Jump to content


Photo

4K vs 2K scan... is it worth it?


  • Please log in to reply
8 replies to this topic

#1 Rene Renault

Rene Renault

    New

  • Basic Members
  • Pip
  • 4 posts
  • Cinematographer

Posted 11 October 2013 - 09:24 PM

Assuming I have good glass, will I see a difference when finishing for a 2K digital projection, for an art house theater run on smaller sized screens?

 

Thoughts?

 

Thanks.

 

Rene


  • 0

#2 David Cunningham

David Cunningham
  • Basic Members
  • PipPipPipPip
  • 1049 posts
  • Cinematographer

Posted 11 October 2013 - 09:40 PM

Depending on lots of things such as grain structure, number of fine/small details or really wide shots... Lots of other things come into play. The most important thing is grain resolution and aliasing.

A 4k scan may not yield you more detail, but even when downrezzed to 2k it may appear sharper because of reduced grain aliasing and/or smudging.
  • 0

#3 Zac Fettig

Zac Fettig
  • Basic Members
  • PipPipPipPip
  • 338 posts
  • Other
  • Boston

Posted 11 October 2013 - 09:56 PM

You certainly will see the difference in your wallet. It really depends on what you're filming, but for a straightforward film, a 2k transfer is probably more than adequite. A 4k will be better, but is it better enough? For a small scale art house run, I wouldn't think it's worth the added expense.


  • 0

#4 Will Montgomery

Will Montgomery
  • Sustaining Members
  • 2030 posts
  • Producer
  • Dallas, TX

Posted 14 October 2013 - 01:39 PM

Until recently many if not most effects shots done at ILM were in HD resolution; not even 2K. Mainly because of time and expense vs. the quality jump (as in not worth it; at least to George Lucas ; )

 

If you're working in 16mm, 2k will be just fine. Spend the extra time and money on script and art direction. If you're working in 35mm I'd still go 2k but if needed you can always re-master in 4k as needed.


  • 0

#5 Robert Houllahan

Robert Houllahan
  • Sustaining Members
  • 1582 posts
  • Industry Rep
  • Providence R.I.

Posted 15 October 2013 - 08:38 PM

I was just talking to the main guy who builds our Xena scanner in LA and he has done some extensive tests of "Super-2K" and "Super-4K" scanning and by "Super" I mean a 3K sensor area down-sampled to 2K and a 6K area for 4K. They have also tried 4K for 2K and he said that there were clear advantages to the "Super" scanning approach in terms of resolving grain more finely and anti aliasing. I think if the scanner is scanning oversampled to 2K you get 95% of the advantages of a higher resolution scan without all of the data management which comes with higher resolution digital files.

 

-Rob-


  • 1

#6 Carl Looper

Carl Looper
  • Basic Members
  • PipPipPipPip
  • 1425 posts
  • Digital Image Technician
  • Melbourne, Australia

Posted 21 October 2013 - 03:43 AM

Is it worth it?

 

That is such an impossible complex question but here's another approach:

 

One might say for each increase in quality, that one might achieve, one might note that the corresponding cost increases at a faster rate.

 

Is the quality worth the cost?

 

Ignoring charlatanism, the cost of something is effectively it's worth.

 

For example, the worth of gold is related to the cost of obtaining it, ie. finding and digging it out of the ground.

 

Likewise the worth of obtaining a higher definition scan will be related to the cost of obtaining it.

 

Cost is one thing. Whether anyone buys it is another. I assume if machines have managed to sell for x thousand dollars there is a market there.

 

The only other metric one can use is one's own consumer instincts - would you buy it? Perhaps not, but that won't tell you much. You might find that even though you don't buy it, you could sell it it to someone who does buy it. In whcih case you might buy it because then there is worth to be found after all, in the difference you can pocket.

 

C


  • 0

#7 David Cunningham

David Cunningham
  • Basic Members
  • PipPipPipPip
  • 1049 posts
  • Cinematographer

Posted 21 October 2013 - 10:05 AM

For example,

 

I find it is "worth" it to get a flat scan to DPX at 2K of my Super 8 films.  The reason is 2 fold. 

 

1.) Slight increase in resolution - Sharper final images but especially important for reduced aliasing

 

2.) More control over the final color/grade.  The highlights are less likely to be clipped or the shadows muddied.  

 

It's more work, but a superior result even though I finish in HD or even 720p.

 

My "worth it" trade off is I tend to get fully scene-to-scene graded transfers done in 1080p for my wedding shoots.  This allows me to work more quickly. 

 

Getting a 2K scan on a ScanStation to flat DPX or Uncompressed QT files is actually cheaper than fully graded 1080p scans.  But, now I need grade and the jury is still out on whether the Bayer Pattern area sensor "really" looses anything.

 

Anyway, my 2 cents.


  • 0

#8 Will Montgomery

Will Montgomery
  • Sustaining Members
  • 2030 posts
  • Producer
  • Dallas, TX

Posted 22 October 2013 - 02:05 PM

Another thought is if we are close to a "golden age" of scanning film right now. In other words, there will always be film scanning and all the archives will never be completely scanned, but will film scanning technology ever get "better" than it is now? Will over sampling at 3k to go down to 2k or HD makes sense to me, scanning 16mm at 8k really doesn't seem like it's going to give us that much information since most 16mm lenses won't resolve to 8k, right? There maybe some incremental improvements but it appears we are close to getting the most out of film that we can (technically) and much of what remains is the subjective skill of a colorist.

 

I guess what I'm saying is that it's not going to get better than today for scanning film and I'm not sure that the cost will come down dramatically either. The real cost is the colorist's time; not even the equipment so much these days. Although Paul at Cinelicious may have other ideas (maybe 50/50 people & machine costs).


  • 0

#9 Carl Looper

Carl Looper
  • Basic Members
  • PipPipPipPip
  • 1425 posts
  • Digital Image Technician
  • Melbourne, Australia

Posted 25 October 2013 - 06:26 PM

This is purely speculation based entirely on my own experiments in this area, but I reckon transfers will get better. And of course, will get cheaper.

 

The issue is not so much how many K you need to get a good transfer - but how many Ks you need to transfer for a particular image processing algorithm to do a better job at reconstructing the image, ie. in the digital domain.

 

As the algorithms get better the more they will demand of the transfer stage.

 

Algorithms are necessary because there is not a 1:1 mapping between film and digital. The information in film is encoded differently to that of digital. Simple 1:1 pixel transfers take a brute force approach and you hit data handling limits before you hit lens/film limits. And you can't really characterise film or lenses in terms of Ks. MTF curves do a better job but the real curves don't just stop at 30% (or wherever they do in an MTF chart) - they go all the way down. But without algorithms capable of exploiting the faint analog signal below the chart cutoff point neither the charts nor transfers bother to elaborate the curve beyond the point that they do.

 

There will be some sort of technical cutoff but the cutoff is not really characterisable in terms of a single point because it is a statistical rather than mathematical attribute. All measures of cutoffs are typically done through subjective tests because apart from the statistical nature of it, any candidate objective ones (proposed mathematically) tend to give absurdly high numbers in terms of the Ks you would need to reach the candidate cutoff.

 

The problem with subjective tests is that they don't really take into account the subconscious perceptions that might be going on (especially when film is playing rather than being viewed as a single frame) - and it could be in such subconscious perceptions where those absurdly high K numbers do have physical meaning.

 

C


  • 0


CineLab

rebotnix Technologies

Abel Cine

Gamma Ray Digital Inc

Broadcast Solutions Inc

Tai Audio

Glidecam

Paralinx LLC

The Slider

Media Blackout - Custom Cables and AKS

FJS International, LLC

Ritter Battery

Metropolis Post

Visual Products

Willys Widgets

Technodolly

CineTape

Aerial Filmworks

Wooden Camera

Rig Wheels Passport

Opal

Opal

Aerial Filmworks

CineTape

Metropolis Post

Rig Wheels Passport

Media Blackout - Custom Cables and AKS

Visual Products

The Slider

Paralinx LLC

FJS International, LLC

Ritter Battery

rebotnix Technologies

CineLab

Broadcast Solutions Inc

Abel Cine

Wooden Camera

Tai Audio

Gamma Ray Digital Inc

Willys Widgets

Technodolly

Glidecam