S35 Anamorphic Lenses with Alexa LF

[Dominik Bauch]

Obviously as the Alexa LF is a larger sensor the depth of field will be shallower vs a S35 Alexa but.... if I'm shooting anamorphic 2:1 on the LF (2880 x 2880) using a S35 anamorphic lens and I compare the image to shooting 2:1 on the Alexa mini (2160 x 2160) with the exact same lens / T-stop, I would imagine that there would be negligible depth of field difference between the 2 images, the Alexa LF image would have a slightly wider FOV than the Alexa Mini image. Same for a 2.39:1 image, Alexa mini would have a slightly smaller FOV.

Is this correct or am I missing something, I'm primarily interested in whether there would be an appreciably shallower DOF when shooting Anamorphic on an LF vs a regular S35 Alexa. The difference in pixels used seems so small that I can't imagine it would make a visible difference. I can totally appreciate that a Full frame spherical lens does have a shallower DOF than an equivalent S35 spherical lens given the greater difference in sensor size being used vs the example above.

Thanks in advance for any insight.

[David Mullen ASC]

If it’s a 1.3X difference then it’s a 1.3-stop DOF difference if you match FOV.

[Mark Kenfield]

What David said.

Some S35mm anamorphics will give you a little more leeway though. The Master Anamorphics for example, can cover the full height of the LF sensor down to the 28mm(!), you won’t get the full width of the sensor, but you can get a full-height 2:1 extraction. 

So 3096x3096, or a 1.43x difference (which isn’t insignificant).

[Dominik Bauch]

But if we take Netflix acceptable resolution of 3148 x 2636 for 2x anamorphic, then the alexa mini open gate 4:3 mode spits out 3424 x 2202, so the width exceeds that specification but the height is 400+ pixels short. In that instance surely the appreciable DOF difference between an LF and a regular Alexa is going to be virtually zero as the photosites are the same size and distance apart.

Also the Cooke Anamorphics have an image circle larger than the master anamorphics; 33.54mm vs 29.26mm. So will the Cooke Anamorphics cover the LF's 3148 x 2636? I'm guessing yes, do they also cover the full height of the LF sensor?

[David Mullen ASC]

I would guess that a typical anamorphic lens would cover a 24mm height since most cover a 24mm width.

[Tyler Purcell]

If you're shooting a show FOR Netflix, IE they're paying for it, then yes the camera resolution needs to be 4k or higher. They will actually check your source files. 

If you're shooting a show that may get picked up by Netflix, it doesn't matter. As long as your master is 2k or above, they're fine. 

Personally, I would shoot 1.33x anamorphics on the LF. Thus, you can use the full 16x9 imager and get the best of both worlds. The anamorphic lens flairs and great large format look. You also retain your full imagers width, thus being 4k compliant. 

The only downside is your choice of lenses, Panavision and Hawk are the only companies I know of making the proper lenses. Panavision would be the "right" 1.5x anamorphic for VistaVision cameras like the LF. The Hawks are 1.33x, so only a slight crop top and bottom to get 2.40:1.

[Mark Kenfield]

7 hours ago, David Mullen ASC said:

I would guess that a typical anamorphic lens would cover a 24mm height since most cover a 24mm width.

In hindsight, that seems pretty bloody obvious all of a sudden! ?

[Miguel Angel]

I shot this last year for a friend's company on the LF and the Master Anamorphics and Supreme Primes. 

If I remember correctly it was the 28mm for the anamorphic part. 

 

[Dom Jaeger]

7 hours ago, Tyler Purcell said:

If you're shooting a show FOR Netflix, IE they're paying for it, then yes the camera resolution needs to be 4k or higher. They will actually check your source files. 

Was this even a question? 

In fact, no, you don’t need 4K for anamorphic. Arri released a white paper about anamorphic on the LF, and as the OP Dominic stated, for 2:1 you only need 2880 x 2880, and for 2.39:1 you need 3148 x 2636. The magic number is 8.29 million photo sites.

https://www.arri.com/resource/blob/41434/2f8d035d9729399dcced26f4cb24ac50/download-alexa-lf-anamorphic-whitepaper-data.pdf

This can also be confirmed by reading the Netflix approved camera guidelines:

https://partnerhelp.netflixstudios.com/hc/en-us/articles/360000579527-Cameras-and-Image-Capture

If you’re trying to give advice, it helps to have done more research than the person you’re advising.

7 hours ago, Tyler Purcell said:

Personally, I would shoot 1.33x anamorphics on the LF. Thus, you can use the full 16x9 imager and get the best of both worlds. The anamorphic lens flairs and great large format look. You also retain your full imagers width, thus being 4k compliant. 

The “full imager” isn’t 16:9 (1.78:1). Open Gate mode is 4448 x 3096 which is closer to 1.44:1.

7 hours ago, Tyler Purcell said:

The only downside is your choice of lenses, Panavision and Hawk are the only companies I know of making the proper lenses.

As the Arri white paper states, all Master Anamorphics will cover the 2:1 extraction 2880 x 2880 area, and from 40mm up they cover the 2.39:1 extraction 3148 x 2146 area. Cooke anamorphics with their even larger image circle could possibly cover both. Other S35 anamorphics may also cover, you’d need to test, but you’re not limited to just full frame anamorphics in order to comply with a UHD mandate.

However there are full frame anamorphics now made by Hawk, Panavision, Cooke, P&S Technik and Servicevision, as documented in the white paper.

7 hours ago, Tyler Purcell said:

 Panavision would be the "right" 1.5x anamorphic for VistaVision cameras like the LF. The Hawks are 1.33x, so only a slight crop top and bottom to get 2.40:1.

Panavision don’t make 1.5x anamorphics. The full frame ones are either 1.25x or 1.65x.

Full frame Hawk 1.33x anamorphics on LF Open Gate would give about a 1.9:1 frame and need considerable cropping to achieve 2.39:1. Good option for 2:1 though.

Nice post, just needed the “I didn’t state any facts” caveat attached. ?

[Dom Jaeger]

10 hours ago, Dominik Bauch said:

Also the Cooke Anamorphics have an image circle larger than the master anamorphics; 33.54mm vs 29.26mm. So will the Cooke Anamorphics cover the LF's 3148 x 2636? I'm guessing yes, do they also cover the full height of the LF sensor?

The actual image circle required to cover 2880 x 2880 is 33.6mm, which apparently Master Anamorphics cover.

For 3148 x 2636 it’s 33.87mm which according to Arri is a smidge too large for the 35mm MA but ok for the 40mm. 

Often the published image circle of lenses is conservative, and it will vary with focal length, so the 33.54mm that Cooke lists for their S35 anamorphics will be the smallest measured. There can also be differences between the image circle (where an acceptably sharp image is formed) and the illumination circle, where image quality may drop off but there is still coverage.

Ideally you need to test them. I haven’t checked the Cooke anamorphic image circles yet myself.

[Tyler Purcell]

5 hours ago, Dom Jaeger said:

In fact, no, you don’t need 4K for anamorphic. Arri released a white paper about anamorphic on the LF, and as the OP Dominic stated, for 2:1 you only need 2880 x 2880, and for 2.39:1 you need 3148 x 2636. The magic number is 8.29 million photo sites.

Ahh interesting. In the non-anamorphic but still 2.40:1 world, they require 4k. I do Netflix deliverables all the time. I can't believe they would not require 4k for anamorphic, it seems counter intuitive. 

5 hours ago, Dom Jaeger said:

The “full imager” isn’t 16:9 (1.78:1). Open Gate mode is 4448 x 3096 which is closer to 1.44:1.

Ah ok, so the reason for the 2.8k is to make non-full frame anamorphic lenses work. I thought the LF was a standard full frame imager. I only checked the specs when it came out and guessed on the aspect ratio based on the resolution number. Now the specs are more easy to access and they finally show 1.44:1, which is an interesting choice. So I'm sorry for the confusion, I'm use to full frame imagers being 1.55:1. 

So nobody makes the proper anamorphic lens for a standard full frame 1.55:1 imager and get 2.40:1 out of it? 

Edited November 13, 2019 by Tyler Purcell

[David Mullen ASC]

The 1.6X Panavision lenses are meant for that, similar to how 1.5X anamorphic lenses were used on 8-perf 35mm Technirama to get a 2.35 image more or less.

[David Mullen ASC]

Dom, is there are published data to confirm that the Panavision Ultra Vista lenses are 1.66X, not 1.6X, and that the Hawks are 1.33X and not 1.3X? I’m working on a written document about non-standard squeeze ratios.

[Dom Jaeger]

Hi David,

not really published data, but the Panavision website describes the Ultra Vistas as having a 1.65x squeeze:

https://www.panavision.com/products/ultra-vista

and various large format cameras have the 1.65x de-squeeze monitoring option (rather than a 1.6x de-squeeze). 

The Arri LF anamorphic white paper I linked to earlier for example lists all the monitoring de-squeeze ratios that the LF has built in:

 

Some older promotional material on the web lists 1.6x, but I would assume that was an estimate given during the early design stages and then repeated by others, as happens on the internet.

Hawks are 1.3x, as described on their website and here. I mistakenly wrote 1.33x earlier because Tyler had written it and I inexplicably didn't stop to question it. I guess because 1.33 is an aspect ratio it sticks in your mind.

[Dom Jaeger]

The Hawk65s are a bit hard to find info on, but there is a brochure in the download section that lists the squeeze as 1.3x.

https://www.vantagefilm.com/en/downloads

[David Mullen ASC]

Thanks Dom!

[Tyler Purcell]

22 hours ago, David Mullen ASC said:

The 1.6X Panavision lenses are meant for that, similar to how 1.5X anamorphic lenses were used on 8-perf 35mm Technirama to get a 2.35 image more or less.

Got ya, I get it now. I totally forgot Technirama was 2.20:1, not 2.35:1. 

[David Mullen ASC]

No, Technirama was meant originally for making reductions to 2.35 CinemaScope 35mm, it just a few years later that they realized that they could blow it up to 65/70, whether or not with black borders to maintain 2.35 on a 2.20 70mm print.

Keep in mind that these formats did not have to use 100% of the full aperture negative area.  VistaVision was 1.50 : 1, but usually composed for reduction printing to 1.66 or 1.85 35mm. 

[David Mullen ASC]

It's odd that Ultra Vista lenses are 1.65X instead of 1.6X since 2.4 divided by 1.5 = 1.6.  But it doesn't really matter, what's a few pixels here and there.  Also, the Monstro sensor isn't 1.5 : 1 anyway, it's 1.89 : 1.

[Tyler Purcell]

1 hour ago, David Mullen ASC said:

No, Technirama was meant originally for making reductions to 2.35 CinemaScope 35mm, it just a few years later that they realized that they could blow it up to 65/70, whether or not with black borders to maintain 2.35 on a 2.20 70mm print.

Keep in mind that these formats did not have to use 100% of the full aperture negative area.  VistaVision was 1.50 : 1, but usually composed for reduction printing to 1.66 or 1.85 35mm. 

Ahh very interesting. Was it just the super technirama 70 shows that were 2.20:1? I'll have to re-examine the remaster of the technirama shows, but I could have sworn they were 2.20:1 on video. 

[David Mullen ASC]

I'm not sure there was a standard but probably most Technirama movies opted to not letterbox the 70mm print and just trim 2.35 to 2.20 instead. On blu-ray, it's been mixed too, probably depending on if they had a 35mm scope version to transfer from versus a 70mm version.  "Spartacus" on blu-ray is 2.20 but "El Cid, "King of Kings" and "The Vikings" are 2.35.

[David Mullen ASC]

http://www.widescreenmuseum.com/widescreen/wingtr3.htm

Some history here. Interesting to note that the negative area was 2.25 : 1 unsqueezed so making 2.35 35mm versions or 2.20 70mm versions were not that hard.  2.25 makes sense... 1.50 : 1 negative with a 1.5X squeeze equals 2.25 : 1.

[Tyler Purcell]

1 hour ago, David Mullen ASC said:

http://www.widescreenmuseum.com/widescreen/wingtr3.htm

Some history here. Interesting to note that the negative area was 2.25 : 1 unsqueezed so making 2.35 35mm versions or 2.20 70mm versions were not that hard.  2.25 makes sense... 1.50 : 1 negative with a 1.5X squeeze equals 2.25 : 1.

Ahh so that does make more sense. What an odd intermediary resolution! 

[Manu Delpech]

Speaking of S35 on Alexa LF, there's some dope coverage on Sasaki adaptating C-series anamorphics to cover the full Alexa LF sensor for Phedon Papamichael on Ford V Ferrari.