50mm Lens Comparison Wide-Open & Does Canon Boost ISO Sneakily?

[Fleetie]

I did these tests this evening. I was as careful as possible to keep room lighting the same for each shot.

See the thread about "50mm lenses that don't suck wide-open", which made me get off my butt and do these tests.

Note that the aim for these tests was only to compare lens brightnesses (T-stop) and bokeh, with all lenses wide-open.
The scene is too dark to determine relative sharpnesses.

I tested the Canon 50mm f/1.4 wide-open, both connected and NOT CONNECTED electronically, by rotating lens in body while holding the lens-release button down. Mechanically it is still completely mounted, but the camera cannot communicate with the lens, so does not know what it is, so it cannot "decide" to sneakily increase the ISO.

Canon 5D Mark 3
Manual
t=1/15s, ISO400 (nominal)

Focus was on the same point on the lantern, using live view, magnified x10.


RESULTS:

Dude was right. The "CONNECTED" Canon 50mm f/1.4 image is noticeably brighter, and I took the picture both ways, several times, and the results were completely repeatable. So it seems that the 5D3 DOES sneakily SEEM TO boost ISO without telling you, if it detects that the 50mm f/1.4 lens is connected. When I review the images on the camera, they all say ISO400.

Dude was also right that the Olympus 50mm f/1.2 is significantly brighter than the Canon 50mm f/1.4 wide-open, EVEN brighter than the "CONNECTED" image. That makes me feel slightly better about having dropped just shy of £400 on this (perfect, mint, unmarked) example of this lens, back in Jan. 2010!


Here's the album on FB:

https://www.facebook.com/media/set/?set=a.4505135767033.1073741834.1849695638&type=1&l=7e5db91bd2

Here are the files on Dropbox, including raw files:

https://www.dropbox.com/sh/rhtr3k0ru902sbr/yMKl0KWANf?lst

The tests I did were:

Zuiko (Olympus) 50mm f/1.8 - Dimmest, as expected

Zuiko 50mm f/1.4 Silvernose

Zuiko 50mm f/1.4 Non-Silvernose (later model) VERY SLIGHTLY brighter than Silvernose

Zuiko 55mm f/1.2 - Has the "biggest" bokeh / OOF blur. NOTE! This is a 55mm lens, not a 50mm one.

Zuiko 50mm f/1.2 This is the brightest of them all

Canon 50mm f/1.4 NOT CONNECTED, because lens was rotated to disengage electronic contacts

Canon 50mm f/1.4 CONNECTED - Should be same, but IS BRIGHTER

 

[Drizzt321]

Re: 50mm Lens Comparison Wide Open & Does Canon Boost ISO Sneakily?

Do you have simulate exposure or whatever it's called turned on for LiveView? When it communicates with the lens it may get more information and so might also apply the peripheral correction information (which generally will affected the outer portions of the frame, not the overall frame).

As for Olympus f/1.2, it should be brighter than a f/1.4. You're talking a full stop there, no duh it'll be brighter.

 

[Fleetie]

Re: 50mm Lens Comparison Wide Open & Does Canon Boost ISO Sneakily?

Do you have simulate exposure or whatever it's called turned on for LiveView? When it communicates with the lens it may get more information and so might also apply the peripheral correction information (which generally will affected the outer portions of the frame, not the overall frame).

As for Olympus f/1.2, it should be brighter than a f/1.4. You're talking a full stop there, no duh it'll be brighter.

No, I never use "simulate exposure".

I know it should be brighter but last time I tested it, it came out looking slightly dimmer than the Canon 50/1.4 lens. but no, the results here are clear.

BTW, how on Earth do you get that f/1.2 is a FULL STOP wider than f/1.4?! No, it's not!

 

[Drizzt321]

Re: 50mm Lens Comparison Wide Open & Does Canon Boost ISO Sneakily?

Do you have simulate exposure or whatever it's called turned on for LiveView? When it communicates with the lens it may get more information and so might also apply the peripheral correction information (which generally will affected the outer portions of the frame, not the overall frame).

As for Olympus f/1.2, it should be brighter than a f/1.4. You're talking a full stop there, no duh it'll be brighter.

No, I never use "simulate exposure".

I know it should be brighter but last time I tested it, it came out looking slightly dimmer than the Canon 50/1.4 lens. but no, the results here are clear.

BTW, how on Earth do you get that f/1.2 is a FULL STOP wider than f/1.4?! No, it's not!

Maybe send Roger over at LensRentals.com an email and see if he can find something. He loves figuring out mysteries like this.

Hmm...right, it's a 1/2 stop, sorry. I know it's a log scale so I went right to a full stop instead of really thinking. Anyway, just because it's a 1/2 stop doesn't say anything about how transmissive the different lenses are. See T-stops. The Olympus might have a better efficiency than than Canon (or maybe vice-versa), which would account for more actual light than can be accounted for the 1/2 stop difference.

 

[Fleetie]

Yes, I know about T-stops. That's partly what prompted this test. See the "50mm lenses that don't suck wide open" thread.

I'd been meaning to do a more scientific test of all my 50mm lenses like this for a while now, so tonight I finally got a round tuit.

Last time I did a similar but less rigorous test like this, the Canon f/1.4 seemed more transmissive, i.e. brighter, than the Zuiko 50/1.2.

But no, Zuiko 50/1.2 is much brighter, as you rightly say, one would expect.

 

[noisejammer]

@Fleetie - thanks for doing the test.... I'll take a bow

@Drizzt321 - wide open an f/1.2 lens collects (1.4/1.2)^2 = 1.39 times as much light as a f/1.4 lens. This corresponds to 0.47 stops (not 1 stop.)

Of course, you would expect the f/1.2 lens to be brighter than a f/1.4 lens but this is NOT what Fleetie observed. When the camera communicates with the lens, it boosts its internal ISO (without telling you) so that the f/1.4 image becomes brighter than the f/1.2.

(IMO) This is a swindle that's perpetrated to persuade people to continue buying fast glass.

 

[Fleetie]

Of course, you would expect the f/1.2 lens to be brighter than a f/1.4 lens but this is NOT what Fleetie observed. When the camera communicates with the lens, it boosts its internal ISO (without telling you) so that the f/1.4 image becomes brighter than the f/1.2.

No. That's not exactly what I am saying:

YES, the "connected" Canon image is definitely brighter.

But even the connected image is less bright than the Zuiko 50/1.2 image. That was not what I found last time, but this time I was more careful to keep lighting constant than last time.

The Zuiko 50/1.2 image is the brightest one I got (IMO).

 

[Fleetie]

One day I might do another, similar test of the same lenses, but this time to compare their sharpness wide-open.

But don't hold your breaths; round tuits are scarce here.

 

[comsense]

I did these tests this evening. I was as careful as possible to keep room lighting the same for each shot.

See the thread about "50mm lenses that don't suck wide-open", which made me get off my butt and do these tests.

Note that the aim for these tests was only to compare lens brightnesses (T-stop) and bokeh, with all lenses wide-open.
The scene is too dark to determine relative sharpnesses.

I tested the Canon 50mm f/1.4 wide-open, both connected and NOT CONNECTED electronically, by rotating lens in body while holding the lens-release button down. Mechanically it is still completely mounted, but the camera cannot communicate with the lens, so does not know what it is, so it cannot "decide" to sneakily increase the ISO.

Canon 5D Mark 3
Manual
t=1/15s, ISO400 (nominal)

Focus was on the same point on the lantern, using live view, magnified x10.


RESULTS:

Dude was right. The "CONNECTED" Canon 50mm f/1.4 image is noticeably brighter, and I took the picture both ways, several times, and the results were completely repeatable. So it seems that the 5D3 DOES sneakily SEEM TO boost ISO without telling you, if it detects that the 50mm f/1.4 lens is connected. When I review the images on the camera, they all say ISO400.

Dude was also right that the Olympus 50mm f/1.2 is significantly brighter than the Canon 50mm f/1.4 wide-open, EVEN brighter than the "CONNECTED" image. That makes me feel slightly better about having dropped just shy of £400 on this (perfect, mint, unmarked) example of this lens, back in Jan. 2010!


Here's the album on FB:

https://www.facebook.com/media/set/?set=a.4505135767033.1073741834.1849695638&type=1&l=7e5db91bd2

Here are the files on Dropbox, including raw files:

https://www.dropbox.com/sh/rhtr3k0ru902sbr/yMKl0KWANf?lst

The tests I did were:

Zuiko (Olympus) 50mm f/1.8 - Dimmest, as expected

Zuiko 50mm f/1.4 Silvernose

Zuiko 50mm f/1.4 Non-Silvernose (later model) VERY SLIGHTLY brighter than Silvernose

Zuiko 55mm f/1.2 - Has the "biggest" bokeh / OOF blur. NOTE! This is a 55mm lens, not a 50mm one.

Zuiko 50mm f/1.2 This is the brightest of them all

Canon 50mm f/1.4 NOT CONNECTED, because lens was rotated to disengage electronic contacts

Canon 50mm f/1.4 CONNECTED - Should be same, but IS BRIGHTER

Curb your enthusiasm!!!!
You need to care of these things:
1) Put the metering off and shoot in M mode
2) Put the tape (super thin one) over contacts and click it all the way. Rotating changes lens light entry and image plane to sensor angles and could effect light reaching sensor and hence brightness (not a perfect control - you are comparing apple with oranges)
3) Capture 5-10 photos with each condition and take an average

 

[Fleetie]

I don't need to do anything of the sort.

I took several shots in both conditions, and the result was unambiguous.

Rotation does not change the angle of lens axis to plane of sensor. The lens is still mechanically mounted hard up against the flange, so no angular movement is possible. Clearly you have not tried this, or it would be obvious to you that mechanically, the mount is identical to the connected case.

You might "need" to. So please do.

In fact, let's have more people do this.

 

[Loswr]

The 'sneaky' ISO boost with fast primes it done to compensate for loss of light transmission through the microlenses over the sensor pixels that occurs at high incident angles of light. As such, it's specific to digital (vs. film), and more boost is needed for smaller pixels. The issue has been documented (and quantified) by DxOMark. It's not just Canon, by the way - Nikon and Sony do it, too.

http://www.dxomark.com/index.php/Publications/DxOMark-Insights/F-stop-blues

 

[Fleetie]

The 'sneaky' ISO boost with fast primes it done to compensate for loss of light transmission through the microlenses over the sensor pixels that occurs at high incident angles of light. As such, it's specific to digital (vs. film), and more boost is needed for smaller pixels. The issue has been documented (and quantified) by DxOMark. It's not just Canon, by the way - Nikon and Sony do it, too.

http://www.dxomark.com/index.php/Publications/DxOMark-Insights/F-stop-blues

Thanks, Neuro.

I do remember years ago reading that the Leica M9 has special angled microlenses to mitigate this effect; I think it was the same effect.

It still does seem a little "sneaky" of the camera to not show the modified ISO used.

I'm not going to lose sleep over it; it's something interesting I learned today. I still like my 5D3!

 

[Mt Spokane Photography]

The 50mm 1.4 is a old lens (1993) designed for film, so the effects of light drop off at the edge of a digital sensor are not compensated for. Newer lenses do a better job of collimating the light so it comes in at a steeper angle, and the effect is less.

http://www.nikonians.org/html/resources/nikon_articles/body/FF_vs_DX_sized_sensors/

After I thinking about it, there would be no fixed distance to the sensor focal plane for collimated light, so this seems to be incorrect.

 

[Etienne]

The 50mm 1.4 is a old lens (1993) designed for film, so the effects of light drop off at the edge of a digital sensor are not compensated for. Newer lenses do a better job of collimating the light so it comes in at a steeper angle, and the effect is less.

http://www.nikonians.org/html/resources/nikon_articles/body/FF_vs_DX_sized_sensors/

I don't think it's possible to collimate the light and still end up with a sharp image. There's urban-legend BS at work in that link.

 

[WillThompson]

Re: 50mm Lens Comparison Wide Open & Does Canon Boost ISO Sneakily?

Hmm...right, it's a 1/2 stop, sorry.

No it is 1/3 stop!

The 1/3 stop steps are: 1.0 1.1 1.2 1.4 1.6 1.8 2.0 .........

 

[Loswr]

Re: 50mm Lens Comparison Wide Open & Does Canon Boost ISO Sneakily?

Hmm...right, it's a 1/2 stop, sorry.

No it is 1/3 stop!

The 1/3 stop steps are: 1.0 1.1 1.2 1.4 1.6 1.8 2.0 .........

No, it's both, or if you prefer, neither. On the half-stop scale, f/1.2 is 1/2-stop wider than f/1.4, but on the 1/3 stop scale, f/1.2 is 1/3-stop wider than f/1.4. See the wikipedia page on f/stops.

Mathematically, f/1.2 really is closest to a 1/2-stop, since 21/2x0.5 = 1.1892), whereas 22/3x0.5 = 1.2599, which personally I'd round to f/1.3.

 

[CharlieB]

Very good point about the sensor acceptance angles. I'm also wondering if periferal illumination correction also might have had an effect in this case.

 

[Fleetie]

Very good point about the sensor acceptance angles. I'm also wondering if periferal illumination correction also might have had an effect in this case.

No, because I always leave it switched off. And I just turned on the camera to double-check, and it was, and is, off.

The jpgs are SOOC, too.

 

[Mt Spokane Photography]

The 50mm 1.4 is a old lens (1993) designed for film, so the effects of light drop off at the edge of a digital sensor are not compensated for. Newer lenses do a better job of collimating the light so it comes in at a steeper angle, and the effect is less.

http://www.nikonians.org/html/resources/nikon_articles/body/FF_vs_DX_sized_sensors/

I don't think it's possible to collimate the light and still end up with a sharp image. There's urban-legend BS at work in that link.

I think you are right, now that I think about it, its nonsense, there would be no fixed distance to the sensor focal plane for collimated light.

I've revised my post.

 

[rs]

As higher pixel density sensors and faster lenses seem like a bad combination according to this, I wonder if it makes small sensor high MP bodies such as the OM-D E-M5 with lenses such as the SLR Magic 50/0.95 hyperprime not all as advertised?

With the pixel density of a 16MP m4/3 body (pixel pitch equal to a 64MP FF body) it is probably a long way off the f0.95 light capturing levels it claims. Could this effect make it nearer to f2.8?