Perry Paolantonio

Why use a screen at all?

When I was in college we set up a very crude aerial image system, where we projected onto a screen to get it as focused as possible. Then facing the projector was a camera that we also focused on that screen. Now remove the screen and like magic, your camera is picking up the image out of thin air because both lenses are focused on the same point in space - no screen necessary.

Mostly we did it because it sounded like BS when someone told us about it and we just wanted to see if it really worked. With the hardware we had back then it looked awful, but it's worth investigating, since it'll completely remove the screen texture from the equation.

You can find more about this in animation and special effects books.

-perry

Merely calling it 1080p and feeding it with a highly compressed stream is not the way it works.

You cannot have proper 1080p from a 6mbps stream. It just like upscaled DVD of 10 years ago.

I understand what you're getting at, but you're bringing a quality judgement into the definition of 1080p that simply doesn't belong there. "HD" does not define the quality of the image - it defines frame rate and picture size. That's all it does, nothing more. Within that image you have a range of possible qualities and that range is broad (HDV vs HDCAM, for instance). But in all cases, if it's 1920x1080 and it's progressive, it's 1080p.

That said, you're incorrect about the quality/bit rate correlation. Yes, 6mbps MPEG2 looks terrible in HD. Yes, 6mbps AVC isn't as good as 20mbps AVC, but it can still look very good. In the early days of DVD, you couldn't encode at lower than 7mbps (SD MPEG2) and have it look good. Now you can encode below 5mbps and have it look just as good or better than the old encoders at 7mbps, because the encoders have matured significantly in the past 15 years. Same with H.264/AVC - the quality of the picture gets better as encoders get better, and that means lower bit rates.

New codecs come into being all the time, in pace with new hardware that can handle it. These codecs all have the same goal: better quality at lower bit rates. H.264 promised to deliver the same quality as MPEG2 at about half the bit rate. And it pretty much does that - maybe it didn't at first, but once the encoders matured it did. H.265 is making similar promises in relation to H.264 - same quality, half the bit rate. But it also supports 10 bit color, which is a massive leap forward in picture quality (goodbye, banding!). Give it a little time and I'm betting you'll see 1080pHD streamed video that's practically indistinguishable from the original, in H.265 at extremely low bit rates.

Have you actually watched Netflix streaming in "SuperHD" mode? (that's their marketing for 1080p at the two highest H.264 bit rates they support - the max being something like 5.85mbps). It's not perfect, but it's pretty damned good looking. From normal viewing distances without any network congestion it can look nearly as good as Blu-ray (this is content dependent, of course).

-perry

The may claim and label their streaminfo as 720 or 1080 or 1080p.

The content is definitely not what they say it is. It is totally impossible to stream 1080p in full resolution.

I.e. they compress with loss and just expand it at the receiving end.

You're talking about different issues here. You can make a 1080p file that's easily streamable on the slowest of broadband connections. It won't look good, but it'll still be 1920x1080 pixels. Yes, it will have lots of compression. But it will still be 1080p, because that number refers to nothing other than the dimensions of the image. "1080p" is not a measure of quality, it's a measure of size.

That said, it's possible to make 1080p look quite good on broadband connections even at relatively low bit rates. At home I don't have cable TV but I do have a typical 20mbps cable modem connection to the internet and a Roku box that can stream 1080p video from a variety of sources. From Netflix (I believe just under 6mbps) it usually looks quite good on a 46" screen from normal viewing distances (about 10 feet away), as long as there aren't network issues that cause the highly compressed versions to kick in. From Amazon it's hit or miss - sometimes it looks great, sometimes the encodes are unwatchable. From Hulu Plus it's usually pretty bad with lots of network problems and outright stalling.

OTA broadcast tends to look pretty great compared to Cable TV, because a single channel has a lot more bandwidth to play in - ATSC MPEG2 is just under 20Mbps, and that means that most content will look good. On Cable TV, you have hundreds of channels competing for the same small pipe, so they all have to be more compressed than the signal that the local broadcaster is sending out over the air.

H.265 is going to up the ante significantly, by the way. I'd say within a year or two we should see much more widespread adoption of that format, now that encoders are starting to appear. The picture quality is pretty amazing even at very low bit rates.

And for what it's worth, bit rate is not in and of itself a measure of quality either - it's a measure of how much data there is. But every codec is different (10Mbps HD MPEG2 looks pretty awful, but 10Mbps HD AVC can look really good - the files are identical sizes, but the compression is different).

-perry

I'm sure there is a practical limit, yes. While we do all of our feature film Blu-ray encoding work from Uncompressed 10 bit files we've done tests on ProRes HQ and 4444 files made from the same HDCAM SR tapes, and we haven't found any difference in quality between the two. ProRes HQ is about 1/10 the size of uncompressed 10 bit files. If you're talking about uploading a large clip to YouTube, then ProRes is a good starting point and is more reasonable than Uncompressed, given the relatively slow upload speeds most of us have. And I don't think you'd see any difference in the final product.

A couple months ago we stopped uploading H.264 files to YouTube and switched to ProRes HQ. Even with H.264 files that looked great on the desktop at high bit rates, we could clearly see the quality difference vs. a ProRes or Uncompressed test clip once YouTube recompressed them.

The bitrate in mbps is for the linespeed. 50 and 20 are hispeeds which are not commoin in the US and neither in many other.s

Again, (and Josh can clear this up), I believe you and I are talking about different things here, Andries.

When one uploads a video to YouTube or Vimeo, that is NOT the file one downloads. Those services re-compress the video to a variety of bit rates, so they can fine-tune the playback depending on network speed issues (check out this video for a good explanation of the concept: http://www.gammaraydigital.com/blog/how-youtube-handles-variable-network-speeds). You have zero control over the bit rate of the resulting file, that's entirely handled by YouTube or Vimeo, and it's totally automated encoding. The only thing you have control over is the quality of the source file you give them. The higher the quality, the better the resulting streaming file.

If you upload a 1080p file that's compressed to H.264 at 10Mbps, and you upload that same file as ProRes (100+Mbps), you will see that the streamed file looks a lot better with the version that was uploaded as ProRes. (I'm just using ProRes as an example here - it could be another high quality format). But the point is that you want to upload the *least compromised* file you possibly can, since you don't have control over the encoding parameters of the file that's actually streamed. One way of doing this is uploading a file that's got a very high bit rate, which theoretically has less compression artifacting, though that really depends on the type of compression in the file you're uploading. A better solution is to get as close to Uncompressed as is practical, but that's tough given slow upload speeds on most internet connections. ProRes is a decent middle ground, which is why we use it for this.

We're not talking about the internet connection speed of the end viewer here at all, it's about the bit rate of the file provided to the streaming service. Higher is usually better.

Are you aware what these 50 or 20 mbps speeds are? In the US most would be glad to have real 10 mbps downstream.

Unless I'm mistaken, Josh is referring to the bit rate of the UPLOADED file, not the bitrate of the file you view on YouTube, after they've compressed it. The quality of the uploaded file makes a massive difference in the final quality of the encodes that YouTube makes. We no longer upload H.264 files if we can help it, we upload ProRes or ProRes HQ. The files are significantly bigger going up and the uploads take forever. But the results are far and away better than when we upload high bit rate H.264 (plus it takes less time to make a ProRes file than an H.264 for us, so it's usually a wash).

Experiment with it - upload the same movie as H.264 and ProRes and you should see the difference pretty clearly. It makes perfect sense too - I mean, we wouldn't make a Blu-ray from an already highly compressed file format and expect it to look good, because it's being re-compressed. YouTube is the same concept, but with more extreme compression.

Garbage in, Garbage out...

I graduated from an art school film department 20 years ago. Even back then, the school was tired of businesses (and even other artists) taking advantage of its students. All anyone wanted was free labor, and the school took a stand on that. They came up with a policy that has stuck to this day: the Career Services department will not accept any "internships" that are unpaid, and payment has to be legit - not just minimum wage.

I absolutely support that policy, because starting a career off (even as an intern) doing work for free usually leads to lower salaries later on. Most people assume what they do isn't worth more, and are afraid to ask for more money even when it's warranted. The culture of free labor (in any industry) makes it hard for people to ask for *any* compensation. Taking advantage of interns just devalues them and reinforces the idea that they should be afraid to ask for what they're worth.

If a business is required to pay a living wage, they're more likely to take the time to have that person do something meaningful, and everyone benefits from that. The point of an internship is to learn something in the real world that a school can't really teach you, but there's no reason to do it for nothing.

This, by the way, is worth a read. It was in the NY Times a few weeks ago: http://www.nytimes.com/2013/10/27/opinion/sunday/slaves-of-the-internet-unite.html

Just to be pedantic:

Usually "telecine" refers to realtime transfer systems that convert film to video (tape or files, but with the inherent limitations of video: pulldown, color space, frame size, etc). They all work in realtime (and typically only in real time), and usually require a color correction system to be somewhere in the middle between the telecine and the target recording format. Think Rank, Shadow, Spirit, and on the low end, machines like the Elmo TRV or those mirror-based setups for transferring home movies off a projector.

A scanner is designed to go from film to digital files, and the speeds at which it does this are non-realtime (faster or slower, depending on the scanner, resolution, etc). The ScanStation is not a telecine - it doesn't even have a video (SDI, that is) output. What these two scanners share in common with telecine is that they're continuous motion (vs. an intermittent motion like you get in some scanners). The Scanity's line sensor is a closer cousin to a telecine, but it's still not the same thing and it's not a fair comparison.

-perry

David - Bring that 16mm film by next time you bring something in! I'd like to add the ScanStation to the mix on that test, and we'd be happy to run a free test scan of it for you.

 

the dynamic range of the Director, with triple flash HDR turned on, is 5.6.

Actually, it's 5.1, according to Lasergraphics.

-perry

I brought 4 rolls of Vision3 200T to Italy this summer. In the US, no problem with TSA hand-checking the film.

In Italy, however, they insisted it go through the machine, and flat out refused to hand check it. He looked at the film, pointed to the speed and said "not 800" then tossed it on the conveyor belt.

It came out fine though - no discernible fogging.

-perry

Hi Dave,

Cool - just let me know when you've got something!

-perry

As some of you may know, we recently acquired a Lasergraphics ScanStation at Gamma Ray Digital.

We're putting together some demo videos, and would like to include as many formats as the scanner can handle. We're all set with Regular 8, Super8 and 16mm, but if you have high quality (sharp, well lit, G-rated) footage in Super 16, Ultra 16 or Max 8, please send me a PM. We're looking for things like landscapes/cityscapes, architectural footage, nature footage, slow motion or timelapse. Basically, footage with relatively little camera motion so that we can easily put titles over it.

In exchange for letting us use your footage on our website and YouTube/Vimeo channels (with credit, of course), we'll scan a couple rolls (50 footers for Max8 or 100 footers for Super/Ultra16) at no charge to 2k DPX, TIFF or ProRes files - your choice. As a thank-you, we'll also give you a deep discount rate on additional scanning. We can only offer this to the first couple of people who send us film in each format, so if you do have this kind of footage, please let me know so we can make arrangements.

Here's an example of one of the videos we've already made:

I hope this doesn't come across as spammy - we're really just looking for some footage to show off the machine and I'm too impatient to wait for the right material to come along!

Thanks!

-perry