Why Bigger is Better (but not much else) S35mm vs. Vistavision Comparison

[Misha Engel]

99% that cooke says 30mm in diameter for their set. They dont mention for individual lenses do they? Normally a long lens has much greater coverage than needed. It´s always the wide end it gets problematic.

Maybe but 30 mm vs. 35.42 mm is a big gap

 

[Björn Benckert]

Maybe but 30 mm vs. 35.42 mm is a big gap

Nope its not it would be quite normal. Also its not full 16:9 but more like 2:1 so then it fits better.

 

[Jacek Zakowicz]

That makes more sense. 2:1 aspect ratio could possibly bring the horizontal pixel count on the Alexa 65 sensor close enough to 5K to call it 5K. Which ties in nicely with the image clarity that Bjorn is seeing...

 

[Misha Engel]

2:1 is 35,42 mm on alexa65 3840x1920

 

[Les Hillis]

To me at least it looks like a barrel distortion makes the image look more "fisheye-ish" for the lack of better term ;-) Maybe picking lenses with pin cushion distortion would create more of a "tele" look... just a thought...

...theoretically a lens designer can use this knowledge and design any lens as a large coverage lens...

Thanks.

Wondering too whether the difference is the result of the kind of lens distortion that can be corrected in post.

 

[Jacek Zakowicz]

2:1 is 35,42 mm on alexa65 3840x1920

The Canons cover 44mm circle being full frame. I think we can go off that maybe up to 50mm lens. Then the Cookes kick in with the tele lenses having extra coverage. They usually take care of all that during the prep...
The question is: how many horizontal pixels in 2:1 aspect ratio on Alexa 65 sensor inside 44mm circle?

 

[Misha Engel]

Thanks.

Wondering too whether the difference is the result of the kind of lens distortion that can be corrected in post.

Almost all except chroma and sharpness, Sony is pretty good in that in camera.

 

[Misha Engel]

The Canons cover 44mm circle being full frame. I think we can go off that maybe up to 50mm. Then the Cookes kick in with the tele lenses having extra coverage. They usually take care of all that during the prep...
The question is: how many horizontal pixels in 2:1 aspect ratio on Alexa 65 sensor inside 44mm circle?

4770x2385

 

[Björn Benckert]

If the canons are full frame you could probably eak out those last 230 pixels to get to full 5k. Or they simply crop of the sad corners in post. Again it´s on the 20mm and below you normally hit the edge first.

 

[Jacek Zakowicz]

Thanks.

Wondering too whether the difference is the result of the kind of lens distortion that can be corrected in post.

Distortion as well as other aberrations have been corrected via software for a while as you probably know. Canon EXIF data is one standard, Sony has their real time lens correction in their FSx cameras while using Sony lenses...
Oops Misha beat me to it ;-)

 

[Phil Holland]

I propose that someone test a 50mm on 8K compared to 25mm on 4K at the same position and show that the geometry changes (because of the lens, of course). And then find the entrance pupil/no-parallax point of both lenses. Keep the tripod in the exact same position but slide the camera forward or back on your dovetails to line up the entrance pupils, demonstrating that the two images can be matched nearly identically for geometry representation.

You'd also have to shoot distortion charts for each lens to eliminate that variable as well.

In the end it always comes down to the optics themselves. If you like the way a certain lens looks and it's only achievable with a large sensor behind it, go for it. If you like the way a different lens looks and it only projects a smaller image circle, go for it.

I'm with David and the others.

My first post on the matter was in regards to Optically Matching Multiple Format Sizes here:

You sir are hitting the nail on the head.




What's interesting to me is Yedlin's tests clearly show some of these variations and nuance, but are not discussed for whatever reason. I've ran into him twice recently, but next time I hope we can chat for a few minutes about all this stuff, which will likely lead to hours of fun

On the topic, Panavision actually explains this in a similar fashion in a nice concise 5 minute video with examples showing some of the differences:

Link for Ipad

I think my next post, as this level of ping pong is getting bizarro, is going to discuss the actual differences that people are either ignoring, not seeing, or defining as "negligible" (I'll let that last one go).

The thing that disturbs me is the differences and most aren't "seeing them", but they are also "not seeing them".

Dipping out for now, I don't want to take away from Mark's test, which is well shot and executed for what he is illustrating.

 

[Jacek Zakowicz]

4770x2385

There you go- mystery solved ;-)

 

[Jacek Zakowicz]

My first post on the matter was in regards to Optically Matching Multiple Format Sizes here:



I think my next post, as this level of ping pong is getting bizarro, is going to discuss the actual differences that people are either ignoring, not seeing, or defining as "negligible" (I'll let that last one go).

The thing that disturbs me is the differences and most aren't "seeing them", but they are also "not seeing them".

Dipping out for now, I don't want to take away from Mark's test, which is well shot and executed for what he is illustrating.

I must be the only one that finds the pictures and the video contradictory. In fact the video is a bit of a mishmash.. Unlike Phil's posts I must add.

 

[Björn Benckert]

Phil could you please explain the difference in judder, perspective and different panning speeded needed and also what you base your statement "moving the camera trough 3D space" on.


As I see it all those comments relate to FOV not to sensor size. None of it matters if the fov is set equal.


Panning judder is due to things moving at "judder speed" across the frame. IF you pan with a wide lens it takes more of a pan in degrees to make something move from one side of frame to the other, if you got a longer lens you dont need to pan as much to have something coming in from one side and going out on the other side. FOV is then, again, key. Sensor size used, not so much.

Then you got judder or sensor redraw issues, the more pixel lines you use normally the slower redraw. But thats not relevant to the size of sensor. Helium for example has as much or more judder / skew when panning due to slow sensor redraw than dragon even though it has a smaller sensor.

Normally the rule of thumb (actually also depening a lot on size of the viewing screen) you should not pan faster than having one thing entering frame on one side and leaving frame on the other faster than in 8 seconds if playback and recording is normal 24-25p. If framerate for some odd installation is double then you can cut the time in half.


The Panavision clip... not sure what to say about it.

 

[Misha Engel]

The Panavision clip... not sure what to say about it.

Well maybe Panavision can't, but lots of others can

40.96x21.6 46 mm F4 (DXL2 vista vision)
24x12.65625 26.953 mm F2.344 or the same lens with a 0.586 speedbooster.

 

[Michael Hastings]

snip:
But I also contend that this difference exhibited is not significant. Mainly what people initially perceive with larger format photography is (usually) increased resolution, lower noise/grain, and typically lower depth of field in most live action moving images, not this slight difference in "roundness" or compression/expansion. Also, if we're talking about the difference between Super-35 and Full-Frame / VistaVision, it's not a big enough jump to create dramatic perceptual differences compared to, let's say, Super-35 and IMAX.

This has been my impression throughout - my Epic-W Helium sensor is also 8K and Red VV is only 37% larger - and I doubt if that even achieves the JND (Just noticeable difference) that is normally used in perceptual tests.

I hesitated to say it before in case it seemed like trolling (and I hate to doubt Phil because he is usually right) but I think any difference beyond the marginal differences in DR, noise and sensitivity, is mostly marketing hype.

Marketing - whether it is for a product or for an individual - is a lot easier if you can point to something specific to differentiate. Red monstro VV is $79K vs $29K for Epic-W (and many of us got Epic-W for $19.5K) so I'm sure it is easier to sell cameras or get higher rental rates if you can push the idea that there is something special about "roundness" or compression/expansion.

Just last week I had the marketing discussion with my partner about shooting a project with my Epic-W vs his GH5 on a project where we would be traveling and needing to shoot quickly. He is an excellent cinematographer and has produced really beautiful stuff with his GH4, but pointed out that while for the most part if we exposed correctly the end result wouldn't be noticeably better on the 1080p or less resolution deliveries we anticipate - clients tend to feel better about spending $30K or more on a shoot where we show up with the latest greatest 8K professional looking camera than they would when we show up with a little camera like their nephew uses.



As someone that spent most of his career shooting on 2/3" sensors (in fact I was on a test team for Sony's first 3 chip (vs. tubes) camera) I was happy to have the option of a shallow DOF S35 sensor on my RedOne (in fact I have often said that Red probably would have sold just as many R1s if it had been S35 1080P - whatever sales they would have lost in not having 4K they would have made up selling to those that didn't buy because they didn't want to deal with the data rates required.) But it seems like beyond S35 it is diminishing returns - the vast majority of classic movies before the eighties were shot on the even smaller Academy size and through the 90s to today the DOF of S35 has been shallow enough. If that is true then bigger sensors are often more trouble than they are worth - given that to get equivalent DOF and reasonable focus pulling on the bigger sensors for most situations requires twice the light. And we shouldn't forget that one of the big parts of Citizen Kane - often considered the greatest movie ever made - was the deep focus.

But I'm also aware of the advantages in many situation for smaller sensors. Smaller sensors make it much easier and more practical to make reasonable size 10x to 20x zoom lenses, or fast primes. Setting aside the Citizen Kane example, there are many types of shooting situations - long lens nature shots, live sports, macro, and my particular area - underwater where greater DOF is important.

I have had to explain that because in U/W shooting we are pretty much stuck with dome ports for the best option for partial underwater optical correction (without getting far into it, a fully corrected underwater optic for even S35 is huge) smaller sensors help us because the dome creates a curved image so the edges tend to be out of focus and have significant aberrations so a smaller sensor gives greater DOF and corrects a lot of the problems.

 

[Björn Benckert]

This has been my impression throughout - my Epic-W Helium sensor is also 8K and Red VV is only 37% larger - and I doubt if that even achieves the JND (Just noticeable difference) that is normally used in perceptual tests.

I hesitated to say it before in case it seemed like trolling (and I hate to doubt Phil because he is usually right) but I think any difference beyond the marginal differences in DR, noise and sensitivity, is mostly marketing hype.

Marketing - whether it is for a product or for an individual - is a lot easier if you can point to something specific to differentiate. Red monstro VV is $79K vs $29K for Epic-W (and many of us got Epic-W for $19.5K) so I'm sure it is easier to sell or get higher rental rates if you can push the idea that there is something special about "roundness" or compression/expansion.

Just last week I had the marketing discussion with my partner about shooting a project with my Epic-W vs his GH5 on a project where we would be traveling and needing to shoot quickly. He is an excellent cinematographer and has produced really beautiful stuff with his GH4, but pointed out that while for the most part if we exposed correctly the end result wouldn't be noticeably better on the 1080p or less resolution deliveries we anticipate - clients tend to feel better about spending $30K or more on a shoot where we show up with the latest greatest 8K camera than they would when we show up with a little camera like their nephew uses.



As someone that spent most of his career shooting on 2/3" sensors (in fact I was on a test team for Sony's first 3 chip (vs. tubes) camera) I was happy to have the option of a shallow DOF S35 sensor on my RedOne (in fact I have often said that Red probably would have sold just as many R1s if it had been S35 1080P - whatever sales they would have lost in not having 4K they would have made up selling to those that didn't buy because they didn't want to deal with the data rates required.) But it seems like beyond S35 it is diminishing returns - the vast majority of classic movies before the eighties were shot on the even smaller Academy size and through the 90s to today the DOF of S35 has been shallow enough. If that is true then bigger sensors are often more trouble than they are worth - given that to get equivalent DOF and reasonable focus pulling on the bigger sensors for most situations requires twice the light. And we shouldn't forget that one of the big parts of Citizen Kane - often considered the greatest movie ever made - was the deep focus.

But I'm also aware of the advantages in many situation for smaller sensors. Smaller sensors make it much easier and more practical to make reasonable size 10x to 20x zoom lenses, or fast primes. Setting aside the Citizen Kane example, there are many types of shooting situations - long lens nature shots, live sports, and my particular area - underwater.

I have had to explain that because in U/W shooting we are pretty much stuck with dome ports for the best option for partial underwater optical correction (without getting far into it, a fully corrected underwater optic for even S35 is huge) smaller sensors help us because the dome creates a curved image so the edges tend to be out of focus and have significant aberrations so a smaller sensor gives greater DOF and corrects a lot of the problems.

I agree to a lot of the above. But even if you gave me two heliums I would not trade you my monstro... There is differences in noise and colors but monstro, just as when 5D came out, is a one of a kind digital camera that fully utilise FF glas, and there is a lot of nice FF glas out there. And for high speed and anamorphics helium and dragon is simply a bit small... Not for all anamorphics, most of them where made for academy, but my Baltar Cinemascopes where actually designed for full frame S35. which is about the same height as monstro. And even though I have not done any proper side by side tests. Monstro 6k, to me looks far better than what I manage to eak out of my helium at 8k, Sharpness color rendition, color fringing and simply better pixels at higher bit rate per pixel.

But yes, Panavision sales talk, at 56 min they are out in the deep end of the pool... Moving camera around change the prespecitive, "roundness of faces" and such, changing sensor size does not, swapping focal length does not.

https://vimeo.com/248235757

 

[Aaron Lochert]

But yes, Panavision sales talk, at 56 seconds they are out in the deep end of the pool... Moving camera around change the prespecitive, "roundness of faces" and such, changing sensor size does not, swapping focal length does not.

https://vimeo.com/248235757

You mean 56 minutes, right? The point when Michael says the words "same lens" and then Dan says the reason for the change is because of using a longer focal length lens?

 

[Björn Benckert]

You mean 56 minutes, right? The point when Michael says the words "same lens" and then Dan says the reason for the change is because of using a longer focal length lens?

Sorry about that. Yes where He claims the field of view is the same even though the camera jumps back and forth. Things that change due to perspective does not really have anything to do with sensor size. Again same could be achieved with different focal lengths and different distance to target.

 

[Hrvoje Simic]

To me at least it looks like a barrel distortion makes the image look more "fisheye-ish" for the lack of better term ;-) Maybe picking lenses with pin cushion distortion would create more of a "tele" look... just a thought...

In David's shots central part general bulge of the image is barrel distortion.
And monitor and table geometry are not.