Jump to content

Underexposure Pixelation: Why can't black just be Black? (Film vs Digital)


Recommended Posts

While doing some low light test on my new camera, I noticed a phenomenon that when the area falls off certain degree of underexposure, it renders into this pixelated blocks, and clutters. I just happens to include heavy shadows, in my composition with high light ratios. So when i saw it, I really freaked out by how awful it looks, especially when the camera moves, the pixelated clutters/ blocks moving along with it. 

Then I went to check a few movie files, and it seems that all of them have the same problem, whether it is shot on film or digital. Only difference this phenomenon of the clutter is less patterned in film, as opposed to digital things looks like moving in rectangular blocks. 

https://ibb.co/g679NMx
https://ibb.co/ypbbWbS
https://ibb.co/N26mz0N

https://ibb.co/QK55QTX
https://ibb.co/HLS06Mg

Shot on film: less pattern blocks
https://ibb.co/4s5xdrx

Why and How!?!? Is there a way to avoid it? 

I have seen these films many many times, and I guess the quality of the story had camouflaged this problem. But now I can not unsee it and is becoming an obsession to avoid it. Why can't black just be freaking black!!! 

 

PS: one sequence in The Leopard, with heavy dark velvet suits appears to be free of this problem. I wonder why? 

Zero such problem (why?)
https://ibb.co/b37dnwR

Edited by Wendy Sanders McDonlad
reformat
Link to post
Share on other sites

Let's look at that first shot in extra-contrast-o-vision, so it's really obvious what we're talking about.image.thumb.png.8a57b9ceeacbad34c194cbb29c8ff1d3.png

What you're seeing there is compression.

Uncompressed images are enormous. I could do the mathematics, but high definition images as sent over the network by services like Netflix would take up an impossibly huge amount of data if they weren't compressed.

The mathematical compression uses various techniques to reduce the amount of data. Different systems use different approaches, but most of them break the image up into blocks and treat them individually. Most of them, for instance, try to figure out if it can estimate part of an upcoming frame just by moving those blocks around. That's sort of what you get when the camera pans across a static scene.

Other techniques include just filling a block with a solid colour, which often you get in very dark areas of the scene, because there's not much actual image data to compress.

This is why we like low compression ratios; if we ask the mathematics to  reduce the data rate only slightly, the amount of compression artefacts that we see is small. What you're looking at there is presumably something that's been compressed for YouTube or Netflix, or a blu-ray disc or something, which requires quite a lot of compression. In my experience Netflix generally looks pretty good, they're quite generous on data rate, but that can vary depending on the situation.

If it's really objectionable on lots of content from different services, it may be that your monitor is displaying shadow areas much brighter than they should be. This is hard to diagnose remotely, but maybe punch up the same video on your phone and the TV and compare the two. Neither is likely to be amazingly accurate from a technical point of view, but if they look wildly different then that may be an indicator of where the problem lies.

  • Like 1
  • Upvote 1
Link to post
Share on other sites

The problem is I shot RAW on my BMPCC 4k, and RAW stills images on my Canon 5D MK II... .and this problem still exists. what are the formats that will eliminate this problem? 

 

I mean when we export the footage from whatever editing softwares, compressions are commonplace, so does that mean there is no way to avoid it? 

Link to post
Share on other sites

You can't really avoid compression when you distribute your finished work. If it's too visible, you're possibly doing something wrong, but as I say, verify it's a problem on all displays. Everything looks bad if the black level is too high.

Things to look at:

- Use a higher bitrate, up to the maximum allowed by the delivery service you're using

- Use a better encoder, particularly selecting two-pass or multi-pass encoding where it's available.

- Use a better codec, if your delivery service allows it. H.264 is less good than H.265/HEVC, etc.

- Shoot with low or no compression (which you're doing) and avoid recompression during postproduction.

 

  • Like 1
  • Upvote 1
Link to post
Share on other sites
Posted (edited)

That's why DCPs use JPEG2000;  wavelet encoding rather than Discrete Cosine Transforms (DCT block encoding). 

The problem is, you can encode however you like, but once it leaves your hands, it will be re-encoded and displayed probably using a DCT type encoding to make the signal small enough to transmit over the Net or on a portable medium.

This is no different than making a perfect 35mm print and shipping it to Uncle Joe's cinema; cobbled together with bailing wire and spit.  We are at the mercy of the exhibitor and the limits of their/our exhibition equipment.

Edited by Frank Wylie
  • Upvote 3
Link to post
Share on other sites

The shite-storm over the last season of Game of Thrones was an unfortunate example of this - the compression artifacts were so extreme it was unwatchable. I recently re-watched the entire series on 4k UHD Blu-ray Dolby vision and it looked incredible. Most streaming services are prone to this, Netflix and Amazon 4k UHD are the exception. Even 4k uploads to YouTube can have significant banding.

As far as camera files, especially properly exposed RAW I agree if your viewing on a decent monitor with as little compression as possible this should be a non issue.

Phil has some good advice on encoding. That and Vimeo can produce better results.

  • Like 1
Link to post
Share on other sites
18 hours ago, Wendy Sanders McDonlad said:

The problem is I shot RAW on my BMPCC 4k, and RAW stills images on my Canon 5D MK II... .and this problem still exists.

It doesn't. You're probably describing output or preview files based on those RAW inputs.
As long as you shoot uncompressed or lossless RAW, compressed blocks of pixels are not something that exists in your image, their existence depends solely on your export settings — and in that regard you should follow Phil's advice above.

  • Upvote 1
Link to post
Share on other sites
Just now, Tomasz Brodecki said:

It doesn't. You're probably describing output or preview files based on those RAW inputs.
As long as you shoot uncompressed or lossless RAW, compressed blocks of pixels are not something that exists in your image, their existence depends solely on your export settings — and in that regard you should follow Phil's advice above.

Yes, I meant the noises in darker areas.

There was no compression blocks, yes. 

Link to post
Share on other sites

It's hard when the compression noise is in your camera original and then exacerbated by the compression used by streaming providers. 

Link to post
Share on other sites

OP...that is what separates digital from film. Film does not try as hard to make sense of deep shadows. Digital tries to make sense of it with noise sometimes. 

If it is a still print you can fix the poor blacks with contrast grading. (dodging and burning) But I don't think it is practical with movies. Maybe they have AI software that can / will be able to do it?

Another issue with digital BW is it has too many shades of gray and it sometimes looks kinda plasticy or artificial. 

Edited by Daniel D. Teoli Jr.
Link to post
Share on other sites

Broadcast/streaming compression artifacts in dark areas are a problem whether you shoot film or digital -- if anything, it's worse with film due to grain.

  • Like 1
  • Upvote 1
Link to post
Share on other sites
38 minutes ago, Wendy Sanders McDonlad said:

It's got black all over, but absolutely no pixelation. 

In that screenshot you shared, shadow detail was completely replaced with absolute black, tossing away any information within almost half of the frame, histogram below
image.png.2ad1c905e51a380358e3ac2b841f55b5.png

(pasting a boosted version below to highlight the issue)

image.png.0cea459cefcac6a9bf887d2e533bbb5d.png

  • Upvote 1
Link to post
Share on other sites
3 hours ago, Wendy Sanders McDonlad said:

could someone explain why the last photo from the Leopard has zero such issue? It's got black all over, but absolutely no pixelation. 

It really depends on the quality of the source that you pulled the image from. If it’s a TIFF frame grab from a 4K Blu-Ray, then the compression will be much less than a jpeg frame grab from a web stream. Hence the additional compression artifacts, or lack thereof. 

Link to post
Share on other sites
14 hours ago, Tomasz Brodecki said:

In that screenshot you shared, shadow detail was completely replaced with absolute black, tossing away any information within almost half of the frame, histogram below
image.png.2ad1c905e51a380358e3ac2b841f55b5.png

(pasting a boosted version below to highlight the issue)

image.png.0cea459cefcac6a9bf887d2e533bbb5d.png

yes, but why aren't the other images toss away and replace with black, instead of the ugly clusters. 

Link to post
Share on other sites
11 hours ago, Satsuki Murashige said:

It really depends on the quality of the source that you pulled the image from. If it’s a TIFF frame grab from a 4K Blu-Ray, then the compression will be much less than a jpeg frame grab from a web stream. Hence the additional compression artifacts, or lack thereof. 

but it is just a mac screenshot from a ripped DVD9. Maybe this particular copy is especially good? I skipped to other scenes with blacks... they look consistent with this one. Magical...

Link to post
Share on other sites
11 minutes ago, Wendy Sanders McDonlad said:

yes, but why aren't the other images toss away and replace with black, instead of the ugly clusters. 

In your other examples, there was shadow detail that got deteriorated because of the compressed output.
In this example, there was no shadow detail whatsoever.

As for the "why" — any and every step of the pipeline could have contributed to this result — from a potentially underexposed original negative, through timing, printing, scanning and eventually preparing this digital copy that you're sharing (with its own bitdepth and bitrate limitations).
  

8 minutes ago, Wendy Sanders McDonlad said:

Maybe this particular copy is especially good?


Usually when a copy throws away half the potential dynamic range of the scene, we don't call it "good", but I understand that everything is subjective.

 

Edited by Tomasz Brodecki
  • Upvote 1
Link to post
Share on other sites
17 hours ago, Tomasz Brodecki said:

In that screenshot you shared, shadow detail was completely replaced with absolute black, tossing away any information within almost half of the frame, histogram below
image.png.2ad1c905e51a380358e3ac2b841f55b5.png

(pasting a boosted version below to highlight the issue)

image.png.0cea459cefcac6a9bf887d2e533bbb5d.png

This is true.

At the same time, it is obviously possible to have shadow detail with minimal compression artifacts in home video releases, as long as the bit rate is reasonable. 

 

Link to post
Share on other sites
2 hours ago, Wendy Sanders McDonlad said:

but it is just a mac screenshot from a ripped DVD9. Maybe this particular copy is especially good? I skipped to other scenes with blacks... they look consistent with this one. Magical...

It’s not magical - there is no shadow detail, so there is nothing to compress. That’s fine if there’s no important detail there to begin with, but it won’t suit every film - films like ‘Her’, ‘Arrival’ or ‘Panic Room’ wouldn’t look as intended with crushed blacks, so it’s not a cure-all.

Depending on your disc, the film may have been transferred from a print or some other high contrast source, which would explain the lack of shadow detail. Or it was simply graded to look like a Technicolor print.

Link to post
Share on other sites

You know film is film and digital is digital. Film has organic grain. Added digital grain is not organic.

Now, this is an example of the digital plasticy look. No matter what you do it, it looks like digital. 

Click to view them...

campsite-jersey-city-2016-daniel-d-teoli

That is where I camped out in Jersey City when working in NYC before the corrosive virus hit. You can improve things if you use a big aperture for low depth of field. Although this was shot with a wide angle.

Sometimes you can get a good digital shot that looks like film...if you are lucky. Here is an example with a wide open 50mm, maybe 1.5, can't remember. I also added some digital grain. If you can tone it slightly, so much the better for trying for the film look.

social-documentary-copyright-daniel-d-te

Both were shot with a true monochrome sensor.

Here is a film shot from back in the early 70's. I snuck into a bar near skid row in L.A. when I was 17. 

12-img033-2-2-spot-mr.jpg

Here is an example of a 100% analog silver print  (L) versus a digital scan and digital print. (R)

left-silver-gelatin-print-right-hahnemue

Of course, working with photo images is different than the movies. But still, it gives you an idea.

Edited by Daniel D. Teoli Jr.
Link to post
Share on other sites

I believe photographers (film or digital), filmmakers/HD movie makers (digimakers) should just embrace their format and not try and make it look like something else...  I also believe that If you shoot on film, call yourself a filmmaker. If you shoot on HD, call yourself an HD or digital movie maker (digimaker). There is a difference between the two formats. Portray yourself accurately because people within the movie/ photography business and consumers have the right to know what they are watching and paying for.

  • Upvote 1
  • Downvote 2
Link to post
Share on other sites
On 4/11/2021 at 12:13 AM, David Mullen ASC said:

Broadcast/streaming compression artifacts in dark areas are a problem whether you shoot film or digital -- if anything, it's worse with film due to grain.

Careful don't let scientific truth get in the way of film being better for everything ..  🙂 

Link to post
Share on other sites
9 hours ago, Lance Lucero said:

I believe photographers (film or digital), filmmakers/HD movie makers (digimakers) should just embrace their format and not try and make it look like something else...  I also believe that If you shoot on film, call yourself a filmmaker. If you shoot on HD, call yourself an HD or digital movie maker (digimaker). There is a difference between the two formats. Portray yourself accurately because people within the movie/ photography business and consumers have the right to know what they are watching and paying for.

So Roger Deakins is not a DoP when he shoots on an Alexa ..  or you are joking yes ? please say you are .. 

Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...