SALT III- High Speed Lens Shootout "WFO" Results

[Nick Morrison]

Have been spending hours upon hours, trying to correct some lens distortions on a heavy vfx shoot (using the LensDistortion node in nukex), will probably need to actually get the lenses and do test shots with the lens on a checkboard grid(I'm lucky that I can actually get the lens back to do the lens distortion charts).

What I noticed from the SALT III test, is that in all cases the higher cost lenses had less lens distortion problems. I could easily see someone spending $100k in vfx hours on a movie trying to undistort lenses for match merges that need camera solves(I'm finding this very important for 3d mesh generation and deep pixel operations ... the meshes turn out curvey on flat surfaces without running the lens distortion nodes). Just another thing to think about if your doing a lens test for a film and there will be a decent amount of vfx.

Patrick, were these distortion problems only on wide angle primes? Or teles as well? (telephoto lenses can have quite a bit of measurable distortion, too; it would appear the more glass elements you add, the more prone the lenses can be to distortion). That being said, are you noticing distortion on any primes besides the wide-angles? Because if you're not, then for VFX shots it may be prudent to stick to the back end of the lens set: a 35, 50 and 85 for example (just thinking out loud...).

Sorry, these VFX considerations are kinda compelling. I wanna know more!

 

[Clayton Burkhart]

Bonus!
There happened to be a set of Schneider Cine-Xenar IIIs at Duclos on our test day and it just wouldn't have been right to leave without at least throwing them up. My very quick impression is that these are the "Sleeper"set of 2013. They looked the closest to the "Optimo" look of any prime set I've yet tried. The lenses were larger than Ultras but not as big as RPPs or MPs. Breathing was non-existant, sharpness was stellar and the bokeh was luscious. I'm hopefully going to be able to borrow a set for a music video coming up so with any luck you'll have a chance to see them in action.

E.

I have been saying this for a while. They really are an excellent set of lenses. If I didn't have to shoot FF sometimes and had a little extra scratch around, these would be my choice as well.

[referring to CP.2's and Canon Cine lenses] Both these sets are re-housed still glass from "L" or "ZF" lines and it shows, the barrels are nice and much more usable that the still equivalents but they do not really measure up to the purpose-designed cine lenses from the big boys. They still breathe too much and have too much CA, especially the Canon primes.

One other thing which I find disconcerting about the CP.2's is that even though they have an EF mount alternative, there is actually an enormous amount of play when mounted on a Canon camera. This is really problematic as it defeats the whole purpose of a Cinema lens and creates image shift which resembles breathing. So even Zeiss needs to tighten up their tolerances with regards to this aspect.

 

[PatrickFaith]

Patrick, were these distortion problems only on wide angle primes? Or teles as well? (telephoto lenses can have quite a bit of measurable distortion, too; it would appear the more glass elements you add, the more prone the lenses can be to distortion). That being said, are you noticing distortion on any primes besides the wide-angles? Because if you're not, then for VFX shots it may be prudent to stick to the back end of the lens set: a 35, 50 and 85 for example (just thinking out loud...).

Sorry, these VFX considerations are kinda compelling. I wanna know more!

Primarily when shooting below 20mm (I only use primes for similar reasons), so you are right ... was with wides. Of course as a newb, I didn't realize this, so have a ton of shots with wides using them often with close focus (which even makes it worse). In Puddles of Light, there are of course, animated puddles everywhere that have to be match moved with the shot. Fortunately nukex ( http://www.thefoundry.co.uk/articles/2011/02/07/213/cameratracker-for-nukex-lens-distortion/) has a node that will undistort the lens before i do the camera solution and point cloud (it requires btw nukex, just not nuke, to do the full distortion solution - it also works much better when you do special checkerboard controlled shots to figure out the lens distortions - ). The nukex nodes will correct the lens, but it takes some time and effort for every composite (and even add's a bit of blur in the processing). Lens problems is something the new crowd of vfx experts can fix in post, but fixing in post starts taking more time/money - so your right again ... about the high end glass being compelling for heavy vfx integration shots(especially were the vfx has to be match moved onto skin and/or a detailed moving surface).

Interesting. I rarely do that at the tail of a production. Usually either before production starts or during the first couple weeks of filming we shoot checker boards and create our distortion geometry for either correction or addition. It's a big time saver getting this done at the beginning of a project as CG elements are created during the production these days. As approved elements come in and surveying/tracking data is culled or created we can get right to work often before the production even wraps.
I've certainly seen variation across primes from different sets, usually slight though. The worst if the set data collector labels a lens as "zoom" instead of marking the actual focal length used on that zoom. iPads and tablets have greatly enhanced that on set workflow because they now have a "checklist" for each shot to go through.

Yep, that's for sure ... often with zooms sometimes people don't write down the exact setting for the shots ... which means that the vfx guy has to spend time to derive the mm setting (which is a hassle). I think on the bigger movies which you work on that have heavy vfx compositing, they probably use cinema glass. As vfx becomes more affordable (nukex cost $8k, realflow i think is 3k, modo 1k) ... then small indie's who normally don't care about the lens quality and aren't really the best when it comes to vfx, will need to start thinking about lens problems. I like Nick's suggestion of staying with 35-80 mm primes for low budget vfx heavy movies.

 

[Phil Holland]

I think on the bigger movies which you work on that have heavy vfx compositing, they probably use cinema glass.

Yep. Lots of PV glass and PL. But, you'd also be surprised how much "other" glass is used. Especially if it's a guy like me capturing elements. Sometimes hundreds of shots. Depends on the project though.

 

[PatrickFaith]

Yep. Lots of PV glass and PL. But, you'd also be surprised how much "other" glass is used. Especially if it's a guy like me capturing elements. Sometimes hundreds of shots. Depends on the project though.

Phil, what do pro's use for the checker boards, I'm using just a junk board from a chess set ... do you guys make something your self (i don't see anything online for purchase) ?

 

[Phil Holland]

Phil, what do pro's use for the checker boards, I'm using just a junk board from a chess set ... do you guys make something your self (i don't see anything online for purchase) ?

I make them, we use another studio's, or we use PV's. The one I'm using the most is mounted on 1/2" gator board (so it doesn't warp). This one is 36x48". I have a few others. One is about 6'x4' and folds into 1' squares and is assembled on location. The larger targets well for high resolution/S35-ish cameras as well as film up to 65mm and can assist with correcting tricky lenses that have complex mustache distortion.

 

[PatrickFaith]

Thanks Phil ... man those are some sweet boards. I'm sure the VFX guys and the executive producers love ya.

 

[Phil Holland]

Thanks Phil ... man those are some sweet boards. I'm sure the VFX guys and the executive producers love ya.

Only when I save them money. Otherwise they just like me

 

[PatrickFaith]

I have been playing around with Anamorphics, but have realized how complex anamorphic testing can be. I mean a lot of lens problems are multiplied on anamorphics such as breathing and distortion. Some lenses actual change distortion as they breath, and a bunch of software is non-functional on 3-d compositing of anamorphics. So far, only the new anamorphics seem usable to me for vfx compositing (but I really have not compared very much). Also I don't know if it is a fundamental issue of the specific lens, or if LOMO type distortions are "fixable" lens alignment issues. Anyway, LMAO, my vote for a SALT IV is some type of anamorphic testing. It would probably need a bunch of subjective categories too. So far, I can see if I do a bad decision on a anamorphics lens choice, it could multiply by a factor of 2 the cost of compositing fx and practicals. Anamorphic lens testing has just turned out way beyond my brain (i was expecting it to be somewhat similar to spherical, and now realize it requires a major upgrade in cinematography skills that I do not have).

 

[Matt Ryan]

I have been playing around with Anamorphics, but have realized how complex anamorphic testing can be. I mean a lot of lens problems are multiplied on anamorphics such as breathing and distortion. Some lenses actual change distortion as they breath, and a bunch of software is non-functional on 3-d compositing of anamorphics. So far, only the new anamorphics seem usable to me for vfx compositing (but I really have not compared very much). Also I don't know if it is a fundamental issue of the specific lens, or if LOMO type distortions are "fixable" lens alignment issues. Anyway, LMAO, my vote for a SALT IV is some type of anamorphic testing. It would probably need a bunch of subjective categories too. So far, I can see if I do a bad decision on a anamorphics lens choice, it could multiply by a factor of 2 the cost of compositing fx and practicals. Anamorphic lens testing has just turned out way beyond my brain (i was expecting it to be somewhat similar to spherical, and now realize it requires a major upgrade in cinematography skills that I do not have).

Anamorphic will be coming. We have zooms, T2 Primes, and Anamorphics to do. Waiting till the newer anamorphic lenses come to market so we have all the available anamorphics to test (which is a lot!)

 

[Nick Morrison]

I like Nick's suggestion of staying with 35-80 mm primes for low budget vfx heavy movies.

Another thing to consider is using MEDIUM FORMAT glass, which even on Dragon would render almost distortion free imagery.

Because the image circle is so vast, and most distortion lives on the edges, shooting through the heart of the glass presumably gives you impeccably flat imagery.

This is probably why people usually notice an imperceptible "magic" to shooting Medium Format on S35.

 

[PatrickFaith]

Anamorphic will be coming. We have zooms, T2 Primes, and Anamorphics to do. Waiting till the newer anamorphic lenses come to market so we have all the available anamorphics to test (which is a lot!)

Awesome, I reference these posts atleast once a month and appreciate the work you guys are doing.

Another thing to consider is using MEDIUM FORMAT glass, which even on Dragon would render almost distortion free imagery.

Because the image circle is so vast, and most distortion lives on the edges, shooting through the heart of the glass presumably gives you impeccably flat imagery.

This is probably why people usually notice an imperceptible "magic" to shooting Medium Format on S35.

Awesome idea, also the 3D camera solves would be real nice. These anamorphics are super cool, yet I'm not sure they "tell" the story that I need ... while I think medium format's could (especially with the 6k). Do you have any thoughts on even how to hook up a medium format to a Epic/Dragon?

 

[Nick Morrison]

Awesome idea, also the 3D camera solves would be real nice. These anamorphics are super cool, yet I'm not sure they "tell" the story that I need ... while I think medium format's could (especially with the 6k). Do you have any thoughts on even how to hook up a medium format to a Epic/Dragon?

Patrick, I personally use Hasselblad lenses (I have a complete set from 50 to 250), which are relatively easy to get onto an EOS mount. You just need a simple adapter.

The large image circle of these lenses makes them ideal for tilt shifting also, and (after trying numerous adapters) I have decided that the MIREX adapter is by far the best. Impecable all metal German engineering.

PS, if you are interested in Hasselblads, look for the "F" series...they are shutterless and are often a stop faster than ordinary Hassy lenses.

 

[Mike Nagel]

One other thing which I find disconcerting about the CP.2's is that even though they have an EF mount alternative, there is actually an enormous amount of play when mounted on a Canon camera. This is really problematic as it defeats the whole purpose of a Cinema lens and creates image shift which resembles breathing. So even Zeiss needs to tighten up their tolerances with regards to this aspect.

Clayton, did u notice less breathing and image shift with the CP.2's on a PL mount system ?

Thanks.

- M

 

[Ryan Patrick O'Hara]

Another thing to consider is using MEDIUM FORMAT glass, which even on Dragon would render almost distortion free imagery.

Because the image circle is so vast, and most distortion lives on the edges, shooting through the heart of the glass presumably gives you impeccably flat imagery.

This is probably why people usually notice an imperceptible "magic" to shooting Medium Format on S35.

This is one of the many issues I have with larger image circle lenses... and it's not with the lenses, it's with the people make general assumptions as if all lenses are created equal.

This is not aimed at Nick, and as medium format lenses are usually made with great care while being of really good image quality, this argument I make is more or less for those who have little experience with optics and make this (or very similar) argument but with general run FF photography lenses, which Nick did not do.

I hear a few people make the argument that because a FF lens has a very large image circle, just using the center is the 'sweet spot' and thus is better. Although this is true in theory and certainly true when just discussing a particular lens.... it does not always factor in that the image standards of some lenses throughout the circle are less than, say other lenses from edge to edge. So although this statement can be true at times, it's a bad idea to assume it all the time, which I find many FF kiddos doing. Distortion and other optical issues do not simply appear on some arbitrary image circle edge to center ratio. Just because the image circle is wider, doesn't mean the center is sweeter. A poor FF lens can have issues well inside 5k or APS-C territory, despite there being lots of bleed-off room in the FF image circle. Distortion, as in barrel or pincushion is hardly dependent on image circle size, in the manner that a FF lens center will have less distortion than an equivalent focal length lens but one that only covers 30-31mm diagonally. Most higher end cinema, especially the wider lenses, don't have much bleed-off room in the image circle from center to intended usable edge, however, despite the center being better than the edges, the amount of degradation is usually much less than other lenses with much larger image circles. It's just how they are made.

So while it can sometimes be true that a really well made FF lens will have a very nice center, perhaps better in some or many ways to other high quality cine lenses, it is not a general statement that FF centers are always sweet and healthy from optical issues. I find many shortcuts are made with photography lenses, which little effect the photographer but make cinematography difficult, and these issues do tend to live just as much, if not more, in the optics than they do with the now varying range of quality we find in marketed 'cine-lenses'; a word who's descriptive nature of lens class is widening at great speed.

So yes, the center of FF lenses is better than the edge, especially with photography lenses lacking some of the consistency criteria some higher end cinema manufacturers aim for, however the misinformation I am fighting is that we assume the center of a FF is always better than a lens with an image circle that only covers the intended format. What I am saying is that assumption is not an always true or false statement and testing is always encouraged. Sometimes it is true, but sometimes it's not.

My morning rant.

 

[PatrickFaith]

Patrick, I personally use Hasselblad lenses (I have a complete set from 50 to 250), which are relatively easy to get onto an EOS mount. You just need a simple adapter.

The large image circle of these lenses makes them ideal for tilt shifting also, and (after trying numerous adapters) I have decided that the MIREX adapter is by far the best. Impecable all metal German engineering.

PS, if you are interested in Hasselblads, look for the "F" series...they are shutterless and are often a stop faster than ordinary Hassy lenses.

This sounds great for the silky shallow depth of field black and white photography i like to do with medium format, I have been searching for a similar lens system that can do that with RED. I really like your idea, only thing is that I do a lot of vfx on the frames (specially 3d solves), and if I did that I am pretty sure everything in nuke would have to go through a lens corrector before i did any camera solves or vfx merging (and I am late on almost everything as it is right now). I also don't have a collimator handy, which I would want to use before every major shoot(i really need to get some lens testing gear, it's just so hard to justify, 27 years back i had all that stuff ... but not now ). I do follow the medium format groups and clubs on the camera side, but haven't found a good source for medium format cinematography.

 

[Jacek Zakowicz]

I'm not sure what you mean-50mm lens will have exactly the same depth of field at the same stop and distance in 16mm, APSC, S35, FF or medium format. Same goes for every FL lens be it 20mm or 150mm

 

[PatrickFaith]

I'm not sure what you mean-50mm lens will have exactly the same depth of field at the same stop and distance in 16mm, APSC, S35, FF or medium format. Same goes for every FL lens be it 20mm or 150mm

Well your right at some level, and I didn't mean for this to get into a medium format DOF thing (since there are more then three variables that change on medium format lenses). In general though, the closer the object is to the lens that your focusing on, the shallower is the DOF(i.e. things focused on far away have a very large depth of field of many feet at the same T stop, as things focused on a foot away may have only a few inches of dof at the same t stop). So with a larger sensor, you need to get closer to a object your focusing on to have it framed the same. i.e. if your doing a head shot on a s35 50mm lens you stand farther away from the subject then you would with a medium format. So your completely right, if the 35mm lense has a COF that can cover the larger format(i.e. sensor size), there is no change. But as a super nice "feature" of the dragon, since it will have a slightly larger sensor, then a picture framed exactly the same with a 35mm at 6k will have a shallower dof then if it was shot at 5k with the same framing(since at 6k you needed to get closer to the object to get the same framing). This is why with the dragon sensor, I will always be shooting at 6k on the dragon, since I like shallower dof. Hopefully long term, the sensor size will even get bigger, which then makes sense to invest in lenses with large COF diameter limits.

 

[Jacek Zakowicz]

With all due respect Patrick- the Dragon sensor is onlty 1.5mm wider than the MX. That's 5% gain in the angle of view. Do you really think that anyone will notice the "medium format effect" with so little change?

 

[PatrickFaith]

With all due respect Patrick- the Dragon sensor is onlty 1.5mm wider than the MX. That's 5% gain in the angle of view. Do you really think that anyone will notice the "medium format effect" with so little change?

I hadn't done the math but that sounds right for the dof change between epic 5k to dragon 6k(this stuff get's non-linear on closeups ... it hurts my brain). On the dragon though, difference between shooting at 5k and 6k will have a bigger change in feel (dof) if framing identically, and I am on the hunt for my middle 2014 to 2016 lens set - which will use the dragon plus what will come after it. Trying to get a total medium format feel on the 6k will be tricky(it is possible using some old school bellows optics with intermediary focal plane - but vfx would be a nightmare and it looses stops). I wish there was something like the Leica 50mm f/0.95 Noctilux-M but at 35mm and at a price about half of the noctulix (also since i don't need the fanatical resolution of the noctulix ... i could deal with something a bit softer and cheaper as long as the optics didn't have any aberrations). I also have moved to computer controlled focus (integrated to a 3 axis motion control system), so I worry about breathing more then i did in the old days.