Film-Tech Forum ARCHIVE: Comparing the importance of resolution to perceived quality

my profile | my password | search | faq & rules | forum home

»	Film-Tech Forum ARCHIVE » Community » Film-Yak » Comparing the importance of resolution to perceived quality

Author

Topic: Comparing the importance of resolution to perceived quality

Scott Jentsch
Phenomenal Film Handler

Posts: 1061
From: New Berlin, WI, USA
Registered: Apr 2003

posted 03-02-2018 03:13 PM

Steve Yedlin, cinematographer for Star Wars: The Last Jedi, Looper, and others, compares several cameras/formats of various resolutions to determine whether resolution alone is the best way to judge the quality that ends up on the screen.

http://yedlin.net/ResDemo/

There are two videos. The first one establishes his approach, and the second one is an hour-long thesis of his process and how he came to his conclusions.

I probably can't fully appreciate everything he's talking about because I'm not a cinematographer, but it's very interesting to watch.

If I'm understanding what I've seen correctly, it would explain why the pixel counts of how a movie was produced, post-produced, and distributed isn't always an indicator of how it's going to look. Reaching into the topic area of the After-life section, this would influence how the 4K Ultra HD discs look on movies that come from a variety of sources, and the hand-wringing that some get into about whether a movie is "fake 4K" or not is unfounded.

| IP: Logged

Bobby Henderson
"Ask me about Trajan."

Posts: 10973
From: Lawton, OK, USA
Registered: Apr 2001

posted 03-02-2018 05:13 PM

Yes, there are lots of variables in digital cinematography. The number of pixels on the camera's image sensor are just one of those many variables. Color depth is another important variable. Different camera sensors vary in their quailty of performance (particularly in low light). Not all camera lenses are created equal.

Ultimately the capability of current generation digital cinema camera systems far outstrips the capabilities of digital cinema projection systems. And the imagery being shown in d-cinema is of considerably better quality than the product one can buy/rent for home.

Steve Yedlin's comparison of cameras, formats, etc in cinematography is interesting, but it does not provide an excuse for the Ultra HD Blu-ray video format being mostly a waste of money for consumers right now thanks to most UHD discs being authored from 2K resolution sources.

Cinematography is only one part of what goes into a movie's final product. CGI effects and digital back lot elements are playing bigger roles than ever in production. All that stuff has to be composited together and color graded. The final digital intermediate is the end result of that work. It consists of pixel-based digital images. If a movie's final post production product is a 2048 X 858 pixel image there really isn't anything that can be done to turn that 2K image into a true 4K product without completely re-doing the post production process.

I stand by my harsh criticism of 2K sourced UHD discs and will continue to call such products "fake 4K." UHD discs carry a significant price premium above standard 1080p Blu-ray discs. When the UHD is sourced from a 2K master it's very difficult to see any improvements in video quality when the UHD is compared to the same movie on Blu-ray. The only advantage the UHD offers is maybe some improvements in video compression. But that's only if the disc authors take advantage of the higher available bandwidth rather than using the HEVC codec to crush file sizes down even smaller.

In the end when it comes to 4K viewing there is no substitute for material in native 4K resolution. This is easy to see when comparing the few UHD titles with native 4K material versus so many 2K hack jobs. Up-rezzing material to 4K is just plain bullshit, even when using the latest image enlargement filters and tricks. I see it enough in the graphic design field, like people expecting some low res image they snagged off the Internet to look good on a vehicle wrap or billboard. Passing off 2K material as 4K is even more lame since it's a product being sold to the public as "Ultra HD" and "4K."

| IP: Logged

Scott Jentsch
Phenomenal Film Handler

Posts: 1061
From: New Berlin, WI, USA
Registered: Apr 2003

posted 03-05-2018 01:47 PM

When increasing resolution while keeping the sensor size the same, the photosites typically have a harder time doing a decent job, especially in low-light. This has been an issue in the digital still camera industry as it chased ever-increasing megapixel numbers.

I thought the bit about halation was quite interesting. I had never heard the term before, but watching the comparison of how the different cameras handled the sequence with the backlit blinds was quite interesting.

I'm all for high resolution, and I don't necessarily buy all of his arguments that we've reached the practical limit of how many megapixels are necessary to properly resolve detail. However, he does demonstrate some very effective up-rezzing of 2K material to 4K and, I believe, 6K, and the difference wasn't as stunning as one might assume.

He also touched on bandwidth, where cranking too hard on the compression caused banding, which can be much worse, and much more noticeable than some lack of fine detail.

Interestingly enough, Yedlin makes the case for displays to have more resolution than the source material. I agree with his comments that adding resolution for resolution's sake doesn't pay off if it means that other aspects are compromised as a result. A camera that is high resolution but doesn't perform for the task at hand shouldn't be used. I think there's a case to be made for doing visual effects in higher resolutions because they are inherently non-photographic and not affected by all the other characteristics of a camera (sensor, lens, etc.).

Could you tell a noticeable difference between the sample clips that he created? There were some examples of drastically different source resolutions, but very little differences in the end product. I wouldn't call any of them unacceptable, and I was watching it full-screen on a 4K display at a distance of about 2.5 feet.

I don't subscribe to the "fake 4K" viewpoint because I think it's too narrowly focused. There have been examples of well-rated UHD titles sourced from 2K material, and I'm sure that there are also mediocre titles sourced from 4K material. So much goes into the production process that I feel it's ultimately most important to look at the end product.

Personally, I've been buying 4K UHD discs most often over Blu-rays. I'm willing to pay the extra $5 to get Atmos/DTS:X soundtracks (which are often exclusive only to UHD) and to get the future-proofing that the 4K disc provides, even though I don't have a 4K display in my home theater or living room.

I watch the reviews closely, and if a title gets knocked for not being any better, then I'll save the $5 upcharge and get the Blu-ray (assuming the sound formats are the same). In many of those cases, I'll just hold off and get the disc from Netflix/Redbox and wait for the inevitable re-release which will probably do a better job, just like what has happened with DVD and Blu-ray releases before.

| IP: Logged

Aaron Garman
Phenomenal Film Handler

Posts: 1470
From: Toledo, OH USA
Registered: Mar 2003

posted 03-05-2018 03:27 PM

Some of the new masters they are using for catalog UHD discs is important too. For instance, a Blu-ray of some film may have had some edge enhanced transfer that was made 10-15 years ago whereas the 4K UHD disc has a new scan of the film elements.

That too can be a major factor in the end result.

I recently purchased Groundhog Day and the 4K disc is stunning by comparison. Truly film like and a perfect representation of what the film likely looked like in 1993.

AJG

| IP: Logged

Bobby Henderson
"Ask me about Trajan."

Posts: 10973
From: Lawton, OK, USA
Registered: Apr 2001

posted 03-05-2018 04:26 PM

quote: Scott Jentsch
I thought the bit about halation was quite interesting. I had never heard the term before, but watching the comparison of how the different cameras handled the sequence with the backlit blinds was quite interesting.

Bright light sources and back-lit objects are problem issues in photography. Aside from how halation can happen on the image sensor camera lenses are prone to issues like flaring and chromatic abberation (colored fringing on the edges of bright objects, often appearing green and purple). Very expensive cinema lenses usually handle those problems very well, along with several other issues common to less expensive lenses.

quote: Scott Jentsch
I'm all for high resolution, and I don't necessarily buy all of his arguments that we've reached the practical limit of how many megapixels are necessary to properly resolve detail. However, he does demonstrate some very effective up-rezzing of 2K material to 4K and, I believe, 6K, and the difference wasn't as stunning as one might assume.

There is a pretty tremendous difference between 2K and 4K material, enough so that even many consumer level cameras have no problem capturing 4K still and video images containing detail well above the capacity of 2K to hold. When you get up into the 6K and 8K range the quality of lens optics becomes a serious variable. And that's on top of other features in the camera itself. Video compression, when ramped up to increasing levels of severity, will eat both image detail and color value (chromatic sub-sampling is a big part of squeezing images down to smaller file sizes).

I can agree there is a point of diminishing returns as camera resolutions are pushed higher. But what is that threshold point? I certainly don't think 2K is that threshold at all. Any real life photographic subject has infinite levels of image detail. It doesn't matter if it's an intricate landscape or a human portrait. A camera system as well as our own human eyesight has a capacity to see only so much of that subject's detail and color. Not all cameras, human eyes or viewing situations are equal. What it sounds like here is a way to rationalize doing things just good enough.

Then there's the issue of film formats and film-based projection and the digital formats that have effectively replaced them in both production and exhibition. IMHO, the d-cinema camera systems, movie projection systems or other technology (like LED-based jumbotron style screens) ought to be striving to at least equal what those 35mm and 70mm formats could do. The puny letter-boxed 'scope of 2K d-cinema projection hasn't been a great replacement for 4/35mm 'scope and we don't have anything digital-based that can approach what 70mm could do.

quote: Scott Jentsch
Interestingly enough, Yedlin makes the case for displays to have more resolution than the source material. I agree with his comments that adding resolution for resolution's sake doesn't pay off if it means that other aspects are compromised as a result.

The Bayer pattern pixel grid has its own effect on harming image detail and introducing image artifacts, like moiré on repeating object patterns (fabric weave, fencing) or pixel stair-stepping on thin curving lines (hair strands, electrical power lines). Higher native image resolutions in the pixel grid can avoid some of those problems. Larger, higher resolution monitors can help reveal flaws in imagery or video material being edited without having to zoom in and pixel peep.

quote: Scott Jentsch
I think there's a case to be made for doing visual effects in higher resolutions because they are inherently non-photographic and not affected by all the other characteristics of a camera (sensor, lens, etc.).

From what I can tell it's mostly the digital post production work process that has been forcing so many movies down to 2K. The tyranny of 2K is still very much alive and well. Over the past decade many performance increases have been made computing technology. Many improvements have been made in creative software. Both have made generating 4K content much faster, easier and cheaper. Instead those same improvements are mostly being used to generate 2K material at even faster/cheaper rates. Instead of improving visual quality the productions are opting to boost profit margins instead, pocketing that difference.

quote: Scott Jentsch
Could you tell a noticeable difference between the sample clips that he created? There were some examples of drastically different source resolutions, but very little differences in the end product. I wouldn't call any of them unacceptable, and I was watching it full-screen on a 4K display at a distance of about 2.5 feet.

While Steve Yedlin's presentation is very interesting, the camera footage comparison video in Part 1 isn't what I would consider a fair comparison. Everything is down-sampled into a 1080p resolution MP4 file (that's what the site let me view/download). That alone creates a serious bottleneck for the comparison. I can't really judge how well 6K footage from an Arri Alexa 65 compares to the 2.8K material from an Alexa XT while viewing both samples in mere 1080p resolution.

I've played around with down-sampling and up-sampling still camera imagery to experiment with different methods of enlarging lower resolution image material (such as from customer provided images) for large format printing purposes (billboards, vehicle wraps, etc). Once you down-sample a high resolution image (such as a 5.6K image from a DSLR) down to a level like 2K that down-sampled image is never going to compare well to the original image when re-enlarged at or near the original's size. Enlargement tricks work only so well.

quote: Scott Jentsch
I don't subscribe to the "fake 4K" viewpoint because I think it's too narrowly focused. There have been examples of well-rated UHD titles sourced from 2K material, and I'm sure that there are also mediocre titles sourced from 4K material. So much goes into the production process that I feel it's ultimately most important to look at the end product.

Which 2K blow-ups to 4K have been well-rated? I haven't seen any UHD reviews of movies from 2K sources where the UHD version was a clearly noticeable improvement above the standard Blu-ray. I can understand opting for the UHD version if it contains other unique things (like Atmos/DTS:X, etc). Still the whole format and its prevalence of 2K sourced titles is just asking for trouble. Physical media is having a real tough time just surviving right now. The misleading advertising in UHD Blu-ray only adds to those problems.

| IP: Logged

Scott Jentsch
Phenomenal Film Handler

Posts: 1061
From: New Berlin, WI, USA
Registered: Apr 2003

posted 03-16-2018 07:00 PM

Like I said, my movie viewing happens on a 1920x1080 JVC HD250 sourced by a Samsung UHD player, so everything I watch is not in 4K. That caveat aside...

My understanding of the situation with 4K is that the improvement the people see and appreciate is primarily due to the HDR component. I'm sure that there is little difference between the resolution quality of a movie that was sourced in 2K (because how could there be), but that's the state of technology as it stands right now. I have yet to see HDR in action to judge for myself, and if money were no object, I would be happy to upgrade to one of JVC's e-shift models, as they have gotten very good reviews, even when compared to native 4K devices. As it stands, I have a very hard time justifying buying a new projector when mine is working well. I still find it worthwhile to purchase 4K discs for the future, and to get Atmos/DTS:X soundtracks.

The question that Yedlin was attempting to answer is whether sourcing and displaying in 4K was better than sourcing in 2K and displaying in either 2K or 4K. It would seem the answer he came up with was mostly "no" but he also states that there were other characteristics of the various cameras (and lenses) that would be a larger influence over which he chose for which purpose.

As I was looking into your original question, I did come up with another reference point in the review of Coco 4K by The Digital Bits:

We’ve learned from director Lee Unkrich that Pixar has tested rendering their animation at full native 4K and found that there isn’t enough of a visual benefit vs 2K. Doing so would also increase both the processing power required and rendering time significantly.

http://www.thedigitalbits.com/item/coco-uhd

| IP: Logged

All times are Central (GMT -6:00)

Printer-friendly view of this topic

The Film-Tech Forums are designed for various members related to the cinema industry to express their opinions, viewpoints and testimonials on various products, services and events based upon speculation, personal knowledge and factual information through use, therefore all views represented here allow no liability upon the publishers of this web site and the owners of said views assume no liability for any ill will resulting from these postings. The posts made here are for educational as well as entertainment purposes and as such anyone viewing this portion of the website must accept these views as statements of the author of that opinion and agrees to release the authors from any and all liability.