Home
SearchSearch MemberlistMemberlist RegisterRegister ProfileProfile Log in to check your private messagesLog in to check your private messages Log inLog in

Small test: 2x teleconverter vs cropping (FD 300 f/4 L)
View previous topic :: View next topic  


PostPosted: Sat Feb 10, 2018 11:34 pm    Post subject: Small test: 2x teleconverter vs cropping (FD 300 f/4 L) Reply with quote

I wanted to see how much sense a 2x teleconverter would make with a (very good) old(er) lens on a modern 24Mp sensor, in real life, I think the answer is pretty clear!

I used the Canon nFD 300mm 1:4 L lens with the 2x-B Extender (which is meant for lenses <200mm) as I don't have the 2x-A yet.

I was lucky enough to get to test this on a subject slightly more interesting than a brick wall because this lioness seemed quite interested in (her reflection in?) my lens and was in exactly the right spot so I could rest the lens as well as the camera (I'm no big tripod user) so I could get away with slow shutter speeds.
This meant respectively 300mm f/4 (the f/5.6 image was focused slightly behind her head, this one might be slightly as well, but I don't think it affects the resolution, maybe the microcontrast a bit!), 1/80s, iso 100 vs 600mm f/9.5 (? Lens stopped down half a stop), 1/60, iso 400.

This is of course not a scientific test at all, I used the best comparable pictures, the bare lens would have had better contrast if it was stopped down half a stop as well, with less CA, but all in all I was quite surprised by how well this lens did on the wrong extender!

Outside the image center this combination gives some LaCA and the (micro) contrast suffers a bit, but I'm not sure if the 2x-A extender would be much better, need to find one, and the 1.4x-A as well!

The difference would probably be smaller on higher MP body's

I used FastStone Image Viewer to compare and the images were all SOOC jpg's from my Sony A7 with all image settings neutral except for Auto DRO.

Edit: Left = bare lens - Right = Lens + TC

1: image with teleconverter downscaled to 50%:

Canon_FD_300_f4_L-2x-B_Tele_extender_vs_cropping_2 by Koen Nieuwenhuize, on Flickr

2: both at 100%:

Canon_FD_300_f4_L-2x-B_Tele_extender_vs_cropping_3 by Koen Nieuwenhuize, on Flickr

3: image without teleconverter upscaled to 200%:

Canon_FD_300_f4_L-2x-B_Tele_extender_vs_cropping_1 by Koen Nieuwenhuize, on Flickr

Is my reasoning correct that it seems this lens is capable of delivering enough resolution in the center to perform rather well on a +-100Mp sensor (2²x24Mp)?


Last edited by Koen Nhz on Sun Feb 11, 2018 1:01 pm; edited 1 time in total


PostPosted: Sun Feb 11, 2018 6:38 am    Post subject: Reply with quote

Well, I found your test confusing at first because you do not state which is the image without TC and which is the one with. But when I saw the "both at 100%" set, I finally got it figured out. Good thing you included file names too or else I'd have been lost going from one set of exposures to the next.

Anyway, my take is this: while your camera's images hold together very well when viewed at 200% (the last set of images), the one on the right, the one taken with the TC, is the clear winner. I'm not surprised by this.


PostPosted: Sun Feb 11, 2018 10:43 am    Post subject: Reply with quote

A7II is not a demanding camera as it is only 24MP. Adding a good TC will improve a lot if there lens has more resolution the sensor can resolve. Your result proves that.


PostPosted: Sun Feb 11, 2018 12:50 pm    Post subject: Reply with quote

cooltouch wrote:
Well, I found your test confusing at first because you do not state which is the image without TC and which is the one with. But when I saw the "both at 100%" set, I finally got it figured out. Good thing you included file names too or else I'd have been lost going from one set of exposures to the next.

Anyway, my take is this: while your camera's images hold together very well when viewed at 200% (the last set of images), the one on the right, the one taken with the TC, is the clear winner. I'm not surprised by this.


Yes, you are right, I forgot to clearly state which is which, I'm sorry, but indeed, left: bare lens, right: lens + extender.
I wasn't really surprised that the TC image has more detail, but I didn't expect those images to be almost as sharp at pixel level (image 2).
I'm not sure if you ment is this way, but in the last set only the left image is at 200%, the one on the right is just 100% and much better imo.

calvin83 wrote:
A7II is not a demanding camera as it is only 24MP. Adding a good TC will improve a lot if there lens has more resolution the sensor can resolve. Your result proves that.


You are right (A7 I, but +-same sensor), but this one I did state quite clearly I think (my quote: "The difference would probably be smaller on higher MP body's ")

It's the amount of difference between the 2 that really surprised me, imo it doesn't look like the TC image is much less sharp at pixel level, so that would mean the bare lens would probably do just fine on future 100Mp body's. I remember the times when the Canon 5D just came out and people tought they would need new (EF) glass because of this 'gigantic' 12Mp sensor Laughing


One other thing I forgot to mention is the color difference, that's my fault for using AWB in changing light conditions (sun was quite low already), I don't think it has anything to do with the TC, but will check this later.


PostPosted: Sun Feb 11, 2018 4:16 pm    Post subject: Reply with quote

Koen Nhz wrote:
cooltouch wrote:
Well, I found your test confusing at first because you do not state which is the image without TC and which is the one with. But when I saw the "both at 100%" set, I finally got it figured out. Good thing you included file names too or else I'd have been lost going from one set of exposures to the next.

Anyway, my take is this: while your camera's images hold together very well when viewed at 200% (the last set of images), the one on the right, the one taken with the TC, is the clear winner. I'm not surprised by this.


Yes, you are right, I forgot to clearly state which is which, I'm sorry, but indeed, left: bare lens, right: lens + extender.
I wasn't really surprised that the TC image has more detail, but I didn't expect those images to be almost as sharp at pixel level (image 2).
I'm not sure if you ment is this way, but in the last set only the left image is at 200%, the one on the right is just 100% and much better imo.


Yes, that's the way I saw things. And that's why I stated that I thought the camera's image at 200% held up very well. Typically, at 200% I expect to see at least some pixilation and jagged (ie, stair-step) diagonals, which I don't really see with the Left image. But even with a 100mp sensor, if evaluating an image at 200%, I would still expect to see some loss of detail. 200% is 200%, regardless of the sensor.

As for the right image, the one taken with the TC, I'm not surprised that it held together well. I have tested several good-quality TCs and have found in my tests that they do not result in loss of detail, in fact just the opposite. A good-quality TC will slightly add to the detail of an image because it improves a lens's resolving power.


Last edited by cooltouch on Sun Feb 11, 2018 4:20 pm; edited 1 time in total


PostPosted: Sun Feb 11, 2018 4:19 pm    Post subject: Reply with quote

Koen, is there a reason why the image on the right shows the lioness with less long whiskers?


PostPosted: Sun Feb 11, 2018 5:38 pm    Post subject: Reply with quote

cooltouch wrote:


Yes, that's the way I saw things. And that's why I stated that I thought the camera's image at 200% held up very well. Typically, at 200% I expect to see at least some pixilation and jagged (ie, stair-step) diagonals, which I don't really see with the Left image.



I think it depends mostly on the program used to show the images (with the condition that the original is sharp enough at 100% of course) and the method of resize, if I checkt the "verfijn" (refine) checkbox it would have looked even smoother (you can choose between Lanczos and Bilinear), but a bit less sharp.


cooltouch wrote:


But even with a 100mp sensor, if evaluating an image at 200%, I would still expect to see some loss of detail. 200% is 200%, regardless of the sensor.



Yes, you are right, but I didn't mean exactly that when I wrote about the 100Mp, if I had a 100Mp sensor I could have compared it (with the bare lens) at 100% (instead of 200%) to the 24Mp sensor with the lens on the TC (I think, see next)


cooltouch wrote:


As for the right image, the one taken with the TC, I'm not surprised that it held together well. I have tested several good-quality TCs and have found in my tests that they do not result in loss of detail, in fact just the opposite. A good-quality TC will slightly add to the detail of an image because it improves a lens's resolving power.


I am not sure, but I think we still don't mean the exact same thing, but it's a bit hard to explain, I mean that the image with TC looks almost as sharp as the one without, both viewed at 100%, not that it resolves more detail from the scene (which I totally expected and it should).
So basically what I mean is that the image with TC resolves almost almost twice as much detail, the theoretically maximum. So if I had stood back twice as far for the TC shot it would still have resolved (almost) as much detail as the bare lens has from the closer distance, if that makes more sense?

That's why I wrote about the 100Mp, that is what the sensor would need to be to get the same 'crop' with the same amount of pixels as the TC gives me, but that's only correct if the images are indeed as sharp and for that I would need to use a testchart instead of a real life image.

Sciolist wrote:
Koen, is there a reason why the image on the right shows the lioness with less long whiskers?


Yes, I thought they looked a bit untidy so I trimmed them down a little bit between the images! Twisted Evil
No, I think what you are seeing is that she held her whiskers a bit closer to her head in the second shot, they can move them around a bit, especially move them more forward if they are investigating something or are curieus.


PostPosted: Sun Feb 11, 2018 7:48 pm    Post subject: Reply with quote

Koen Nhz wrote:

Sciolist wrote:
Koen, is there a reason why the image on the right shows the lioness with less long whiskers?


Yes, I thought they looked a bit untidy so I trimmed them down a little bit between the images! Twisted Evil
No, I think what you are seeing is that she held her whiskers a bit closer to her head in the second shot, they can move them around a bit, especially move them more forward if they are investigating something or are curieus.


Ah, I get you. The 'missing' whiskers are pointing away from me in the 2nd shot, giving the appearance of being shorter.

Thanks.

"So basically what I mean is that the image with TC resolves almost twice as much detail, the theoretically maximum. ..."

It's interesting you say that. Playing devils advocate for you, the only thing that came to mind that you could eliminate for the difference is whether you missed critical focus on the non-TC lens and the head of the lioness was not placed on the focus plane. Obviously, if you missed focus by a hair it could lose detail, but still look in focus.

Does that make sense? In the hypothetical scenario I describe, the focus point ended up just a smidge in front of the lionesses nose.


PostPosted: Sun Feb 11, 2018 9:44 pm    Post subject: Reply with quote

You are correct that in the image without TC the focus isn't 100% accurate (I stated this somewhere in the opening post as well), but it's not in front of the lioness's head, the best focus seems to be slightly behind her eyes, on the back of her head, you can see this in the hairs there, they have a bit more microcontrast than her nose, so it was indeed at a disadvantage because in the TC image the best focus seems to be exactly on her eyes.

I checked a dof simulator and if the subject was 10m away from me (don't know exactly), the dof is about 25cm, so the detail on the nose might be affected slightly, but I don't think it makes a real difference in amount of detail captured. I compared it to perfectly focused images where she wasn't looking into the lens and the difference seems really small.

Also movement blur can't be ruled out completely because of the slow shutter speeds.

One of the reasons for me for doing this test was the fact that I somehow doubted if this lens could resolve as much as my camera, it usually looks really sharp, but sometimes it looks as if not all fine detail is captured, especially low contrast detail, something I also noticed in the past with my EF 70-200 f4 IS L lens (which I never grew fond of), but I think this might have something to do with the very high 'edge' contrast which seems to hide some of the finer lower contrast detail in some subjects.
Perhaps that's even one of the attractions I (and perhaps many of us) have with these older lenses, that they give a less 'digital' look because of the gentler contrast with less emphasis on edges.

This is probably all irrelevant when shooting RAW anyway..

This FD 300 f4 L lens at least has much nicer bokeh than the 70-200 f/4 IS, and the rendering is gentler wide open than stopped down a bit. And for me this is almost purely a nature/zoo/... lens, so being sharp is almost always an advantage in contrast to a 70-200 that's more versatile and better suited for portret (where I didn't like it very much at all).

The fact that the 300 L looks so good on a TC proves to me that it's resolving power certainly isn't the problem, so I'm happy with that. Smile