Welcome to our community

Be a part of something great, join today!

Log in
  • Hey all, just changed over the backend after 15 years I figured time to give it a bit of an update, its probably gonna be a bit weird for most of you and i am sure there is a few bugs to work out but it should kinda work the same as before... hopefully :)

Single Chip 2/3 cameras?

Viper
(Zodiac, Curious Case of Benjamin Button)
 
2/3" C mount machine vision lenses >100lp/mm

2/3" C mount machine vision lenses >100lp/mm

If you look at some of my other posts you will see some information on the subject of 2/3" lenses for Scarlet and Acam dII.

There are many 2/3" C mount machine vision lenses that are quite sharp and resolve better than 100 lp/mm when stopped down to prime stop.

You can also get 17mm, 25mm, and 50mm f/0.95 lenses form Goyo Optical for less than $400 each (it seems) and other also make f/1.2 and faster lenses for 2/3" format that would look about the same as lenses of that speed used on academy 35mm T500 and probably better than many older 16mm lenses used for Super16. I am doing some tests of a few of these machine vision Megapixel lenses I picked up now, they look quite crisp, there are several classes of C mount lens,

1) TV lens, maybe resolves 300 to 500 lines on 2/3" format.

2) High Quality, maybe resolves 400 to 600 lines on 2/3" format.

3) Megapixel, maybe resolves 500 to 800 lines on 2/3" format at prime stop.

4) 5 Megapixel, maybe resolves 700 to 900 lines on 2/3" format at prime stop.

The measured resolution depends on how much you sharpen the image during the de-Bayer (de-mosaic), video uses low resolution lenses and over sharpens to get better numbers, Digital Cinema should use the best lenses and not sharpen as much.

Kowa makes a 5mm that will cover 2/3" that is low distortion and high resolution, it should be something to look into,

http://www.prosecuritys.com/lm5jcm.html
 
Steve Ellington said:
. . .are there any other single chip 2/3rds cameras . . . what will be the best lenses
Click to expand...

The 16mm film frame is almost the same as 2/3".

There is, and will be, a lot of naysayers, who say that old 16mm lenses will not be sharp enough to take full advantage of your sensor. But if some 16mm films have been shown on the big screen, and if I can buy a used lens for a few hundred dollars, um, I would be a nerd if I turned it down just because technically I am not using all of my pixels.
 
16mm Cine lenses...

16mm Cine lenses...

combatentropy said:
The 16mm film frame is almost the same as 2/3".

There is, and will be, a lot of naysayers, who say that old 16mm lenses will not be sharp enough to take full advantage of your sensor. But if some 16mm films have been shown on the big screen, and if I can buy a used lens for a few hundred dollars, um, I would be a nerd if I turned it down just because technically I am not using all of my pixels.
Click to expand...

Some Cine lenses are good enough to get maybe 900 V lines, which is about what the 2/3" sensor will put out with a OLPF on it.

A bigger issue is that the sensor can be sharper than film ever was, so you should not stop the lens down past f/5.6 if that is the best stop if resolution matters since from f/8 down to f/32 the image will get softer everywhere. From f/5.6 the image will also get softer as you open up past f/2.8, so you really only have three stops you can use for 3K resolution f/2.8, f/4, and maybe f/5.6.

You need realy good lenses to be diffraction limited at f/2.8, most are not until f/5.6 or f/8.

When shooting S35 diffraction does not become a big issue until you stop down to f/11 or smaller, so you can stop down about two more stops when shooting S35 rather than 2/3". When shooting EE35 you can get away with f/16, something you would not use on 2/3" since at f/16 a 2/3" sensor will be reduced to SD or sub-SD resolution.

I am looking into if (5) Megapixel 2/3" Machine Vision lenses are a better value and look better than older Cine lenses. From what I have seen so far, the results from some look quite crisp, the computer optimized and better coated newer lenses may work out better on a digital camera than 50 year old Cine lens desigins, especially at stops larger than f/4.

One thing to think about is that when shooting with a 75mm f/1.2 lens on a real 3D subject rather than a test chart, is that 95% of the frame is very out of focus, so what little difference there might be between lenses in the focal plane only affects a very small slice in the image the other 95% of the frame will look about the same.

Video lenses of the past, and some cheep security lenses would not be a good choice, and some of the better new C mount lenses are made with poor plastic mounts, but not all of the lower cost C mount lenses are poor quality as far as the glass part is concerned, so if you need a few low light shots, a cheep f/1.2 8mm 2/3" Megapixel lens might get the shot for you better than a f/2.8 lens costing 100 times as much, since on the screen anything with more than 300 or 400 V lines look like 16mm film or better.

http://www.kowascope.com/frontend/Proddisp.asp?co=10000349

http://www.kowascope.com/frontend/proddetail.asp?pn=LM5JCM&co=10000349

The spec sheet for this Kowa 8mm f/1.4 2/3" lens says it gets 120lp/mm in the center, since the sensor is about 4-5.5mm high that mens you get about 960 to 1320 video lines resolution. Very few lenses that cover 2/3" would resolve better than 150lp/mm, if you have a link to specs for one please post it I would like to look into that lens...

http://www.kowascope.com/images/product/LM8JCM.pdf

Some of the other brands such as Tamron still have feet and meter markings and f/ stops with clicks, but others are not marked in part because the cameras used lack a backfocus adjustment to compensate for the effects of the OLPF and sensor cover glass on the change in backfocus for different f/ stops on the setting for "best focus" at high resolutions.
 
Dan Hudgins said:
... I am doing some tests of a few of these machine vision Megapixel lenses I picked up now, they look quite crisp, there are several classes of C mount lens,...
Click to expand...


Dan it will be great if you can post 1:1 framegrabs from your tests. Because without seeinig is no believing. :biggrin:

I has same feeling that most of old 16mm will not keep well with high resolution chips and its worth to wait what will RED came up with their small primes and especially in what pricerange.

I was looking for what is available for 2/3" and besides of best ones ARRI Ultra 16 primes for about USD 80.000,- (not sure) http://www.arri.de/prod/cam/ultra16/ultra16.html
Wonder if DigiPrimes are better for use with Scarlet than Ultra 16 ones?

I found Optar Illumina S16 Ultra Prime Lenses Set of 6 for USD 16.000,-
http://www.optarillumina.com/lenses/super16.html

c-mount lenses from Linos
http://www.pstechnik.de/en/optics-c-mount.php,

Kinoptic http://www.kinoptik.com/optik_obje_cine.htm

Also dudes from Ikonoskop used their Ikonskop Kinoptix 9mm PL lens for thier demo footage and it looks very good. But I dont know if Kinoptix = Kinoptic or if sell them.
 
combatentropy said:
But if some 16mm films have been shown on the big screen, and if I can buy a used lens for a few hundred dollars, um, .
Click to expand...

Hi,

I don't know of any 16mm films that have been blown up to 35mm for 25 years. You are probably thinking of S16 that uses newer and rather expensive lenses. Those lenses were until RED came along more expensive than 35mm lenses, one reason I bought 35mm equipment & not 16mm.

Stephen
 
What is > 120lp/mm?

What is > 120lp/mm?

Petr Dvorak said:
Dan it will be great if you can post 1:1 framegrabs from your tests. Because without seeinig is no believing. :biggrin:

I was looking for what is available for 2/3" and besides of best ones ARRI Ultra 16 primes for about USD 80.000,- (not sure) http://www.arri.de/prod/cam/ultra16/ultra16.html
Wonder if DigiPrimes are better for use with Scarlet than Ultra 16 ones?

I found Optar Illumina S16 Ultra Prime Lenses Set of 6 for USD 16.000,-
http://www.optarillumina.com/lenses/super16.html
Click to expand...

I was thinking about doing some 1:1 blowups to post here, but it will have to wait until next year since I do not have a C mount camera sharp enough to measure better than 120lp/mm right now.

$80000 seems insane for a 25mm C mount lens (?), as I pointed out in some of my posts here stops smaller f/5.6 cannot resolve 3K on a 2/3" sensor (depends if you measure using IR-RED or BLUE-UV since the UV gives some extra resolution). If the lens is diffraction limited in the corners at 2.8 it might be sharper than 3K but at f/4 you are about even and at f/5.6 you are getting on the down hill slope.

At stops bigger than 2.8 most of the frame is out of focus for real subjects, so having the lens better than SD at f/1.4 would not change much in the image and the OLPF will soften the small details anyway.

You can download the free version of OSLO and see for yourself what changing the f/ stops do to the MTF curve for a lens that only covers 2/3" sensor size (+/- 5.5mm image field angle).

The Kowa lenses are about 2.8 since it is very hard to make fast lenses that are 120lp/mm in the center. So they would only be used from like f/4 to f/5.6 the exposure would need to be controled by ND and shutter angle.

If you are going to shoot at f/5.6 and f/8 I do not think there would be much difference to see between a good distortion free 120lp/mm megapixel lens and the $$$$ lenses, but as time goes on I and others I am sure will try to post images taken with various C lenses so we can see how Scarlet will do.

Its too bad the manfactures of those lenses do not post images shot with their lenses so you can tell what you are getting before you buy one. Some dealers will let you test the lens or have samples for testing, so it does not hurt to ask. Some lenses may be good for f/1.2 but others may be better in the same focal length at f/4 so you realy need to get a feel for what range each lens will work best at.

I have noticed that some of the newer lenses have a three blade iris, what a stupid way to save money, having a triangle iris reduces the resolution at a given T/ stop because the points of the triangle are in outer zones that would be closed off with a 12 blade iris for the same light transmission. The points also cause diffraction streeks, a little like a star filter. Also you get triangles in the out of focus areas, so sometimes the old movie lenses with their multi-blade iris might be good for out of focus background shots etc.
 
Dan Hudgins said:
I was thinking about doing some 1:1 blowups to post here, but it will have to wait until next year since I do not have a C mount camera sharp enough to measure better than 120lp/mm right now.
Click to expand...

For testing I was thinking about to make C-mount to Nikon/Canon DSRL adapter.

$80000 is price for whole set of 5 ARRI/Zeiss Ultra 16 Mk. III Primes which are not c-mount. Its from Cinematechnic site
http://209.85.129.132/search?q=cach...chnic.com+Optex+lens&hl=cs&ct=clnk&cd=1&gl=cz
 
2/3" Lens testing...

2/3" Lens testing...

Petr Dvorak said:
For testing I was thinking about to make C-mount to Nikon/Canon DSRL adapter.

$80000 is price for whole set of 5 ARRI/Zeiss Ultra 16 Mk. III Primes which are not c-mount. Its from Cinematechnic site
Click to expand...

I think I have seen a PL to C mount adapter once, maybe that would work if they come in PL, but $80000 is quite a bit of money for maybe 50% more resolution at two of the f/ stops? Can you find the MTF and lp/mm curves for that set at all stops?

C mount is quite a bit closer to the sensor than Nikon or Canon, 17.525mm Canon EOS is more like 44mm (see link below) and Nikon is 46.5mm according to this link,

http://www.bobatkins.com/photography/eosfaq/manual_focus_EOS.html

I use good M42 lenses on a Canon XTi sometimes and that works but you cannot focus to inf with a simple adapter on a Nikon as far as I can tell from the f2f numbers.

You would need to lock up the mirror and push the lens into the camera, which probably would not work since it would be too big.

Another way is to make a sort of reverse DOF adapter where you put a Macro lens on the Canon or Nikon and shoot a small ground glass behind the C mount lens, that would let you get 12Megapixel images of the C mount lens image. You could also add a field lens or just use a field lens, but the resolution photographed using just the field lens would give better results than when the lens is on a CMOS C mount camera perhaps.

You cannot use film to test the lenses for two reasons,

1) Color film is not sharp enough to measure >120lp/mm

2) very small errors in focus will make some lenses look worse than they are

You realy need a small megapixel C mount camera that has a 2/3" sensor to shoot stills with that gives 4K resolution. There are some for technical use, but when Scarlet comes out may be a better time to start testing lenses since you can gauge the losses caused by the OLPF, redcode, and other issues.

If you use an industral camera without OLPF you might see that one lens is a tad cleaner than another at f/2.8, but when you test them again later on Scarlet you may not see as much difference, and if you purchased the one that you thought was better, only to later find out that you cannot see much improvement on Scarlet you might then wish you went with another option.

I was bringing up the machine vision lenses because they are used on high resolution Megapixel C mount cameras and are made so that you can get clean sharp images, if not wide open at least at prime stop. Many older movie lenses for 16mm probably look quite soft on a megapixel C mount camera since even on a SD C mount camera you can see that they get sharper when you stop them down to f/4 or f/5.6, if you can see that on a SD camera, then you would surely see it on a 3K camera that has maybe 2.5 to 3.5 times the V resolution.

C mount zoom lenses are another can or worms, the ones I have looked at do not track well, if you set the backfocus for one focal length and zoom the focus does not hold it can be made sharper if you tuch up the focus ring. This is not so much a problem with the lenses I have, as far as I know all zoom lenses have a tolerance on their optical and mechanical compensation for the zoom optics movement, that lets the focus float around the right value with the focus getting sharper and softer as you zoom over the range. The tolerance was not too big of an issue for film cameras since the resolution of the film and projector were not high enough to notice the errors much, but on a 3K camera you will probably see the focus weave around and be hard to get tack sharp at all points alonge the zoom range, and worse if you turn the backfocus for the center of the range both ends may go soft, and the other way around.

The way some HD camcorders get around the zoom lens tracking errors issue is to use auto focus, that wat it does not matter if the lens has too large a tolerance since the feedback loop in the auto focus can push the focus back and forth as the lens zooms or the stop is changed, auto focus also compensates for the "plate effect" issue caused by the OLPF and sensor cover glass introducing negative spherical aberration.

If the $$$ Super16 lenses are not corrected for the thickness of the sensor cover glass and the thickness of the OLPF (which can be 2mm or more) then much of its advantage at wide stops may be lost since the "plate effect" of having glass behind the lens will introduce negative speherical aberration etc., at 150lp/mm it does not take much extra aberration to end up with <100lp/mm if there is glass in the cone of light or the focus is off a tiny amount. Even a lens focus error of 0.02mm could cut the resolution in half, even more for fast lenses. Lenses made by RED should be computed for the plates in Scarlet, i.e. made just for that camera and not of general use.

Focus error is a big issue for lens testing, you need to do a slow rack focus and pick the best frame, the numbers on the lens scale if it has them, would "never" be so trustworthy as to give a true result on the first try.

People used to working will film cameras probably underestimate the critical nature of getting 3K off a 2/3" chip, the focus tolerance will be two to three times as critical as it is now for the RED ONE, and maybe up to six times as critical as for a 16mm camera.

Will the chip be in the flat field within one micron? If not the focus could be uneven center to corners, making measurments for a flat field lens maybe look worse than one with a curved field. So when you see tests, it would be best not to make judgments based on small differences between lenses that may or may not be the result of testing errors, but rather to look at a range of real world tests on 3D subjects and see if the end result looked at with a full frame image at a resonable distance looks pleasing.

I had endless complaints from my Brother about film prints not being sharp enough with him looking at tests over a light box with a 20x magnifier, when viewed in a 35mm movie projector much if not all of the extra resolution I got into the result was invisible, so I got him a 5x magnifier and told him to use that so that he would see what the theater audience would see from a normal distance, not the view of someone standing on the stage six feet from the screen...
 
I was looking through this thread and wondered about conveters for 2/3 lenses. We use the Kowa 10m pixel 2/3 lenses which are very sharp on small 1 Cmos cameras. If you wanted to use a 1 chip camera with a 2/3 lens such as an HJ11 or HJ22 would a simple mechanical adaptor work ? or should something like an abekus adaptor with corrective optical be better ? I also wonder if using say a Canon 7 - 63mm S16mm lens be more suited as it images on the film plane rather than a 2/3 block. Ive not got any lenses to test here right now byt any ideas would be appreciated.
 
If you Google C-mount lenses on M4/3 cameras you will find quite a few stills taken with various vintage 16mm/TV lenses and newer lenses. Some do extremely well, some are very soft. On a Lumix GH1 the best vintage 16mm lenses are quite sharp and have good color contrast. The images degrade into aliasing before you see any real softness in the lens. I have played with some of my 16mm zooms on a 1/3" 570(H) line SD security camera and a late model 12-120HEC Angenieux zoom shows no visible difference in sharpness at any stop below T11, even wide open. Here is a link to a video (not mine)shot with an old Angie 12-240 on a GH2 at 2x digital zoom http://vimeo.com/22717894
I think many old lenses are capable of producing quite acceptable results on a Scarlet though they will not exhibit the overall MTF and color contrast of newer lens designs. From playing with the R3d files of Felix shot with a Scarlet fixed the OLPF is designed to pretty much eliminate any visible aliasing at its native 3k pixel resolution. The image has a more organic film like texture to it than any native 1080p camera I have ever seen. A lot of shooters like older 35mm glass on Red 1 because it also results in a smoother more film like texture to the image compared with the razor edged sharpness of modern glass.

A good option for new(or used) PL mount S16 primes are the Illuminas http://www.lumatechinc.com/16.html But Red's own mini-primes are likely to be the best and most economical choice for high quality primes at around $1k each
 
David,

Thanks for the link. I think that using older glass is real important with Red and other cameras, its something always been sucsessful in the area that we work and Id like to learn more. My question really is how best to couple them to a single sensor camera, I realise that there is an issue of design between the newer lenses designed for Digital sensors. But Im interested to know if using new glass designed for 3 chip cameras is best directly coupled or is best used with some form of optical device to reduce any fringing and also best practice of using older glass designed for film.

Also im interested in using high end 2/3 lenses as some do have an extended zoom range which can be essential in some areas. Im sure if coupled corectly they will perform very well on Scarlet but if used with the Abekus 260 device Id be interested to know how good the image is on a Red One or Epic, it may not be an ideal match but may be workable.
 
3-chip HD camera lenses need optical correction to focus properly on a single sensor camera. Otherwise you will get some chromatic aberration. I don't know that B4 lenses would be any better than premium modern S16 glass. Canon makes a couple of good zooms but you would have to add aftermarket motors for servo controls. With a corrected B4 mount, high end HD lenses should perform very well on a Scarlet though.
Red 1 has a reputation for being a bit soft at 2k due to the OLPF design. Epic seems to be significantly better judging by some of the 2k 300 fps clips.
 
You must log in or register to reply here.
Back
Top