Is Film the Great Equalizer?

[Rudi Herbert]

Here's something that's had me thinking for some time now. We always tout the virtues of HDV and DV before as "enough" for TV but definitely never good enough for large screen projection, with HD barely making the cut. It seems that every SD or HD camera can, with a lot of care and expertise, squeeze by and look acceptably good on a small screen, but don't you ever delude yourself in thinking that something like that could give you the resolution and quality to be shown on a big screen without the audience going: "what the f... is wrong here?"

And, as I've navigated this forum for months now, before making the decision of buying a RED and after the decision was made, often times I've been concerned with footage samples where there's noise in the blue channel, or blown highlights there, or lost detail in the blacks over there, and I've obssesed about the fact that if such deffects are obvious on a computer monitor, then they'll be glaringly unnaceptable on a cinema screen, right? Then I read a thread where the VFX supervisor on Wanted made it clear that RED, compared side by side to Kodak film scanned "AT 4K", look virtually identical to trained eyes. Then I also re-read an old thread when the first cinema projection of RED material took place in NY, and EVERYBODY agreed on the fact that RED looked awesome, without noise, clipping or the like, basically, just like film. And then I went back to a print we made from a Sony Z1U and transferred to film, and watched it again on a real large screen, and could not really tell that a "lower" quality camera had been used. And I remembered how impressed I was when I saw Robert Rodriguez's Once Upon a Time in Mexico, which was shot on the Sony F-900, actually cropping the 2/3'' 16 x 9 sensor to the 2.40 x 1 widescreen format AND recording to tape at 3:1:1 fidelity and still looked VERY good on the big screen. Imagine how much better you can do with a RED and a little care in your exposure.

So, instead of being the ultimate test of fidelity and resolution, is theathrical projection indeed a sort of equalizer, a medium that actually forgets, forgives and flattens the imperfections of digital cameras to a very acceptable degree instead of showing how inept they are? Do we obsses too much about how much quality and resolution we need for a film blow out when in reality film will help smooth out the very imperfections we terribly fear? Should we instead be more worried with the nasty artifacts that compression for the web and other small, albeit crappy, formats introduce to our beloved and hard acquired footage instead of declaring our digital cameras not fit for the big screen? I'm beginning to lean more towards this latter way of thinking myself...what say you?

[Bruce Allen]

Yes.

I thought the football in the Leatherheads trailer titles looked fine printed to 35mm... and we shot that on my HV20.

...but then it also looked fine on TV during the Superbowl and on the web in HD.

You have a key phrase - "a little care in your exposure". That's it. You HAVE to nail it, there is no excuse. If you didn't get it right, do it again. Be prepared to admit mistakes and don't be afraid to re-shoot.

Yes, I think we absolutely obsess too much about compression artifacts and resolution needed for film blow up. Just light it right to begin with and compression artifacts won't be that noticeable.

Bruce Allen
www.boacinema.com

[Tom Lowe]

It isn't a matter of resolution, it's a matter of dynamic range. 90% of what sucks about CMOS, CCD and digital sensors are highlights blowing out. There is no way around it, unless you are skillful in using grad NDs or just exposing with great care. RED is essentially a 2K camera, and you should finish at 2K, IMO, while keeping your 4K REDCODE as archive.

Red matches or beats 35mm in terms of resolution right now, but in terms of dynamic range, 35mm still wins. But also keep in mind that the scanning process of film degrades the image more than most people are willing to admit.

[Bruce Allen]

Agreed totally, Tom. The Leatherheads thing worked because it was a studio shoot. Likewise, for Wanted they probably had plenty of control over the light, so Red could totally compete with film.

Luckily it's not that hard or expensive to expose carefully, use grad NDs and light properly... also those skills are handy when doing film projects too!

Totally agree on the 2k / HD finish part too - resolution isn't the problem, dynamic range is. 4K post is just too time-consuming and resource-wasting for indies. Time wasted waiting for 4K shots to render is time you could be spending fixing the edit, doing better matte paintings, tweaking filters, etc.

I know of several theatrical releases where they would take HD originals and apply 20%+ scaleups and it flew. Finish in 3K if you must but 4K is overkill and a counter-productive distraction in most cases I think because as Rudi says, in the end you are likely to be going out to a slightly soft film print. Certainly I've noticed this in film titles and trailer graphics - even at 2K you spend a ton of time tweaking pixels but when you print to film you don't really see those details because they get kinda smoothed and blurred. Instead you wish you'd spent more time on the overall composition / concept / etc.

I've done probably > 20 film finishes - and I still get surprised at how much detail is lost.

Hey, only way to prove this is to make a film, right? How are you guys' projects going?

Bruce
www.boacinema.com

[laguun]

some years ago, i had the revealing experience: i had 35mm film outs looking great, and when we projected the -exact- same masters via 4k many issues became clearly visible which the 35mm print swallowed.

A 35mm print, especially after the distribution copy process, will indeed erase lots and lots of resolution, technical issues etc.

[Tom Lowe]

Bruce, you know our conversation about finishing 2K vs 4K dates back probably a year by now! But yeah, at some point I looked at the camera, at the workflow, at the resolution vs the noise, and at other factors, and thought that a 2K finish makes the most sense.

[David Mullen ASC]

The answer is yes and no. Some things get blunted and blurred by the film-out and print projection process, particularly if it goes through and IP/IN dupe stage. But other artifacts can survive all those steps (clipping, for example, or excessively deep focus) and still be visible on the big screen, if not more visible due to the high degree of enlargement of the image (some lens artifacts, for example, may be more noticeable on a large screen).

Even if you shoot a movie in 35mm, there are a lot of quality control issues to fuss over. That's why you were hired to shoot the movie, to care about those details.

[Hans von Sonntag]

Projected 35mm on the big screen has approx. 720p resolution.

Its perception that counts, not resolution.

Scaling RED footage from 4k to 2K creates a big colour space. Keyword here is oversampling. Yes, I find RED a brilliant 2K/1080p camera and I'm not needing 4K for deliveries at all. Who really is?

Hans

[Bill Anderson]

David, regarding this film transfer issue, what is the impact of "excessively deep focus"?
Great posts guys. Thanks.

[James T Mather]

I believe it was Edward Woodward.

[Dylan Macleod, CSC]

Projected 35mm on the big screen has approx. 720p resolution.

Its perception that counts, not resolution.

Scaling RED footage from 4k to 2K creates a big colour space. Keyword here is oversampling. Yes, I find RED a brilliant 2K/1080p camera and I'm not needing 4K for deliveries at all. Who really is?

Hans

I am.

I am curious to know why people feel 2K is better than 4K for finishing. Is this just a cost/processing time issue?

If you had the budget and time to finish 4K and you had access to a 4K digital projector wouldn't you get superior image quality?

I am in the process of testing this now. We are making two versions of the test for viewing on the 2K projector we have access to, but we are also going to do a 4K version. Finding a 4K projector is the challenge I am presently dealing with.

Is 4K at the "point of diminishing returns" right now?

Dylan Macleod
Cinematographer
Toronto, Canada
www.dylanmacleod.com

[David Mullen ASC]

David, regarding this film transfer issue, what is the impact of "excessively deep focus"?
Great posts guys. Thanks.

There is no "impact" -- it's just a visual aspect of some types of photography that survive any distribution chain. It won't "disappear" if the projection is soft, etc. So if you're thinking that a 1/3" CCD HD camera will look the same as something shot with a 35mm-sized sensor camera because once it goes to film and projected, the differences will be blunted, the depth of field difference is one element that won't be.

[Daniel Browning]

What is a "film projector"? Is that what people used to watch movies on before there were 2K digital projectors? Pity those poor, backwards folk.

[Bruce Allen]

I am.

I am curious to know why people feel 2K is better than 4K for finishing. Is this just a cost/processing time issue?

If you had the budget and time to finish 4K and you had access to a 4K digital projector wouldn't you get superior image quality?

I'd spend the budget on something else. Matte paintings for blown-out skies, another layer of power windows, smart blurs, more time in sound mix, better production design, makeup on set, etc.



Is 4K at the "point of diminishing returns" right now?

Yes, IMHO. It's just not the smartest thing to put your money in. I can think of tons of great-looking films that were 2K DIs. For example, Amelie and Master & Commander - both had heavy, heavy work done in post. If they had tried to do it at 4K they would have had to compromise on the amount of post work they did and would have looked worse, IMHO.

Only reason to go 4K is for marketing hype, or arrogance, or if you're going after a particularly super-crisp-beyond-35mm aesthetic. This is assuming you have ample time, budget and opportunity to get it right on set though, because what you can do in post is compromised by the resources required to process at 4K.

Just my 2c. And of course I break my own rules. At the moment, my favorite workflow for effects stuff is to output from RedCine at 2880x1620 (1.5 x full HD - it seems to be a decent resolution that gets close to the max out of what Red can deliver), do blowups, repos, keying, etc at that res, and then downsample to 1080p on final render.

Bruce Allen
www.boacinema.com

[Bill Anderson]

There is no "impact" -- it's just a visual aspect of some types of photography that survive any distribution chain. It won't "disappear" if the projection is soft, etc. So if you're thinking that a 1/3" CCD HD camera will look the same as something shot with a 35mm-sized sensor camera because once it goes to film and projected, the differences will be blunted, the depth of field difference is one element that won't be.

For sure. You threw me with the "artifact" handle, and, since it was coming
from someone with your ability, I thought I was missing something.
Merci.

And I can't begin to imagine who would expect depth of field issues to be affected (no matter how subtly) by a film transfer and enlargement.

[Bill Anderson]

Bruce. dmcleod might have been pointing the finger (rightly so) at a post I made and quickly deleted -one about film and diminishing returns at a particular enlargement- or something of that nature- because Mcleod's post beat me to the punch and made my point moot.

[Hans von Sonntag]

I am with Bruce.

Invest your money in everything else but in an expensive 4K workflow. Highend productions may consider 4K but before this they spend their money in a great script, cast, locations, director etc...
4K is the the least asset if we talk about production value.

Hans

[laguun]

I am suprised to see that many folks think that 4K onlining must be complicated or expensive.

for us the cost difference is at ~5-15% 2K vs 4K.
DI & VFX sources in 4K often make things as pan & scan, key, track, denoise etc easier and therefore cheaper-
3D rendering is indeed taking much longer, but hey, in CGI render time seems to become the only ressource growing constantly.

and i dont think one should look at it as in "reduce script to get more light, reduce lens to get more cam, reduce resolution to get more audio bitdepth".

Every dept should work to its maximum quality in budget, and 4K editing, DI, vfx and titles are no different from 2K DI, vfx and titles except for computing power and diskbandwidth. 4K and 2K are both realtime anyhow, and if you need to render a grade etc, a solid renderfarm (cpus are soo cheap today) should easily munch through a day of work overnight today.

Its different with a-budgets, thats for sure, when 15 studios have to work on a WAN and have >1.500 shots to coordinate, it might be recomendable to use 2K often. But inhouse, for the flagships, we use 4K sources, even if the master is only 1080P/4K.

[Bruce Allen]

Laguun, totally agree RE degrain, pan-n-scan advantages, etc.

I'm humbly submit however that for most people (maybe not you), the storage, labor and equipment costs for 4K instead of 2K are over your 5-15% figure. We are talking 400% the data that needs to be processed, after all.

Also, Hollywood chooses 2K not solely because of the bulk of the work, but because they also value interactivity when doing complex stuff.

Render farms don't improve interactivity in many cases - but working at lower-res does. If you want reasonable responsiveness working with 4K stuff, you'll be on proxies pretty much all of the time.

I disagree strongly with your statement that 4K and 2K are realtime. If they are, then you're simply not doing complicated work.

Take a look at Imax films, for example - the effects, though crisp, often look kinda "last-generation" - because the creative tools / artistic workflow has been limited by the need to do 4K deliverables.

There are architectural limits in a lot of useful desktop film software that makes it not good for 4K. 3D programs run into real problems when dealing with the textures needed for 4K. Photoshop has difficulty loading those textures to be edited. After Effects is pretty cumbersome at 4K, some plugins simply fail because they weren't tested at that resolution, etc.

Finally, my impression is that when working at 4K, you need to go for the most RAM, greatest # of CPUs, fancy high-end storage systems, etc - in my experience, the bleeding edge and not the best place to be price / performance-wise.

All of this will be solved in a few years I agree. But for most Redusers, I think 4K is silly for now.

But much respect to you - if you can do 4K finishing cheaply now, that is great. You are one of the good guys and you're helping everyone by offering such services. If the world was filled with post houses and production companies with more of your attitude, I'm sure it'd be a better place for filmmakers.

Bruce Allen
www.boacinema.com

[Chris Pickle]

4k vs 2k debate is why I still believe that maybe the Red should have been a 2k camera but with the same physical sized sensor. It would have allowed for larger / faster pixels. Plus the work-flow and data management would have been easier.

Except for the odd FX job, I can't imagine where you would need 4k over 2k.

Just my thoughts.
chris

[Bruce Allen]

4k vs 2k debate is why I still believe that maybe the Red should have been a 2k camera but with the same physical sized sensor. It would have allowed for larger / faster pixels. Plus the work-flow and data management would have been easier.

Except for the odd FX job, I can't imagine where you would need 4k over 2k.

Just my thoughts.
chris

I think there's no way you can get full 2K RGB data from a single-CCD 2K sensor, with an optical low-pass filter. 1.5K probably.

But yes I agree 3K with larger sensor sites (which would make a lovely 2K RGB image) would have been great too.

EDIT: of course my beloved HV20 would fit into your category. So I guess 1.5K is all you need?

Actually, come to think of it, wasn't the original Toy Story only rendered at around 1.5K? Ah yes, 1536 × 922...

Anyone remember watching the original Toy Story and thinking "man, that looks soft"? Hahahaaaa... I certainly didn't.

Bruce Allen
www.boacinema.com

[Chris Pickle]

I think there's no way you can get full 2K RGB data from a single-CCD 2K sensor, with an optical low-pass filter. 1.5K probably.

But yes I agree 3K with larger sensor sites (which would make a lovely 2K RGB image) would have been great too.

Bruce Allen
www.boacinema.com

Yes, you're right. I thought of that after I posted. There is some loss by the time it gets written to a file. And you're right on the 3k, or whatever it needed to be to get down to 2k.

Someone could have down the math to figure out what they would have needed to start with to get to 2k. Say a 20% loss, then a 2.5k sensor to give a true 2k file. It would have allowed for much bigger pixels.

I helped produce on a 3 day commercial shoot, and we were all very surprised at how slow the camera is. I spoke with the camera tech department at a big rental house in Toronto. (they have several dozen Reds on order) The tech said he "bench tested" the Red in the mid 200s. After our shoot, I was inclined to believe him.

Chris