iron bridge was shot at the same time, the one from the right side of tracks was 5D using 24-105 F4, the one from the left side lower down was the EX3. I find the 5D footage needs some softening, I do that in Color. These clips are just sort of a teaser of the final production we are currently working on.

June 10th, 2009, 10:56 AM	#12
Bill Binder Major Player Join Date: Jun 2005 Location: San Francisco Posts: 542	As someone who owns both FF and 1.6x cropped bodies, as well as the 24-70mm and 16-35mm Canon lenses, maybe I'm missing something here, but FF does play a role here doesn't it? True, the lens distortion is what it is no matter FF or crop, but with a crop you are way more in the center of the lens image and with FF you go way farther out to the sides of the lens image. So, with the crop, you're likely to see less "distortion" than with the FF, but of course it won't be as wide as FF either. My point is, it would be pretty easy for someone to use a particular wide lens on a crop and think it doesn't have much distortion, but when someone uses the exact same lens on a FF they would. No? Also, I kind of doubt this is rolling shutter related given the description. Rolling shutter during a pan would create diagonals of verticals, not barrels curving on all boundaries. No? Finally, I'm thinking using the term "barrel" here might be misleading, although I'll let the OP opine on this. I say this because both my 24-70 and 16-35II do a pretty damn good job of keeping barreling to a minimum even at full wide -- meaning lines stay mostly straight. HOWEVER, that doesn't mean there isn't significant distortion at the sides say at full 16mm wide -- meaning a face could look mighty stretched if off in a corner as opposed to the middle of the frame. Also, although this type of distortion can be obvious in a still, and can be mitigated somewhat depending on composition, all bets are off when seen during a panning video because then objects are being panned into and out of the major distortion zones, and it becomes hugely obvious -- to me at least. This is pretty normal from my take on things, but what do I know? You could always try the new wide Tilt-Shift if you have the coinage, although not sure how that would pan either, heh...