RPO's quest to build a f/0.7 lens.

[Ryan Patrick O'Hara]

Call me crazy, but I plan on experimenting to make a lens that will hopefully achieve an exposure equivalent of what a f/0.7 lens would provide or perhaps even a little under.

Let me say that this will be for fun, and if it works, which it very well might not... the lens will be very funky in regards to optical quality. I just want to see if I can do it.

Right now, the plan is to use a Nikon f/1.2 55mm lens, which is full frame. That means the projection circle is at least 11x the needed surface area of 16mm film.

What I plan to do, which has been briefly attempted before, is to take the full frame image circle and condense it to an image circle suited for a 16mm chip size. Thus we get more of the image circle falling upon a smaller sensor, so mathematically, I'm anticipating a two stop gain in light with an increase in the field of view; meaning the 55mm lens will look wider than it normally would on that particular camera.

The F/1.2 lens will still behave like a F/1.2 lens. The depth of field will be the same, however, the exposure will have increased by 4x the amount of light.

As mentioned, this will likely end up as a very 'vintage' looking lens, with aberration and sharpness issues, but I hope to make something that is fun to play with. I'd imagine getting any image at f/0.7 should be a fun time.

Depending on light loss, there is a small chance I could break the f/.07 barrier, by perhaps a sixth or a third of a stop. I'm just wanting to hit f/0.7.

Mind you I'm using F-stop to measure the amount of light, but the lens will always really be a f/1.2 and have all the DoF qualities of a 55mm f/1.2. Only my exposure will change because of the focal reducer assembly.

So my lens will technically be aiming to be a 55mm f/1.2 lens that exposes as if it's letting in light that a f/0.7 lens would.

See what I mean? It won't actually have an aperture opening of F/0.7

I'll update this thread as it comes together over the next month or so. Depending on how much I work.

I'll be starting with a 50mm f/1.4 and will buy the 55 f/1.2 as necessary.

-ryan

 

[Ryan Patrick O'Hara]

* most def not to scale... at all

So here is the first basic game plan.

The obstacles are immediately recognizable.

First, thank goodness for Nikon flange distance! At 46.5mm, it is one of the longer FFD's out there. This gives me some working room. However, if I'm not mistaken, the biggest reason this won't turn out pretty is because I believe the focal reducer should actually come before the focal point which occurs inside the lens. So I am immediately fighting an uphill battle trying to wrangle the light after it's past that point. I could be wrong, and perhaps it is bad to try to design this lens after how it's done in the telescopes, but it is a concern for sure.

Second, to utilize the space as much as possible and to test the focal reducer at different positions, I will remove the flange and any necessary back end parts of the lens. I imagine this won't be much, as I just took apart another Nikon lens of mine to see what I was getting into.

As of right now, the plan is to keep it simple and go from there. I will be suspending the 55mm lens, focal reducer, and camera (now a GH2 in 1:1 crop mode) as my test subject. The GH2 is a wonderful choice as it has one of the shallowest FFD's of the bunch thus giving me more (if there was anything to begin with) room to work. This is under the assumption the focal reducer does not change the back focus distance of the lens. I hope it does by means of making it longer. Of course this runs the risk of letting the image circle get bigger over distance and cancel out the condensing effect. Having all of these parts on a bellows system track will allow me to move them very precisely and by careful amounts to find the optimum distances between the lens, reducer and camera. I will have an assortment of lens elements and achromats on hand to experiment with and attempt to improve the results.

I originally desired a PL version of my creation, but unless everything can fit within the RedEpic PL port, and do so safely, I am afraid the deep flange depth is not helpful. Only helpful when the lens has a deep flange depth. The camera is better to have a shallow depth so I have more room to work in the open, and by open I mean the two inches and change.

Here is a good example of how this works. It was attempted on GH1's awhile back, but in my opinion, failed in many attempts by many people because of improper mounting, which I plan to explore the best I can. However, my deepest fear is that this won't work considering the reducer might be optimal in front of the point of focus which happens in the lens. I don't know if that is the case, but it seems to be how it works in telescops:


I believe the reducer typically goes right before the mirror, which is before the point of focus. Above, I can only put it right after.

If this were easy, it wouldn't be as fun, would it?

 

[Ryan Patrick O'Hara]

reserved.

 

[Matt Gottshalk]

Tagged for the epic mad-scientist glass whoreage.

 

[richg101]

nice idea my 2 pence... Maybe purchase a achromatic doublet (Tokina +0.4 closeup). these are known to improve drastically any chromatic aberration problems from less than optimal lenses. eg. my olympus 50mm f1.4 is a lot better with the achromat when wide open. little difference in brightness too.

 

[Dustin Moore]

It is really hard to make the opposite of a telecompressor (tele-decompressor?) It is also really hard to make
things that are f/0.7 and are not narrow field microscope objectives.

You might want to grab some lens design software like OSLO or Zemax, pull in one of the basic 50mm lens
models and see if you can squeek out an f/4.0 teledecompressor that does what you want in terms of
the image size. Then think about f/2.0 and let us know if you figured out how to make it all work.

Then you will need to find someone to grind the lenses and a machinists to put them in a housing. Hopefully you
are already a machinist or a lens designer. That will save some headaches.

Actually, if you just design a f/1.0 lens for a 16mm camera image circle and build that from scratch you will be
much happier. You can make all of the image quality versus lens complexity decisions when it's easy to
fix the problem. Once you are using a lens someone has already designed your hands are tied when it comes
to speeding the system up more or less.

 

[Dustin Moore]

The F/1.2 lens will still behave like a F/1.2 lens. The depth of field will be the same, however, the exposure will have increased by 4x the amount of light.

You should check your math on this. Assuming that you keep the pixel size the same between configurations, you may find that an
f/0.7 lens acts like an f/0.7 lens and has the depth of field you would expect from an f/0.7 lens. (Assuming the f/0.7 lens has decent
imaging performance and one can talk about depth of field and mean something.)

 

[Ryan Patrick O'Hara]

It is really hard to make the opposite of a telecompressor (tele-decompressor?) It is also really hard to make
things that are f/0.7 and are not narrow field microscope objectives.

You might want to grab some lens design software like OSLO or Zemax, pull in one of the basic 50mm lens
models and see if you can squeek out an f/4.0 teledecompressor that does what you want in terms of
the image size. Then think about f/2.0 and let us know if you figured out how to make it all work.

Then you will need to find someone to grind the lenses and a machinists to put them in a housing. Hopefully you
are already a machinist or a lens designer. That will save some headaches.

Actually, if you just design a f/1.0 lens for a 16mm camera image circle and build that from scratch you will be
much happier. You can make all of the image quality versus lens complexity decisions when it's easy to
fix the problem. Once you are using a lens someone has already designed your hands are tied when it comes
to speeding the system up more or less.

Tele-extender makes an image circle bigger. It's what we use to double a focal length, hence we make the surface area 4x larger and that is a two stop loss. That is why it's always a 2-stop loss for a 2x extender.

What I want to do is the opposite, and those seem to be called focal reducers. It takes the image circle and condenses it into a smaller circle. Therefore the same amount of light is being pushed into a smaller space, which means more photons per photosite.

The goal is to compress full frame image circle down to what is required of an image circle for 16mm sized sensors.

I'm really doing this for fun. I'll be experimenting with different lens elements from broken lenses, focal reducers, condensers and other fun stuff in concert with the already made f/1.2 55mm lens, which I may tamper with, if I find I need to. My goal as of now, was to allow the lens to operate as normal, and simply do my best to wrangle the light coming out of the lens and reducing the image circle down to as small as 16mm sized sensor. I know what you are saying and I'm not out to create a f/0.7 lens from scratch and have it properly housed. There will be no lens designer. I'm just playing with focal reducers from telescopes and various optical elements, using those elements to teach myself more about lenses, and to see if I find anything cool along the way. Most likely the contraption will be temporarily housed by using a bellows system track modified to hold lens elements, and the final design will be made from spacer tubes and adapter rings.

I'm more focused on using a telescope focal reducer instead of the condenser. The condenser is going to be for experimentation, as it seems completely inappropriate for my goals.

You should check your math on this. Assuming that you keep the pixel size the same between configurations, you may find that an
f/0.7 lens acts like an f/0.7 lens and has the depth of field you would expect from an f/0.7 lens. (Assuming the f/0.7 lens has decent
imaging performance and one can talk about depth of field and mean something.)

I don't follow. The comparison is between the DoF from the lens on a 16mm sized sensor and the same lens with the condensed image circle on the same sensor. Why would the CoC change if the frame size and pixel size remain constant? I'm not comparing two different formats. Simply what a 55mm f/1.2 on a 16mm sized sensor would look like when much of the lost image circle was compressed back onto the sensor. The difference is there is more of the image circle on that frame. I have never played with doing this before, so perhaps I don't know. I figured DoF would be the exact same, and perhaps perceived depth would increase as the 'blur circles' would be smaller on the recording medium and thus look sharper (not taking into account the sharpness lost from my crappy element assembly)/


I'm doing this to learn and explore, so I really do appreciate you mentioning this. Can you please go more in depth because I'm trying to wrap my head around what will happen when I condense an image circle.

 

[Dustin Moore]

The comparison is between the DoF from the lens on a 16mm sized sensor and the same lens with the condensed image circle on the same sensor. Why would the CoC change if the frame size and pixel size remain constant? I'm not comparing two different formats. Simply what a 55mm f/1.2 on a 16mm sized sensor would look like when much of the lost image circle was compressed back onto the sensor. The difference is there is more of the image circle on that frame. I have never played with doing this before, so perhaps I don't know. I figured DoF would be the exact same, and perhaps perceived depth would increase as the 'blur circles' would be smaller on the recording medium and thus look sharper (not taking into account the sharpness lost from my crappy element assembly)

The depth of field depends on the circle of confusion, the f-number and the distance to the object that is supposed to be
perfectly in focus. Over on http://en.wikipedia.org/wiki/Depth_of_field#Derivation_of_the_DOF_formulas they define the depth of
field as the difference in distance between the farthest thing that is mostly in focus and the nearest thing that is mostly in focus. Then they
rewrite the distance to the object that is supposed to be perfectly in focus in terms of the magnification m. If c (CoC) and m in constant,
for close objects they find
DOF ~ 2 N c (m+1)/m*m
where N is the F number (for close focus). If you half the F number, you will half the depth of field. For far focus, the formulas are messy but
if you plot them relative to F number you'll find that smaller F number always pulls down the depth of field.

Side note on the telescope telecompressor, remember that most of those are designed for an f/10 or f/6.3 input beam and go to around f/3.0
at best. Going from f/1.2 to f/0.7 is a big deal in terms of lens complexity.

 

[nothing]

The focal reducer will shorten the effective focal length of the lens and consequently (assuming the aperture remains constant) change the f-stop. It should have DOF similar to the faster lens it becomes, or perhaps I'm missing something.

 

[Ryan Patrick O'Hara]

The focal reducer will shorten the effective focal length of the lens and consequently (assuming the aperture remains constant) change the f-stop. It should have DOF similar to the faster lens it becomes, or perhaps I'm missing something.

Interesting. I'm still not certain on this.

To play devil's advocate, if I were to go along with this notion and say that the lens has changed qualities and now inherent the DoF of the faster lens it has appeared to become, wouldn't the focal length change be directly proportional to the gain in sensitivity? For instance, because of the inverse square law, a 2x teleconverter loses 2-stops of light, because the surface area is four times as large. So playing with this reasoning, theoretically, this lens were to have shallower depth of field, but at the same time inversly be a wider focal length of proportional means, and we know that wider focal lengths have deeper stops.

Lol. Something to think about.

I assumed the lens would keep it's same properties, the only difference is that we see more of what used to fall off the sensor, thus it's more of a FoV change than it is a DoF change. It's the CoC part I've got to figure out. I love this talk though. Good way to start my brain in the morning. Thanks to everyone who is participating. Can't wait to figure this all out.

 

[Ryan Patrick O'Hara]

The depth of field depends on the circle of confusion, the f-number and the distance to the object that is supposed to be
perfectly in focus. Over on http://en.wikipedia.org/wiki/Depth_of_field#Derivation_of_the_DOF_formulas they define the depth of
field as the difference in distance between the farthest thing that is mostly in focus and the nearest thing that is mostly in focus. Then they
rewrite the distance to the object that is supposed to be perfectly in focus in terms of the magnification m. If c (CoC) and m in constant,
for close objects they find
DOF ~ 2 N c (m+1)/m*m
where N is the F number (for close focus). If you half the F number, you will half the depth of field. For far focus, the formulas are messy but
if you plot them relative to F number you'll find that smaller F number always pulls down the depth of field.

Side note on the telescope telecompressor, remember that most of those are designed for an f/10 or f/6.3 input beam and go to around f/3.0
at best. Going from f/1.2 to f/0.7 is a big deal in terms of lens complexity.

I'm on my way to LAX, but I'll be plugging away at the DoF formula later to see. It's been awhile since I've actually used the formula and not some DoF calculator online. Going to get back to the ground floor. And yes, the focal reducers are made for F/10 -> F/3.3 telescopes. Things are going to get messy. I'll be trying some achromats out to see if I can negate some of the effects. I said it was going to be funky!

 

[Dustin Moore]

To play devil's advocate, if I were to go along with this notion and say that the lens has changed qualities and now inherent the DoF of the faster lens it has appeared to become,

If you put a bunch of glass together that makes an f/0.7 beam hitting the image plane, you got a f/0.7 lens. The
only caveat is that if you can't get a picture out of the lens because of other aberration problems, depth of field
might not be that important.

wouldn't the focal length change be directly proportional to the gain in sensitivity? For instance, because of the inverse square law, a 2x teleconverter loses 2-stops of light, because the surface area is four times as large.

So it is true that a lens with a 2x teleconverter will have focal length that is twice that of the lens alone and that
therefore the f number will be the original f number divided by two.

This does not mean that you have four times as much light hitting a particular pixel. The doubling of light
goes by f/2.8->f/2.0->f/1.4->f/0.98->f/0.7. An f/0.7 lens is four times as bright as an f/1.4. That's also a
good idea of how much harder it is to make an f/0.7 lens than an f/1.4.

The brightness goes like (f#)^(-2). The inverse square law doesn't work on images and the actual math is a great
way to ruin a perfectly good weekend. If you want to see the proof it's equation 2-77 in
http://books.google.com/books?id=Jg...e&q=irradiance due to lambertian disk&f=false

So the size of the image doesn't really determine the brightness. It's dependent on magnification which
is dependent on focal length which is dependent on f/number which controls brightness.

 

[Ryan Patrick O'Hara]

Before I upload photos and information, just wanted to say that so far, my initial tests have been interesting.

I've been using a f/2.5 105mm lens to test, and so far have gained a stop and a half - a stop and a quarter range on it. At an f/4, I got it to match my Cooke Panchro/i at T/2.8.

The Nikon was definitely the way to go. Because of flange distance, the Nikon has one the longest/deepest, while the camera I'm using (NEX-5 right now, later the GH2) is short. This allows me room to add elements in the middle. Sadly, because of where I have to place the focal reducer, it does not reduce to the amount it does in telescopes. It was advertised to gain at least two+ stops in telescopes, of course placed differently in the assembly. I think a f/10 - f3.3. Instead, I'm seeing a stop and change.

I will be keeping at it, but it seems if the f/1.2 50mm follows suit, we may be looking at something below a f/1 but not a .7. Maybe a .8 or so.

I've taken apart about $2,000 worth of lenses I no longer use. They are really quite fun to take apart. It's a learning experience and I'm having fun.

Not giving up yet.

-ryan

 

[BlackWolf]

Interesting project. AFAIK, Zeiss simply used a focal reducer for the Planar f0.7 to reach its fast aperature.

I've been playing around with some fast lenses myself recently, including this 60lb 130mm f/0.7 monster. Covers 80x80mm sensors!

 

[TimurCivan]

I've seen this in person... its awesome.

 

[Jordan Scott Price]

interesting project. Afaik, zeiss simply used a focal reducer for the planar f0.7 to reach its fast aperature.

I've been playing around with some fast lenses myself recently, including this 60lb 130mm f/0.7 monster. Covers 80x80mm sensors!

Awesome!