Lens baffle equivalent of stopping down?

[gtravis]

First of all, I apologize for not simply attaching this to one of the many threads about lens baffles that already exist -- I simply couldn't decide which one was most appropriate.

I use a Minolta 50mm f1.4 MD lens on my G1 (with an adaptor). One of the disadvantages of that lens is extreme softness/ghosting when wide open -- in fact, things don't get sharp until the lens is stopped down to about f4 or smaller -- which makes the lens' utility as a fast prime portrait lens kinda, well, non-useful.

This weekend I played around with making a "baffle" for the lens, which I fitted inside the adaptor instead of to the back of the lens itself. Theoretically, as I understand it, a baffle should restore the lost sharpness by reducing stray light while not affecting the lens' speed (much).

I simply didn't find this to be true. I found that the baffle did indeed increase sharpness, but it also decreased the lens' speed (wide-open) by an amount similar to, if not identical to, simply closing the lens aperture (i.e. stopping down).

So, is baffling no good? Or did I simply do it wrong. I experimented with a number of different size baffle "holes" and saw a correlation with sharpness increase that pretty much tracked the baffle hole size (smaller=sharper) but also saw a correlation with speed decrease that also pretty much tracked the baffle hole size.

Here are a couple of picture. One, looking from the rear of the lens adaptor and the other showing the rear of the lens itself and looking into the adaptor from the front:

 

[Zaraz]

Yes, this whole lens baffle concept is the exact same as stopping down. I'm not sure if the original poster of that French article was seriously thinking he invented something new or just pulling people's leg...

The stray light theory sounds plausible, but unless your adapter is silvered on the inside the effect of stray light is going to be negligible. If it were not, the more sound approach would be to cover the inner walls with thin felt or something else similar to flocking.

--
http://www.flickr.com/photos/thustra/

 

[DjarumBlack]

A lens baffle is not necessarily stopping down if the size of the baffle is NOT larger than the light cone before the sensor. Telescopes have baffles in them, and they do not stop the lens down. They are there to improve the contrast on images by blocking out stray light.

 

[Zaraz]

A lens baffle is not necessarily stopping down if the size of the baffle is NOT larger than the light cone before the sensor.

Agreed. But if the optical improvements correlate with a reduction in light intensity (longer shutter speeds for equal exposures) then I think it is safe to say that the effect is not that of a baffle but of an iris - and most lenses already have one which is easier and more flexible to use than a cardboard cut-out.

Telescopes have baffles in them, and they do not stop the lens down. They are there to improve the contrast on images by blocking out stray light.

At the same time, for telescope baffles the optical design and location in the optical path is quite different to what is proposed here. Telescope baffles are similar in effect to a flocking of the hood and interior walls of a photographic lens, which of course can be extended (as I proposed) to include the interior walls of any adapter.

--
http://www.flickr.com/photos/thustra/

 

[Bryan Biggers]

In theory, there might be scattered light. The old lenses are designed for 35mm film of course, and the MFT sensor has only about 1/4 the size of the 35mm imaging area. So, in theory, 75% of the light misses the MFT sensor, and has to go somewhere. On the E-P1, the sensor box is black, and the sensor is recessed, so it seems that not much of this light could be scattered onto the sensor, but it could in theory bounce around in there and reduce contrast, however I doubt the claims about 1 stop of light being scattered, you would have to have 50% of the light that misses the sensor ending up back on the sensor.

People do spend a lot of time placing baffles and "flocking" Their telescope tubes to adsorb stray light and improve contrast, and this is much discussed and debated. There is almost certainly some small amount of contrast to be gained in those systems, and in some cases you can see and measure it.

In this case, I've tried making some baffles of the theoretical size, and some smaller and larger just for fun, and have tried vs. placements. Any baffle that obscures any part of the edge of the lens as seen from the sensor is just like stopping down, light from the entire lens forms parts of the image even with the crop factor on this sensor. When I make a baffle for my 28mm lens (widest old lens I have) that is the theoretical size to just cut off the edges of the cone of light that misses the sensor, and have it at the optimal placement in my light path, I don't see any exposure reduction in my informal tests, but I don't notice any image improvement either. When I reduce the size, there is an exposure reduction, and an image improvement, but it seems to be the same as if I just stopped down the lens. I wonder if people are just being fooled by accidentally making a baffle that stops down the lens just a bit; it would not take much. If you see an appreciable exposure reduction that is what you are doing unless you have some crazy reflections of an appreciable amount of light in your system.

I do what this to work, and it seems like it should work, so I'm still messing with it. Maybe the lens that I'm using does not benefit, it might be different for others.

 

[Zaraz]

Two more things to keep in mind regarding scattered light: There are quite a few people that use medium format lenses on their APS-C or FF DSLR. In theory, the same scattered light problem would apply to them because of the significantly larger image circle of a medium format lens. In practice, it's not an issue.

Further, if a baffle of sorts would actually work, it should work equally for all apertures, and not just for the lens wide open. The relative amount of light scatter remains the same at all apertures as the image circle projected by the lens remains fairly constant.

In practice, as much as I would like it to be otherwise, I believe that the "glow" or softness that many of the fast legacy lenses have is really the m4/3 sensor exposing their resolution limits when used wide open. Some will be better than others, but very few will actually be very sharp unless stopped down a bit.

--
http://www.flickr.com/photos/thustra/

 

[Tortellino]

Agreed. But if the optical improvements correlate with a reduction in light intensity (longer shutter speeds for equal exposures) then I think it is safe to say that the effect is not that of a baffle but of an iris - and most lenses already have one which is easier and more flexible to use than a cardboard cut-out.

I am not sure that the longer exposure time is a good indicator. If the baffle blocks the stray light from hitting the sensor then of course you will have less light - isn't it the point? But it's a "bad", defocused light, so you don't want it anyway.

I think a better indicator would be the DOF. Take pictures of a ruler with different baffles, and see if it increases or stays the same as without baffle.

 

[Bryan Biggers]

I put up a comparison sequence, lens with and without baffle. Click through to the original shots if you like (summary: no significant difference). Warning, the original shots are several megs.

http://dcexp.com/gallery/LensBaffle

 

[amalric]

I think your baffle is too close to the sensor, and too small, while it should be close to the exit lens and bigger.

I made one for a Hexanon 50/1.8 and it helped reducing stray light at max aperture, on 4/3.

For the diameter I followed general advice. The baffle makes the difference between being able to use or not to use such a lens at max aperture. provided it is sharp in the first place.

Being larger it shouldn't affect your exposure, but still block haziness.

Am.
--
Photostream: http://www.flickr.com/photos/amalric

 

[Hausdorff]

Maybe, it's better to take test shots with much lighter object in the field of view? In your example, there are no parts of light sky near the sun or something like this.

 

[Zaraz]

I am not sure that the longer exposure time is a good indicator. If the baffle blocks the stray light from hitting the sensor then of course you will have less light - isn't it the point? But it's a "bad", defocused light, so you don't want it anyway.

The exposure time is a very good indicator, but feel free to do a DoF test instead. The result will be the same. If you were having "stray" light in the order of magnitude of 1/3 or 1/2 photographic stop, you would not be seeing much of the actual image! This kind of stray light would also be fairly evenly distributed across the sensor and hence not reduce lens resolution but contrast.

Reducing the aperture of the exit pupil of a lens is the same as stopping down. It also does little to decrease the diameter of the projected image circle. You can verify that fairly quickly by placing a "baffled" lens under a desk lamp and comparing the results with and without baffle

--
http://www.flickr.com/photos/thustra/

 

[Zaraz]

amalric wrote:

The baffle makes the difference between being able to use or not to use such a lens at max aperture. provided it is sharp in the first place.

The baffle is actually preventing you from using the lens at max aperture as it has reduced your effective aperture. Hence the improvement of IQ.

Being larger it shouldn't affect your exposure, but still block haziness.

I bet with you that if you do a controlled test with and without baffle, you will have at least 1/3rd stop difference in exposure.

 

[duartix]

This weekend I played around with making a "baffle" for the lens, which I fitted inside the adaptor instead of to the back of the lens itself. Theoretically, as I understand it, a baffle should restore the lost sharpness by reducing stray light while not affecting the lens' speed (much).

Can you also test producing two equally exposed pictures like this:

1. Picture one: baffle inserted.
2. Picture two: baffle removed and lens stopped down so that exposure is the same as Picture one.

What are the results for contrast/DoF?

--
Duarte Bruno

 

[Bryan Biggers]

Maybe, it's better to take test shots with much lighter object in the field of view? In your example, there are no parts of light sky near the sun or something like this.

Are you talking about my test shots? Maybe, the exposure was already 1/3200 sec, so it was pretty bright, but maybe a really bright object just out of the field of view would be the worst case.

 

[amalric]

amalric wrote:

The baffle makes the difference between being able to use or not to use such a lens at max aperture. provided it is sharp in the first place.

The baffle is actually preventing you from using the lens at max aperture as it has reduced your effective aperture. Hence the improvement of IQ.

That's what simpletons think.

Being larger it shouldn't affect your exposure, but still block haziness.

I bet with you that if you do a controlled test with and without baffle, you will have at least 1/3rd stop difference in exposure.

Even if so, that would be due to stray light. Loosing 1/3 of a stop for more resolution would be a fair trade off.

BTW my argument is that the OP baffle is wrongly placed. Reflections should be stopped at the source, before they become part of the image.

I might be wrong but you don't address the problem. To me a baffle is like an internal hood, it doesn't have anything to do with exposure.

BTW the image circle of a MFL is much larger than a 4/3 lens, so you don't detract at all from its light gathering capacity for 4/3, unless you put the baffle too close to the sensor.

Am.

--
Photostream: http://www.flickr.com/photos/amalric

 

[Zaraz]

amalric wrote:

The baffle makes the difference between being able to use or not to use such a lens at max aperture. provided it is sharp in the first place.

The baffle is actually preventing you from using the lens at max aperture as it has reduced your effective aperture. Hence the improvement of IQ.

That's what simpletons think.

That is a really profound argument. If your baffle makes you happy because you think it is different from stopping your lens down, I'll be the last wanting to stop you from being happy.

--
http://www.flickr.com/photos/thustra/

 

[gtravis]

Four pictures, left -> right:

1. Baffle (as pictured above) in adaptor. Lens set to f1.4, shutter set to 1/500 per camera's exposure calculation.

2. Baffle removed. Lens set to f1.4, shutter set to 1/800 per camera's exposure calculation (i.e. camera thought there was more light).

3. Baffle removed, Lens set to f2.0 to get a 1/500 shutter on the camera (i.e. it appears the baffle makes it an f2.0 lens)

4. Baffle removed, lens set to f4.0, shutter set to 1/100 per camera's exposure calculation.

Lens is a Minolta f1.4 50mm MD.

 

[amalric]

The baffle diameter must be made in proportion of what part of the 35mm FOV will be used by the 4/3 lens, not more not less.

Old MF lenses are not telecentric, but the sensor is. A matter which is apparently ignored, However unnecessary stray light must be blocked.

If you are convinced of the non necessity of a baffle don't use it. However most MF lenses have one. It is only that they are different between 35mm and 4/3, so you must replace them.

Am.

--
Photostream: http://www.flickr.com/photos/amalric

 

[Zaraz]

The baffle diameter must be made in proportion of what part of the 35mm FOV will be used by the 4/3 lens, not more not less.

Putting a baffle at the immediate rear of a photographic lens and blocking part of the light exiting the rear element is not going to reduce the image circle cast by that lens. The full "FoV" is still cast behind the baffle.

Take a lens, set it to wide open, hold it over your desk, have a desk lamp above and watch what happens with the image cast by the lens on your desk as you slide a piece of paper or cardboard in front of the rear element. Vignetting of the cast image will not even occur before you have covered at least half of the rear element! And before that, your image circle remains the same!

Old MF lenses are not telecentric, but the sensor is. A matter which is apparently ignored, However unnecessary stray light must be blocked.

The fact that virtually all digital sensors are "telecentric", meaning sensitive only to rays with a fairly low angle of incidence, in fact ensures that they are even less susceptible to "stray" light as they are much less sensitive to reflected light hitting the sensor at an angle.

The Leica M9 uses a very sophisticated design for their sensor micro lens structure to counter this effect and allow the use of wide angle lenses with the short M register distance.

As to lenses, being telecentric or not is of little relevance for a 35mm lens on a m4/3 body

If you are convinced of the non necessity of a baffle don't use it. However most MF lenses have one. It is only that they are different between 35mm and 4/3, so you must replace them.

Which of your MF lenses have an internal baffle, and where in the optical path is it located?

Out of 10+ MF lenses I have, 2 wide angle lenses have a baffle and with both it is inside the lens, directly in front of the diaphragm, and larger in diameter than the max diaphragm diameter.

--
http://www.flickr.com/photos/thustra/

 

[Antoine Palade]

I tried the baffle solution too. Two observations, one question and one suggestion:

Observation # 1: I found it impossible to place the baffle at the rear of the lens without it touching the glass element, which obviously I don't want to do. There is simply no place to stick the baffle on an elevated part of the rear of the lens, because there is no elevated part at the rear of the lens! Mind you, it might be particular to the lenses I am using - a Zuiko 50mm f1.4 and a Canon FD 55mm f1.2, I don't know.

Observation # 2: Trying to put the baffle inside the adapter didn't work for me as the baffle had to be bent to be able to be inserted within the adapter. And once bent, I can't see how it can work...

Question: This might sound stupid, but why is the middle hole of the baffle supposed to be round, when the sensor is, what, rectangular?...wouldn't it make more sense to match the shape of the middle of the baffle to that of the sensor, i.e. rectangular?

Suggestion: If the baffle principle works in theory, then what we need is an adapter with a built-in controllable iris, which would mimic the aperture of the hole in the middle of the baffle, right? Being adjustable, it would be very easy to determine the most appropriate aperture that would block stray light.