Sharpness

[Mitchell Lewis]

While I'm very happy with our Sony EX3 / Letus Ultimate / Nikon lenses setup, I find myself shooting wide open quite a bit. I'm hoping that Letus will release the EX3 relay lens soon as according to Phil Bloom, you gain almost 2 stops. (that would be a BIG help)

With all this in mind I've noticed that my footage never looks razor sharp. It does look sharp, just not razor sharp. I watch broadcast television and the sitcoms on the networks always look razor sharp but still have shallow DOF. Maybe I'm asking too much of our new rig, after all it's only an $8500 camera with a bunch of extra glass (Letus, GG, etc...) Don't get me wrong, I LOVE IT! But I'm wondering that because I'm shooting wide open my images won't be sharp as a tack? I'm pretty sure this isn't a focus issue. BTW, I've also been shooting with the detail turned off lately and surprisingly I haven't noticed much difference.

Phil Bloom always said that a Letus adaptor gives footage an "organic look" and he's correct. It looks very film like. But if once in a while I want razor sharp images like on the Discovery Channel, maybe I should shoot without the adaptor? I need to do some more testing, but maybe someone else has experience with this?

I shoot with the following Nikon lenses: 17x35mm f/2.8, a 5mm f/1.4 and a 80x200mm f/2.8

[Bob Hart]

Groundglass relay of any form will have a softer, lower constrast look to it.

With the Letus Extreme on the SI2K, the finest pitch resolution bars on a Lemac chart resolve but there is also softness across them compared to direct-to-camera.

You will also observe this with an individual 16mm film frame with the grains of the emulsion although its resolving ability is finer..

This is the nature of groundglass diffusion, a one-generation removed picture of a picture or as P+S Technik describe it, a non-coherent image.

A 5 micron groundglass random texture on a 35mm sized image cannot compete with 35mm film with finer emulsion grain or a 35mm digital sensor which has its straight edged 5 micron pixels arranged in organised rows and can be artificially sharpened.

If you deselect sharpness settings in the EX3 when you shoot direct-to-camera you will discover the native camera image is not quite so elegent after all.

The groundglass adaptor is not an everything tool, just a creative choice.

Try to avoid absolute wide-open with all lenses. With the Nikons, that last 1 to 1.5 stops adds flare. Any defecits in the lens image onto the groundglass are amplified. - A 1 micron pinpoint to the groundglass becomes about 5 microns. A 5 micron pinpoint diffuses to 25 microns more or less. A 25 micron pinpoint obviously diffuses wider. This is only a speculative illustration on my part as the groundglass "grains" are not organised but are random.

If there is a flare around the pinpoint, then this flare is also amplified in area by the groundglass and spoils contrast performance leading to an apparent lower resoution as it bleeds back across the actual sharp pinpoint. This characteristic is even more apparent with tube based IR night-vision.

To some extent this also applies to an image cast across rows of pixels. Whilst one or a small group of pixels collects a pinpoint, a flare around the pinpoint encounters a far larger number of pixels. At a certain brightness threshold, when the flare is dominent, an apparent resolution loss may occur.

The beauty about direct relay is that the complexity of a zoom lens is avoided, therefore the relayed image should be both sharper, of better contrast and brighter, meaning that you can come back off the widest iris setting on the objective lens and the relay lens a stop or two into sweeter performance. You win in about three ways as the performance gains are interactive.

Please do not ascribe authority to my comments, as I am straying into theory I do not have all the knowing about.

[Mitchell Lewis]

Thanks Bob.
So you don't think I'll gain any significant sharpness shooting at higher f-stops? (both the EX3 and Nikon lenses)

[Bob Hart]

I think you will get increased "apparent" sharpness but there will be a point soon reached when the creative advantages of being able to use shallower depths of field will be compromised.

If you have a Nikon which is f1.4 or even f1.2 wide-open, at f2.8 with these lenses you are venturing closer to their sweet spot than with a f2.8 lens wide-open.

The sweet spot for the f2.8 lens might be closer to f4 or even f5.6. However the counter-argument is that some of the modern f2.8 lenses are better wide-open than their ancestors used to be and the difference may not be as noticable. The Nikon 35mm and 28mm f2 lenses are supposed to be as good as their f1.4 ancestors at f2 - f2.8. For much of the time, you would sensibly shoot with the lens set at f3.5 to f4.

Some f2.8 lenses with their smaller exit pupil may vignette or have corner brightness fall-off at the smaller iris settings, which is one other reason to go for the f1.4 and f1.2 lenses which have larger exit pupils.

Running around shooting with paper-thin depth-of-field is the signature of the new owner on his first date with his groundglass adaptor. Just because you can does not mean you should.

I find flare to be as much a curse as soft image with groundglass work. Deeper contrasts convey the illusion of better sharpness.

I did not realise how much of an issue it could be until I tried using Nikon prime lenses for relay lenses when that last 1 to 1.5 stops on the 35mm f1.4 Nikon and 28mm f1.4 Nikon, as good lenses as they are, became apparent in the relayed image.

Admittedly this was a misuse of the lenses but pointed me to the fact that it is there.

My personal preference therefore would be to ND before the Nikon lenses with a mattebox.

This is to keep the lightfall onto the groundglass as close to the iris sweet spot of the camera lens as possible without having to switch camera ND to more than ND1. I understand the camera is happiest with its own iris setting at f4 - f5.6.

The downside of using NDs on front of the lens in a mattebox is that not all NDs reduce the light without letting some infra-red go through. This disproportioned balance of natural versus infra-red may fall outside of the range the camera is intended to deal with.

An additional low-cut IR filter may need to be used as well otherwise the blacks may aquire an orange cast which you can't grade out in post.

[Mitchell Lewis]

Thanks Bob. I'm "hoping" (crossing my fingers) that when Letus finally releases the EX3 relay that it will give more sharpness for the reasons you describe:

1) 2 stops more light means I won't have to shoot wide open = more sharpness
2) Less glass compared to the EX3 lens it replaces = more sharpness

The big question is....will the sharpness improve enough to make it worth the hefty price?

[Bob Hart]

It will be your call on the worthwhileness or otherwise I am afraid. I have not had a play with the EX3 relay as has Philip so I cannot provide that assessment for you.

With groundglass and pretty much a lot of other things, you can get to about 85% of a theoretic 100% performance score fairly quickly. The last 15% represents the impossible.

Getting towards 100% becomes incrementally more expensive for smaller and smaller gains it seems.

I am hoping to get my DNA all over a Letus EX3 relay not too far down the track to try out on a SI2K, which shoots a variety of pixel size and area formats.

At 2048 x 1162 resolution and the widest possible groundglass area I can wring out of the Letus Extreme (yes it is tricked up a bit), on the "A" and "F" blocks of the Lemac chart, the SI2K can see the finest lines which the notes on the back of the chart suggest are film resolutions only. Bring up the sharpness in post and they are definitely there to be seen.

Going for this widest area has some drawbacks. The really wide Nikon mount lenses exhibit edge and corner brightness falloff. These are the ones in the f2.8 and f3.5 zone. It is only f1.2 to f1.8 territory for this arrangement but this was also with a Nikon stills lens for relay. This is partially compensated for by the wider view now yielded by these lenses.

The Le brother's purposed relay may do a whole lot better sharpness-wise on the 2K but maybe not, as the 2K is a single 2/3" sensor camera, not a 1/4", 1/3" or 1/2" 3xCCD/CMOS like most prosumer cams.

The Letus relay has been R&D-ed to deal with the prism split system on these cameras and may therefore be less optimised for a single sensor camera.

I am hoping that it works just as well on the 2K because if it does and it also enables the widest possible view of the Extreme's groundglass, then it also enables some serious creative options as well as the dynamic range in which this remarkable system excells.

Currently the EX3 (1/2") and B4 (2/3") relays are optimised to be faithful to the standard 35mm motion picture frame so the "apparent" resolution cheat of zooming back furthur with the camcorder for relay is no longer available. However, Phil Bloom's demo footage indicates that there is no comparative resolution loss and an apparent gain by relaying direct.

The Extreme and maybe a tricked up Ultimate can possibly go furthur. 35mm groundglass relay may have a use-by date but it has gone about two years past initial predictions and may go furthur yet if somebody discovers a new diffusion material as it will be a cost-effective alternative to 22mm single or 3xCCD or 3xCMOS sensor systems.

[Mitchell Lewis]

Thanks again Bob. I'm hoping to get my hands on an EX3 relay at NAB in a couple of weeks. (cross your fingers)

[Bob Hart]

Mitchell.


If you do get your hands on one at NAB would you do a check for me.

With the relay and attached Letus dismounted from the camera, could you measure the diameter of the image footprint as it falls onto a piece of paper in sharp focus.

A trick for doing this is to draw several concentric rings on the paper of several diameters starting from say 8mm and getting wider in 1mm increments to about 20mm. You then take note of which ring is filled edge-to-edge by the circular image whilst the image remains sharp.

If the image is rectangular, then the corner-to-corner fit whilst the image remains sharp is what matters.

If my experience at the Sony roadshow is any example, the attendants are going to freak out, have plastic pups and want to intervene before you chance damage to anything.

[Mitchell Lewis]

I don't quite understand what you're wanting Bob. Maybe you can draw the circles (rectangles?) and make them an attachment so I can see them?

But yes, I'd be happy to give it a try if I can get my hands on one.

[Bob Hart]

Mitchell.


The circles are drawn one inside the other. Rectangles are not needed as the circles represent the diagonal corner to corner measure of a rectangle.

I want to establish just how big an area the image throw from the EX3 relay covers when the groundglass image is sharp. If you have circles drawn it is easier than holding everything together and trying to measure the image width with a ruler.

The simpler test would be to draw a 2/3" or 18mm diameter circle. If the image thrown by the EX3 relay fills this circle without the edge of the image being all furry and soft, then that is as much as I need to know.

I am after using a wider than normal view through the EX3 1/2" relay to a SI2K 2/3" sensor to cheat a wider groundglass area and improved "apparent" resolution like I can with the EX1.

The Letus 2/3" B4 relay would work immediately out of the box via an IMS B4 mount but the groundglass view would be restricted to the 24mm movie frame at 2048 x 1152.

Then at 1280 x 720, the active sensor area on the SI2K comes down to about 1/2" which would lose groundglass image area and apparent resolution if the 2/3" relay remains in use on the camera.

At NAB there may likely be too many people waiting in line to play with the hardware, so don't fret over trying to do this test.

[Mitchell Lewis]

Okay, so I'm going to print out this graphic (8.5 x 11 sheet of paper) and use it to do the following:

1) Assemble the Ultimate adaptor and EX3 relay
2) Don't mount to camera and don't bother mounting a 35mm lens
3) Shine a bright light through the end where the 35mm lens mounts
4) Point the end that mounts to the EX3 at the paper
5) Move the rig up and down until you get a perfectly focused circle
6) See if that circle is larger or smaller than the 18mm circle on the paper

I'm not sure if I'm going to be able to do this at NAB. But when I eventually purchase a relay, I'll certainly will do with for you.

[Bob Hart]

Mitchell.


It can be easier and more valid to use just the relay lens itself and not the whole Ultimate adaptor. This is because the available groundglass image area of the Extreme/Elite is larger than the Ultimate which is intended to faithfully confer the 35mm motion picture image area.

What I am chasing is being able to aquire the larger area. The pool of light from the Ultimate may be vignetted either by the limitations of the 1/2" relay onto a 2/3" area or a vignette from the Ultimate's visible groundglass area.

I don't know if the view through the Ultimate is rectangular like the Extreme/Elite or circular like the P+S Technik Mini35 but whichever it is, the corner-to-corner bright patch or circular bright patch projected onto the paper might not then represent the actual capability of the relay lens itself.

A torch with a front window of one and a half inches or more will be fine. Held between about 7 - 8 inches from the front of the EX3 relay lens, the lamp of the torch should become sharply defined on the paper when the rear of the relay lens is the correct distance from the paper. A few black crisscrossed sharpie lines drawn on the front glass of the torch will be helpful for getting the rear of lens to paper distance right.

Hopefully when these conditions occur, the circle of light from the lens will fill the 2/3" circle.

If this is the case, then the EX3 relay should work for then SI2K using the larger image area available from the Extreme/Elite and the improved apparent resolution which can then be obtained.


Many thanks in anticipation of your endeavours.

[Mitchell Lewis]

Sorry Bob, but I wasn't able to do any testing for you. Zacuto had the EX3 relay in their booth, but it was always attached to the camera. They were so busy (and so was I!) that I didn't want to bother them to take it apart for me to do your test. Sorry....

I did here that this particular prototype has "academy" framing. They will make another adaptor for the EX3 that will be full-frame....better for 35mm lens use as it won't change the focal length of the lens.

I wonder how long it will take Letus to release these lenses?

[Bob Hart]

Mitchell.


Thank you for taking the time and trouble regardless. I did know of an intention to examine cross-adapting the smaller format relays for wider views on the larger formats, in effect catering for 35mm movie frame purists and those who prefer to use as much of the 35mm still-camera frame as possible. The news was not for public consumption at that stage.

If I can get near to 2K "apparent" resolution with a 40mm or 45mm Nikon stills lens for a relay on the SI2K, a purposed relay should do better.