What aspect ratio/resoloutions for 1.5x

[Lucian]

Hey guys,

 

can anyone tell me what the ideal shooting aspect ratio and resolutions would be for getting proper scope (2.39:1  and 2.35:1) from a 1.5x adapter?

 

 

Thanks in advance.

 

[Sean Cunningham]

V * AR * 1/SQ

 

V= Vertical Resolution

AR = Aspect Ratio (flat)

SQ = Squeeze Factor

 

So, for instance, if your target was 1080 lines of vertical resolution you would multiply that by 2.35 to get 2538.  That would be your target, flat, widescreen image (2538x1080).  To get to the squeezed horizontal resolution you multiply this by the reciprocal of the anamorphic lens compression ratio, 1/1.5 (use full precision on calculator) so that's 1692.

 

1692 x 1080

 

Knowing this squeezed aspect ratio makes it theoretically easier to arrive at a shorter calculation but you'll find that the mix of single, double and greater decimal precision at different steps can alter the result by several pixels, introducing quite a bit of slop.  

 

For instance, the above resolution works forward and backward through the original calculation, proving it's correct.  That's not the case, however, if you were to use the aspect ratio of the frame we know to be correct.  If you were to use the aspect ratio of the squeezed frame and the common accuracy of only two digits of floating point precision you would get the following:

 

1692 / 1080 = 1.57

 

...a 1.57:1 aspect ratio for the squeezed frame.  But that aspect ratio does not give you correct results, even for our starter 1080 lines of vertical resolution example:

 

1080 * 1.57 = 1696    ...and...   1696 * 1.5 = 2544   ...but...  2544 / 1080 = 2.36

 

Not a huge deal but it's something to be aware of, particularly if you're doing something for a client or for eventual DCP or any other application where "close enough" isn't good enough.  

 

When you really want to know use as much real information as you can like actual pixels, known compression ratios, aperture measurements, etc. and stay away from aspect ratio and crop factor short hand.

[Lucian]

Thanks for the reply. I think I follow you, but it might be a little technical for me. Going to have to read it about 10 times :P

 

 

I'm curious because Digital Bolex  have been asking what resoloutions/aspect ratios people want included for anamorphic shooting. They are doing a 4:3 mode for 2x anamorphic shooters, and have been very responsive to community requests. It would be amazing to shoot with the iscorama (1.5x) and not have to dick around with crop in post, potentially fudging up your framing.

[Gábor Ember]

I would start calculating from a different point but the end results are the same:

 

target aspect / squeeze x vertical resolution = horizontal resolution

 

2.35 / 1.5 x  1080 = 1692  (1692x1080)

 

2.35 / 1.5 x 720 = 1128     (1128x720)

 

2.39 / 1.5 x 1080 = 1720.8 (1720x1080)

 

2.39 / 1.5 x 720 = 1147.2 (1148x720)

 

This way you have the least amount of rounding errors.

[Sean Cunningham]

As long as you can get the same results going forward and backward through the calculations, yeah, you're good.

 

You rounded the wrong way in your 2.39:1 examples, however.  1720.8 would round to 1721 and 1147.2 would round to 1147.  It likely doesn't matter to the image but if your desired intent is to reduce "rounding errors" it's counter-productive to add them yourself.

[Rob Bannister]

Sorry if Im not following either. Is this all assuming you are working with a 16X9 or a 4X3, or 3X2 sensor readout?

[Gábor Ember]

As long as you can get the same results going forward and backward through the calculations, yeah, you're good.

 

You rounded the wrong way in your 2.39:1 examples, however.  1720.8 would round to 1721 and 1147.2 would round to 1147.  It likely doesn't matter to the image but if your desired intent is to reduce "rounding errors" it's counter-productive to add them yourself.

 

I'm not rounding 1.5666 to 1.57 in the middle of the calculation, that is what I meant. Also I would not use odd numbers as resolution, that is why I got those end numbers but you are right.