Jump to content




Photo

In-camera MATTE for CINEMASCOPE look - lose resolution?

cinemascope 2.4 resolution matte Sony F5 letterbox aspect ratio native

  • Please log in to reply
15 replies to this topic

#1 Stefano Stroppa

Stefano Stroppa
  • Basic Members
  • PipPip
  • 35 posts
  • Cinematographer
  • London

Posted 23 April 2015 - 03:15 PM

Shooting with Sony F5, the idea is to have a final film with a cinemascope look once it'll be projected.

I know I can't have the choice but to shoot with the native aspect ration of 17:9 of the F5, I'll so put the letterboxing markers (2.4) that crops the native aspect ratio.

 

My question is: do I lose lots of resolution once I put 2.4 mattes on the 17:9 native to create the cinemascope look? What could be the solutions? If I shoot 4K external could be a better compromise to the loss of resolution? Since I'm recording with an higher resolution then the internal 2K? 

 

(When I set the mattes in the camera, do I record already with them? or the markers are just guide lines that help me framing knowing what would be the cinemascope look at the end?)

 

If I shoot with Red Epic (Dragon sensor), what could be the best aspect ratio settings for a final cinemascope look? 6 wide, or do I crop later with the mattes?

I can't use any anamorphic so I can't take this possibility into account.

 

 

Thanks a lot for any help! 


  • 0




#2 David Mullen ASC

David Mullen ASC
  • Sustaining Members
  • 18788 posts
  • Cinematographer
  • Los Angeles

Posted 23 April 2015 - 08:11 PM

Cropping an image to make it 2.40 doesn't mean you are losing resolution necessarily.

 

To simplify things, imagine you shoot with a camera that records an RGB file format that is 2048 x 1152 (16x9) and you need to crop it to 2048 x 858 for a 2K DCP for 2.40 projection.  So where is the resolution loss (ignoring whether that camera records a good 2K image or not)?  You can only project 2048 x 885, so that's the area you extracted from 2048 x 1152 to get that aspect ratio.

 

Same goes with letterboxing a 1920 x 1080 HD image.  If you compare the full 1920 x 1080 image to the letterboxed one on the same HD monitor, the letterboxed image is just smaller vertically but it isn't softer, it's the same quality image just that the top & bottom are covered with black bars.  But an object in the image has the same detail in it, the same number of pixels devoted to it.

 

However, let's say you compare cropping a 2048 x 1152 image to 2048 x 858 versus shooting with a 2X anamorphic lens and recording an image that is 2048 x 1707 with a 2X squeeze.  Now your 2.40 image area is made up of a larger number of total pixels.  You'd have to downsample 1707 to 858 to get an unsqueezed image for 2K DCP scope projection but you may have more resolution (vertically at least) compared to the image shot with normal lenses and cropped to 858 pixels vertically.  So it's not that cropping to 2.40 loses resolution so much that if you use a camera with a taller 4x3 sensor and use anamorphic lenses, you may be gaining some resolution.

 

However, this ignores the fact that many spherical lenses are sharper than anamorphic lenses.  Plus let's say that the camera doesn't have a 4x3 sensor, bit has a 16x9 or 1.89 sensor, meaning that if you use a 2X anamorphic lens, you'd have to crop the sides to get the image to 2.40, as opposed to using spherical lenses and cropping the top & bottom to get 2.40.

 

For example, if you shot on the 6K Red Dragon using 2X anamorphic lenses, you'd record in the 6K 6:5 mode, 3792 x 3160, whereas if you shot spherical and cropped to 2.40, you'd be using a 6144 x 2560 area of the sensor for the 2.40 image. That's a total pixel area of 11.98MP for anamorphic versus 15.73MP for spherical to end up with a 2.40 image.  So in this case, cropping to 2.40 still gets you more resolution than shooting with anamorphic lenses.

 

Keep in mind always that if you shoot with 2X anamorphic lenses and your final project is 2.40 : 1, then the area of the sensor you are using is limited to 1.20 : 1.


  • 0

#3 Mark Kenfield

Mark Kenfield
  • Basic Members
  • PipPipPipPip
  • 712 posts
  • Cinematographer
  • Melbourne, Australia

Posted 24 April 2015 - 01:23 AM

Which format are you recording on the F5? 4K raw? 4k compressed? 2k compressed? HD 4:4:4? Each is going to yield a different end result. If you're able to record in 4k (either raw or compressed XAVC) you'll have tonnes of resolution to play with. If you're recording 1080p 4:4:4, you'll have less resolution, but still an incredibly sharp and detailed HD image.

 

I've shot 2.39:1 projects in almost every recording format the F5 is capable of, and I've been totally satisfied every time. So I don't think you have anything to worry about.


  • 0

#4 Stefano Stroppa

Stefano Stroppa
  • Basic Members
  • PipPip
  • 35 posts
  • Cinematographer
  • London

Posted 24 April 2015 - 05:47 PM

 


thank you so much David! Always professional and passionate in explaining concepts! 

 

 

Mark, if I manage to get the AXS-R5 recorder I'll shoot 4K RAW, but probably for budget reason I'll shoot internal 2K in S-Log XAVC codec, my problem is that I have a final export requirement in 1920x1080 (and some other DCP in 2K) 

 

they want the cinemascope look, so I was concerning if I was going to lose resolution or not! 

when I set up the markers for 2.40, am I going to record them, or it's just a guide lines and I'll need to put them in post during editing? 

 

 


Edited by Stefano Stroppa, 24 April 2015 - 05:47 PM.

  • 0

#5 David Mullen ASC

David Mullen ASC
  • Sustaining Members
  • 18788 posts
  • Cinematographer
  • Los Angeles

Posted 24 April 2015 - 06:01 PM

I've heard of mixed results with the 2K mode of the F55 -- you may want to read this:

http://www.xdcam-use...f55-and-others/

 

You may be better off recording in 4K XAVC if you can't record RAW and then downsampling in post to 2K/HD.  Or record 1080P if you don't have any other deliverable requirement but HD.


  • 0

#6 Satsuki Murashige

Satsuki Murashige
  • Sustaining Members
  • 3081 posts
  • Cinematographer
  • San Francisco, CA

Posted 25 April 2015 - 01:33 AM

Well 1920x1080 is by definition a 16:9 (1.78:1) aspect ratio (assuming square pixels), so in order to deliver a 2.39:1 aspect ratio, you would have to either crop the vertical dimensions to get 1920x803 (2.39:1), or deliver a horizontally squeezed 1920x1080 image which can be later expanded to 2581x1080 (2.39:1). The former will contain less resolution, but the latter to my knowledge is not a DCI standard. Either way, you will have to make a compromise to get the 2.39 aspect ratio onto the 16:9 delivery format.

The F5 can only record 1.89:1 (4096x2160, 2160x1080) or 16:9 aspect ratios (3840x2160, 1920x1080), so pick whichever one works for you and crop in post. You can view 2.39:1 frame guides while shooting either format to help you compose. Also, the F5 with the new 4K license can shoot both flavors of 4K XAVC (1.89:1, 16:9) so you no longer need the Raw recorder just to get 4K.

If you're not sure whether to protect for 16:9 delivery, or whether you will need to deliver 2K instead of 1080, then you may want to just shoot 4K XAVC (or Raw if possible) and compose with near common top frame guides so that you can later extract either format and downscale in post with minimal compromises in framing. 4K XAVC at 23.98p is 32 minutes per 64GB, or 2GB/minute.
  • 0

#7 Mark Kenfield

Mark Kenfield
  • Basic Members
  • PipPipPipPip
  • 712 posts
  • Cinematographer
  • Melbourne, Australia

Posted 25 April 2015 - 02:46 AM

 

 

thank you so much David! Always professional and passionate in explaining concepts! 

 

 

Mark, if I manage to get the AXS-R5 recorder I'll shoot 4K RAW, but probably for budget reason I'll shoot internal 2K in S-Log XAVC codec, my problem is that I have a final export requirement in 1920x1080 (and some other DCP in 2K) 

 

they want the cinemascope look, so I was concerning if I was going to lose resolution or not! 

when I set up the markers for 2.40, am I going to record them, or it's just a guide lines and I'll need to put them in post during editing? 

 

 

 

 

The 2.40 guide markers aren't recorded into your internal footage, they're just for monitoring, so you'll have the extra frame height to tweak framing in your recorded files.


  • 0

#8 cole t parzenn

cole t parzenn
  • Basic Members
  • PipPipPipPip
  • 287 posts
  • Other

Posted 26 April 2015 - 08:50 PM

I've heard of mixed results with the 2K mode of the F55 -- you may want to read this:

http://www.xdcam-use...f55-and-others/

 

You may be better off recording in 4K XAVC if you can't record RAW and then downsampling in post to 2K/HD.  Or record 1080P if you don't have any other deliverable requirement but HD.

 

 

Interesting. I would think that treating each 2x2 square as one three channel pixel would be computationally easiest and give the "truest" image.


  • 0

#9 David Mullen ASC

David Mullen ASC
  • Sustaining Members
  • 18788 posts
  • Cinematographer
  • Los Angeles

Posted 26 April 2015 - 11:10 PM

I don't know, it sounds a bit blunt to me -- but I think one problem is that the OLPF was designed for 4K, hence the optional 2K OLPF.  But it seems to cause a hit in sharpness and I think downsampling 4K in post would lead to a sharper image.


  • 0

#10 Mark Kenfield

Mark Kenfield
  • Basic Members
  • PipPipPipPip
  • 712 posts
  • Cinematographer
  • Melbourne, Australia

Posted 28 April 2015 - 01:20 AM

I've heard of mixed results with the 2K mode of the F55 -- you may want to read this:

http://www.xdcam-use...f55-and-others/

 

 

 

With the F5(5), detail can (though doesn't always) take a hit at frame rates above 60fps when shooting in full-frame mode, and it can become easier to provoke the appearance of aliasing and moire. Below 60fps, everything is rosy, and shooting with the 2k centre-crop mode everything is rosy there too.

 

Also, I'd be surprised if anyone can notice or measure an increase in sharpness from downsampling the camera's 4k to 2k in post versus capturing the lower resolution in-camera. The F5(5)'s internally downsampled 2k and SR444 1080p images are exceptionally, bitingly sharp - so I'm not sure there are any additional line-pairs that you could cram into a 2k/HD image.


  • 0

#11 Stefano Stroppa

Stefano Stroppa
  • Basic Members
  • PipPip
  • 35 posts
  • Cinematographer
  • London

Posted 28 April 2015 - 05:23 AM

Well 1920x1080 is by definition a 16:9 (1.78:1) aspect ratio (assuming square pixels), so in order to deliver a 2.39:1 aspect ratio, you would have to either crop the vertical dimensions to get 1920x803 (2.39:1), or deliver a horizontally squeezed 1920x1080 image which can be later expanded to 2581x1080 (2.39:1). The former will contain less resolution, but the latter to my knowledge is not a DCI standard. Either way, you will have to make a compromise to get the 2.39 aspect ratio onto the 16:9 delivery format.

The F5 can only record 1.89:1 (4096x2160, 2160x1080) or 16:9 aspect ratios (3840x2160, 1920x1080), so pick whichever one works for you and crop in post. You can view 2.39:1 frame guides while shooting either format to help you compose. Also, the F5 with the new 4K license can shoot both flavors of 4K XAVC (1.89:1, 16:9) so you no longer need the Raw recorder just to get 4K.

If you're not sure whether to protect for 16:9 delivery, or whether you will need to deliver 2K instead of 1080, then you may want to just shoot 4K XAVC (or Raw if possible) and compose with near common top frame guides so that you can later extract either format and downscale in post with minimal compromises in framing. 4K XAVC at 23.98p is 32 minutes per 64GB, or 2GB/minute.

 

thank you Satsuki for the input!

 

 

here an interesting article on the 2K Center Scan mode of the F5

http://blog.abelcine...ode/#more-31240


  • 0

#12 Bruce Greene

Bruce Greene
  • Basic Members
  • PipPipPipPip
  • 393 posts
  • Cinematographer
  • Los Angeles

Posted 28 April 2015 - 10:26 AM

Stefano, It seems that you have 2 delivery requirements here: DCP at 2048x858 and 1920x1080.

If you shoot at 2048 pixels wide, (cropping for 2.39:1) you will have your full resolution for DCP, but you will need to downsample to 1920 (if you are letterboxing for the HD deliverable). This scaling will reduce detail. This can be somewhat avoided by a couple of approaches. Shoot in 4096 pixels wide, master in the this format in post, and downsample by exactly 1/2 to 2048 for DCP. This will retain the maximum detail for DCP and the downsample to 1920 will be a fractional division, but from a much higher resolution source, and should also look very good. Of course this means a 4k post finish, which may be more expensive, or beyond your computer power if you're doing the color post yourself.

But, if your delivery is 2.39:1 for DCP and FULL FRAME 16:9 HD (1920x1080), then you could master (uncropped) for DCP (and crop later) and use the 2048 (16:9) master and crop the sides to 1920 pixels for HD release without further scaling and softening of the image. I went this route on my last picture which was 2048x858 DCP and 1920x1080 TV release. There were some shots that needed to be rescaled and framed for the HD release of course, but on average, we were able to avoid scaling for the TV release. There was also a letterboxed HD version, that was only cropped on the sides to 1920 pixels and had no other scaling or repositions. (unfortunately, I don't think this version was released anywhere but my personal copy :)

We had shot in 2k Arri Alexa ProRes, so had no 4k master to work from.
  • 0

#13 Stefano Stroppa

Stefano Stroppa
  • Basic Members
  • PipPip
  • 35 posts
  • Cinematographer
  • London

Posted 28 April 2015 - 02:11 PM

But, if your delivery is 2.39:1 for DCP and FULL FRAME 16:9 HD (1920x1080), then you could master (uncropped) for DCP (and crop later) and use the 2048 (16:9) master and crop the sides to 1920 pixels for HD release without further scaling and softening of the image. I went this route on my last picture which was 2048x858 DCP and 1920x1080 TV release. 

 

Interesting!

thank you Bruce for the reply :)


  • 0

#14 Stefano Stroppa

Stefano Stroppa
  • Basic Members
  • PipPip
  • 35 posts
  • Cinematographer
  • London

Posted 28 April 2015 - 03:09 PM


use the 2048 (16:9) master and crop the sides to 1920 pixels for HD release without further scaling and softening of the image

 

hey Bruce, I'm a little confused: 

 

isn't it a 1.89 master, that becomes 16x9 as you crop the 2048 to 1920? 


  • 0

#15 Bruce Greene

Bruce Greene
  • Basic Members
  • PipPipPipPip
  • 393 posts
  • Cinematographer
  • Los Angeles

Posted 28 April 2015 - 03:22 PM

 
hey Bruce, I'm a little confused: 
 
isn't it a 1.89 master, that becomes 16x9 as you crop the 2048 to 1920? 


the master would be 2048x1152 (cropped to 2048x858 for DCP) and you would crop to 1920x1080 (for full frame 16:9) or letterbox the 1920x? for whatever you think looks best for a letterboxed version. I think we did 1920x818...
  • 0

#16 Stefano Stroppa

Stefano Stroppa
  • Basic Members
  • PipPip
  • 35 posts
  • Cinematographer
  • London

Posted 28 April 2015 - 03:30 PM

the master would be 2048x1152 (cropped to 2048x858 for DCP) and you would crop to 1920x1080 (for full frame 16:9) or letterbox the 1920x? for whatever you think looks best for a letterboxed version. I think we did 1920x818...

 

yes sorry, I forgot you shot with Arri Alexa, I was thinking about the F5!

ok thanks again for the reply! 


  • 0



Also tagged with one or more of these keywords: cinemascope, 2.4, resolution, matte, Sony F5, letterbox, aspect ratio, native

Willys Widgets

Technodolly

The Slider

Media Blackout - Custom Cables and AKS

Tai Audio

Pro 8mm

Broadcast Solutions Inc

Ritter Battery

rebotnix Technologies

Visual Products

Glidecam

CineTape

Aerial Filmworks

Paralinx LLC

Abel Cine

CineLab

Zylight

Rig Wheels Passport

Tai Audio

Pro 8mm

Broadcast Solutions Inc

Ritter Battery

Glidecam

Rig Wheels Passport

rebotnix Technologies

Visual Products

Aerial Filmworks

Technodolly

CineTape

Zylight

Willys Widgets

Media Blackout - Custom Cables and AKS

Paralinx LLC

CineLab

The Slider

Abel Cine