resolution vs redcode trade-off

[Mike Dominic]

Hi. I have a project coming up that is mostly going to be shot at 96FPS. My choices are 5KHD at 12:1 or 4KHD at 8:1. The final delivery is going to be 1920x1080.

I don't have an Epic in front of me to test, but I am wondering if anyone has any experience as to which option would give the better result. Or will they pretty much look identical by the time you rezise them to 1080p?

Thanks for your help!

 

[Ketch Rossi]

I never shoot less then full resolution, no matter of what frame rate is needed...

With MX was 5k with Dragon is 6k, I shoot at maximum resolution with maximum frame rate allowed at lowest Redcode allowed... Then take care of what ever additional Slowing down in Post with software of choice, and always have the original material in full rez, for frame pulling, and to feed any needs at any given time.

Of course this is what I have done and will always continue to do despite of what I would shoot for...

Others might and will of course have different choices opinion...

 

[Mark L. Pederson]

Hi. I have a project coming up that is mostly going to be shot at 96FPS. My choices are 5KHD at 12:1 or 4KHD at 8:1. The final delivery is going to be 1920x1080.

I don't have an Epic in front of me to test, but I am wondering if anyone has any experience as to which option would give the better result. Or will they pretty much look identical by the time you rezise them to 1080p?

Thanks for your help!

It really depends on how much detail there is in what you are shooting. If final delivery is 1920 x 1080 - I'd go for the 4K HD 8:1.

 

[Gunleik Groven]

I am with Mark (happens a lot these days, it seems)

When shooting highspeed, compression becomes a larger part of the equation.

But it is only an "issue" if you shoot plenty details at high aperture and high contrast.

If you are aiming for the "fluids in slomo"... just shoot whatever that is most convenient.

 

[Mark L. Pederson]

When shooting highspeed, compression becomes a larger part of the equation.

But it is only an "issue" if you shoot plenty details at high aperture and high contrast.

Agreed - and I would also add that if you are fighting noise floor with not enough light - you want as little compression as possible. If you loose detail due to compression - downsampling/scaling down isn't going to add it back.

 

[Nawaf Alsabhan]

But it is only an "issue" if you shoot plenty details at high aperture and high contrast.

Gunleik, What are some examples of plenty details?

 

[Justin McAleece]

Detail heavy shots could include: landscapes, moving leaves, moving water, explosions, driving shots or other fast motion shots. Basically the opposite of say, a talking head scene.

 

[Nawaf Alsabhan]

Detail heavy shots could include: landscapes, moving leaves, moving water, explosions, driving shots or other fast motion shots. Basically the opposite of say, a talking head scene.

Thanks for the info

 

[Les Dittert]

Detail heavy shots could include: landscapes, moving leaves, moving water, explosions, driving shots or other fast motion shots. Basically the opposite of say, a talking head scene.

Just a quick note : moving leaves,water is easier than still, as the motion blur is easier to compress.
R3D compression is 'Intra frame' compression. There is no motion used in doing the compression.
Each frame is completely separate and independent from the next frame.

 

[Björn Benckert]

Hi. I have a project coming up that is mostly going to be shot at 96FPS. My choices are 5KHD at 12:1 or 4KHD at 8:1. The final delivery is going to be 1920x1080.

I don't have an Epic in front of me to test, but I am wondering if anyone has any experience as to which option would give the better result. Or will they pretty much look identical by the time you rezise them to 1080p?

Thanks for your help!

You have to do your own testings for the given shot. As there are quite a few parameters in play.

When you crop down on the sensor you stress the resolution of your lenses (and sensor) and when you don't you obviously get more compression. Personally I would aim for FF as far as possible. Then if I want higher speed than whats possible in full frame I would shoot takes with cropped sensor and to cover to cover up all bases also shoot takes with Full frame at highest frame rate.



Then in post you have the options to choose between real high speed with cropped sensor or use any of the motion estimate timewarp softwares available like cinespeed or twixor. But, very important, those programs are not to count on as quite often they do not calculate things right and you might get a lot of artifacts if using them. Which is depending on whats going on in the shot, how much movement you got between frames and what the shot contains. Normally it's quite difficult to tell from just looking at the given shot so you basically need to test it and for the times it does not work then the cropped sensor take will be the better choice.

So shooting 96fps and then think that post will be able to slow it down further is quite a risky way to go. For some shots it will work nicely for other shots... not at all. So better shoot both version and if you only could do one then cropped sensor it is, then you at least know that you have the shot. even though it will not be as crisp.

 

[Gunleik Groven]

What Björn said.

Moving leaves against a bright sky at T8 with a wide lens, is a known R2D killer at higher compression, whatever the motionblur does...

There is a distinct point to Peter Collisters Tree-shot in his Dragon test

 

[Brice Ansel]

Hi. I have a project coming up that is mostly going to be shot at 96FPS. My choices are 5KHD at 12:1 or 4KHD at 8:1. The final delivery is going to be 1920x1080.

I don't have an Epic in front of me to test, but I am wondering if anyone has any experience as to which option would give the better result. Or will they pretty much look identical by the time you rezise them to 1080p?

Thanks for your help!

What Björn said.

Moving leaves against a bright sky at T8 with a wide lens, is a known R2D killer at higher compression, whatever the motionblur does...

There is a distinct point to Peter Collisters Tree-shot in his Dragon test

Hi,
here is a test shot I did on a tree @ T8 under a bright and contrasty day.
Lens is a Leica R 19mm Elmarit V1 shot @ 5K FF and 75Fps; (can't remember if shuter angle was 120 or 180)

EDIT: Of course everybody has its own tolerance, and speaking for myself, I found that pushing beyond 75Fps (and using a higher compression rate) for this type of shot will not make it.
Cheers.
B.

 

[Karl H]

this is an area where I'd love to see an option to downscale to 2K RGB and record for slo-mo. It would retain the detail much better than trying to record 5K at 12:1. It's not in red's philosophy to do so it seems, but it's not in mine to stop asking either

 

[Les Dittert]

Why on earth would 'moving' have any difference ?
R3D ( Jpeg 2000 ) has no motion involved in compression. None. Nada.

Motion actually helps, as it reduces the high frequency component of the image.

What Björn said.

Moving leaves against a bright sky at T8 with a wide lens, is a known R2D killer at higher compression, whatever the motionblur does...

There is a distinct point to Peter Collisters Tree-shot in his Dragon test

 

[Nick Morrison]

This thread is a great read, thanks to all for contributing. Reminds me of when Ridley Scott was shooting Prometheus - they often chose to shoot 3:1 in Iceland (instead of 5:1) because of the compression challenges the foliage and water posed.

 

[D Fuller]

Really? That's interesting. So R3D is always an intraframe codec? I guess that makes sense, given the ability to produce an R3D still, but all these years and I did not know that.

Why on earth would 'moving' have any difference ?
R3D ( Jpeg 2000 ) has no motion involved in compression. None. Nada.

Motion actually helps, as it reduces the high frequency component of the image.

 

[Robert Ruffo New]

It really depends on how much detail there is in what you are shooting. If final delivery is 1920 x 1080 - I'd go for the 4K HD 8:1.

I most certainly would not - as I've said many times before, if most people can't see the resolution difference over HD then 1080p is the same as 5K delivery - it doesn't matter. 1080p will show you many image flaws, if present (and 4K, again, won;t show you any more as yoru audience won't be able to see more, so who cares)

WHat will matter:

- Secondaries will be way noisier in higher compression - HSL will be very, very tricky. Any grading in fact, especially if it digs into the blacks too much. canl give you problems if you go over 6:1

- 4K is a crop on your lens - not nearly as nice as 5K, if your lens can cover that.

- 4K is a zoom-in on any noise - you're losing at least half a stop in the noise floor if you shoot at 4K instead of 5K - think about it, you're using less of teh light the lens saw to record an image, you're discarding some of the available information

- Hard drives are almost free now - why bother with compromises?

OOPS - Edit I didn;t realize Marc was specifically talking about optimal slow-mo - so sorry Mark.

 

[Mark L. Pederson]

Why on earth would 'moving' have any difference ?
R3D ( Jpeg 2000 ) has no motion involved in compression. None. Nada.

Motion actually helps, as it reduces the high frequency component of the image.

This is my understanding as well. There are times, when motion - specifically in a high contrast scene will make you NOTICE the compression artifacts more. So ... all those tiny leaves blowing in the breeze - you notice the edges more than if the tree was 100% still.

 

[Robert Ruffo New]

Motion creates new blobs of incorrect macroblock estimation on every frame, so the macroblocking cannot be hidden.

 

[Mark L. Pederson]

- 4K is a zoom-in on any noise - you're losing at least half a stop in the noise floor if you shoot at 4K instead of 5K - think about it, you're using less of teh light the lens saw to record an image, you're discarding some of the available information

Sorry Robert - but this statement is not technically correct.

You are NOT "zooming in on noise" - you are cropping the sensor. With respct to lenses - other the the very high end glass like Lecias and Masters - fall off occurs at the edges of the glass. On most lenses - cropping to 4K gives you an overall sharper image if you sample detail across the entire image. I have ZERO idea where are are coming up with "you're losing at least half a stop in the noise floor if you shoot at 4K instead of 5K" - please elaborate as that is not my experience at all.

Compression doesn't really know the difference between noise and detail - so ... if you are fighting the noise floor - and you go for higher compression - you are more likely to take a hit on the detail you want to protect. This is why I always recomend LESS compression in very low light - and there are times when having less compression will trump more resolution. Depends on how much detail in the frame, how much noise - and what compression settings we are talking about. 12:1 vs 8:1 is a bigger issue than 5:1 vs. 3:1