|
|
This topic comprises 9 pages: 1 2 3 4 5 6 7 8 9
|
|
Author
|
Topic: Tomorrowland aspect ratio 1:2.20
|
|
|
|
|
|
|
Marcel Birgelen
Film God
Posts: 3357
From: Maastricht, Limburg, Netherlands
Registered: Feb 2012
|
posted 06-07-2015 02:58 PM
quote: Bobby Henderson
No, it doesn't. The pixels on all DLP imagers are square. There is no projection hardware that natively supports the display of non-square pixels. When the Super 2K and Super 4K imagery I describe is re-sampled to fit full frame on a 2048 X 1080 imager or a 4096 X 2160 imager all of those pixels are square. None of the d-cinema cameras in use are recording non-square pixels. None of the post production systems are messing around with non-square pixels. In the past the only reason why people needed to mess with non-square pixels was because SD TV sets only used non-square pixels.
Bobby, you're just not understanding what I'm writing... you're essentially saying the same as I am.
The "non square pixel thing" refers to the DCP, NOT the imager, the actual DMD. I'm not expecting ANYBODY to build an actual imager that uses non-square pixels. Yet, what you're doing is saving non square pixels in the DCP. "Square" doesn't actually mean shit for binary formats anyway, it's not like the bits in them are squarish, rectanglish or roundish... Or... let me say it this way: The pixels in the DCP are supposed to be stretched before presented and they will thus end up as non-square pixels on screen...
This "Super 2K" or "Super 4K" thing like you call it, is essentially the same I'm proclaiming for a while now. And the nice thing is, you don't need to upgrade ANYTHING for systems with anamorphics to make it work.
| IP: Logged
|
|
Bobby Henderson
"Ask me about Trajan."
Posts: 10973
From: Lawton, OK, USA
Registered: Apr 2001
|
posted 06-07-2015 05:23 PM
quote:
Bobby, you're just not understanding what I'm writing... you're essentially saying the same as I am.
The "non square pixel thing" refers to the DCP, NOT the imager, the actual DMD.
I don't think I am. Using Super2K as an example, I would expect the imagery in a 'scope DCP to be 2048 X 1080, all square pixels. The imagery's proportions would look squeezed, but the pixels would still be square. Only the anamorphic lens would unsqueeze the image to its proper proportions (but make the pixels non-square on screen).
If the DCP contained a non-square pixel image, I would assume that would be a 2560 X 1080 image for 2K and 5120 X 2160 image for 4K. The DCP file size would have to be signficantly larger to accomodate the larger images, or it would have to be more data compressed. The image would have to be digitally scaled to fit the 1.9:1 ratio of the projector's imager chips. It would end up being digitally squeezed by a flat hardware function and then un-squeezed again by the projector lens. If scaling artifacts were going to be a problem they would more likely occur via a flat hardware function in the projector than if the image was re-sampled from 2560 X 1080 to 2048 X 1080 during post production by more intelligent re-sampling tools.
| IP: Logged
|
|
Marcel Birgelen
Film God
Posts: 3357
From: Maastricht, Limburg, Netherlands
Registered: Feb 2012
|
posted 06-07-2015 06:16 PM
Bobby, what you're describing are "non-square pixels" in the DCP and is exactly what I'm advocating, since pre-production scaling/squeezing/resampling tools can do a better job at scaling from e.g. 2560x1080 to 2K than a projector or server could do real-time. The non-square pixels refer to the fact that they're supposed to be viewed as non-square pixels. In this case, you're going to "stretch" them in appropriate form using an anamorphic lens, so you won't lose any resolution or detail on the projection side of things.
So, if you create a DCP by taking a 2560x1080 source and squeeze it to 2K, or maybe you film something with an 1.25 anamorphic lens, your end result would be a "squeezed DCP", containing "squeezed" and thus "non-square" pixels.
You can feed this to any DCI compliant projection system with anamorphic lenses, without any need for software or hardware modifications. You just need to switch to your anamorphic lens and show this as a full container. This way, there will be no need for any kind of scaling (besides 2K to 4K or vice versa) inside the projector or server. The system will operate at the best possible resolution and best possible light level.
The only "problem" is that the DCI specs officially requires "square pixels" (pixels which are meant to be projected like little squares and not like little rectangles). So, this "Super2K" and "Super4K" would currently not be DCI compliant.
| IP: Logged
|
|
|
|
Bobby Henderson
"Ask me about Trajan."
Posts: 10973
From: Lawton, OK, USA
Registered: Apr 2001
|
posted 06-08-2015 09:43 AM
quote: Marcel Birgelen
So, if you create a DCP by taking a 2560x1080 source and squeeze it to 2K, or maybe you film something with an 1.25 anamorphic lens, your end result would be a "squeezed DCP", containing "squeezed" and thus "non-square" pixels.
No. The pixels never get non-square. Only the imagery takes taken out of normal proportions. When an image is squeezed from 2560 X 1080 image to 2048 X 1080 every 5th column of square pixels is removed. It is not preserving a 2560 X 1080 image in a 2048 X 1080 space. The image is re-sampled to correspond with that more narrow space and map properly to the hardware that will be projecting the image.
Here's some visual examples if you're still not understanding it.
Image #1: This represents the starting 2560 X 1080 image with its content in normal looking proportions.
Image #2: This is the image reduced to 80% of its original width, re-sampled to the fit the 2048 X 1080 full frame space of the d-cinema imager.
Image #3: This is a blow-up from part of that squeezed image. Notice the pixels are still perfectly square.
The only thing that is made "non-square" is the pixels of detail after they are stretched by the anamorphic lens and projected on the screen. There is nothing non-square going on before then.
Nevertheless, this approach is indeed compatible with current d-cinema projection systems. That's really the easy part, despite what theaters would have to pay for anamorphic lenses.
The harder part would be getting movie productions to go along with this approach. They would have to render 'scope images 157.3% the size of the current ones they're producing -over 1 million more pixels for 2K and over 4 million more pixels for 4K. A Super 4K source image would have over 11 million pixels per frame.
I think there is also a race against time element with this. Many of us prefer common height screens. Those are real movie screens. But those kinds of screens are in danger and the 'scope format is in danger in the long term. If too many movie theater screens have the same shape as a home HDTV set movie makers will eventually shift to fill the extra height of those screens. Scope will go bye bye then.
Stadium seated theater designs deserve some of the blame for the popularity of common width with theater operators. You can make what seems like a big screen without building as huge an auditorium; the screen is just taller. But 'scope ends up being smaller. It never seems epic in any of those rooms. IMAX launched the trend of installing these common width screens without any masking or curtains. D-cinema seems like it was tailored for common width screens and flat ratio movies. It's really disappointing Dolby would reinforce the common width trend in their new Dolby Cinema theaters. They had a real opportunity to correct this situation of 'scope being done wrong.
I don't think the trend of theaters building common width screens will be stopped unless there is a way for 'scope to be presented as the best quality, highest resolution option.
| IP: Logged
|
|
Marcel Birgelen
Film God
Posts: 3357
From: Maastricht, Limburg, Netherlands
Registered: Feb 2012
|
posted 06-08-2015 01:31 PM
quote: Bobby Henderson
No. The pixels never get non-square.
Bobby, that's just a definition. In the concept of a DCP, this would be considered non-square pixels. The end result would be non-square pixels on screen and that's the thing that matters. The second picture you posted essentially contains "non square pixels", but because you encoded it as a simple JPEG lacking the necessary metadata uncommon aspect ratio of the pixels, every viewer/renderer/browser will interpret it as square pixels and thus the squeezed result.
And yes, I know you just rescaled that image, but think for a moment. Pixel aspect ratios are nothing new and what you (and I) are proposing here is nothing different.
The "problem" though is, DCI specs currently do not allow those "non square pixels" or rather this "pre-squeezed" content inside the DCPs. They follow most HD TV standards which also try to avoid non-square pixels in the distribution formats.
quote: Bobby Henderson
When an image is squeezed from 2560 X 1080 image to 2048 X 1080 every 5th column of square pixels is removed.
If you do that without some form of averaging, the end result WILL look dreadful. Just like scaling without filtering looks dreadful while projecting a movie. Visible vertical scaling artifacts will be the result, especially in scenes with lots of panning and zooming.
quote: Bobby Henderson
The only thing that is made "non-square" is the pixels of detail after they are stretched by the anamorphic lens and projected on the screen. There is nothing non-square going on before then.
Thus, you could easily state: The pixels were non-square already when you put them in the DCP. The only thing that defines how a pixel should be rendered is standards, conventions and settings. It's not like a file format encodes "pixels with square bits" or something. In the end the uncompressed datastream is just a long string of binary XYZ color values.
| IP: Logged
|
|
Bobby Henderson
"Ask me about Trajan."
Posts: 10973
From: Lawton, OK, USA
Registered: Apr 2001
|
posted 06-08-2015 02:31 PM
quote: Marcel Birgelen
Bobby, that's just a definition. In the concept of a DCP, this would be considered non-square pixels. The end result would be non-square pixels on screen and that's the thing that matters.
Not when the imagery is being captured, modified in post production or encoded to DCP for the exhibition stage. Any use of the term "non-square pixels" anywhere in those areas is dangerous. Mistakes can be made by way of confusion over irresponsible use of terminology.
quote:
The second picture you posted essentially contains "non square pixels", but because you encoded it as a simple JPEG lacking the necessary metadata uncommon aspect ratio of the pixels, every viewer/renderer/browser will interpret it as square pixels and thus the squeezed result.
No. The second image contains square pixels. It contains fewer pixels than the first image, only 80%. Suggesting the pixels are non-square is trying to claim more pixels are there in that specific space. You can check the pixel dimensions/numbers of both images. Those numbers make the pixels nothing other than square. No extra metadata regarding aspect ratio is needed to re-state that.
quote: Marcel Birgelen
If you do that without some form of averaging, the end result WILL look dreadful. Just like scaling without filtering looks dreadful while projecting a movie. Visible vertical scaling artifacts will be the result, especially in scenes with lots of panning and zooming.
I used the description of every 5th column being dropped as a simple way to explain 2560 X 1080 pixels isn't magically mapping into a more narrow 2048 X 1080 space with all those extra pixels intact. Your insistence that non-square pixels are being stored implies such a thing when it is not true. BTW, I used a modified version of Photoshop's bicubic sharper filter to reduce the width of that second image.
This kind of scaling of 2.5K/5K to 2K/4K has to happen on the post production side, prior to DCP authoring in case there are any nasty looking scaling artifacts present in the imagery. If the theater were to receive a 'scope DCP in 2.5K/5K resolution and leave it to hardware to do the scaling chances would be greater to have nasty reduction artifacts hitting the screen.
quote: Marcel Birgelen
Thus, you could easily state: The pixels were non-square already when you put them in the DCP. The only thing that defines how a pixel should be rendered is standards, conventions and settings.
No. You can't state that at all. Both the hardware and software from production to exhibition, and not to mention simple laws of geometry, is defining those pixels to be square. The imagery hitting the d-cinema projector and its 2048 X 1080 chips has to have nothing but square pixels to map to it 1:1. Nothing in the digital realm is making any of it non-square. It is only after the digital image has been converted into analog light that the pixels are getting stretched out of normal, square proportions.
| IP: Logged
|
|
|
|
Marcel Birgelen
Film God
Posts: 3357
From: Maastricht, Limburg, Netherlands
Registered: Feb 2012
|
posted 06-08-2015 03:54 PM
quote: Bobby Henderson
Suggesting the pixels are non-square is trying to claim more pixels are there in that specific space.
Nobody is claiming that there are hidden pixels or whatever... you get the basic concept behind pixel aspect ratios, do you? It's even something that can be configured in e.g. Photoshop. The thing you're describing in an infinite amount of words is just that: Non-square-pixels. You can wrap it in ever so much more complicated explanations, but damn... Every broadcast engineer will call it "Pixel Aspect Ratio" or "Non Square Pixels", whatever is in your signal path doesn't matter, analog, digital, optical... The pixels in the source (in this case, the DCP) are supposed to be stretched to be viewed, so they are non-square by definition... But don't worry, their bits are entirely square though.
quote: Bobby Henderson
This kind of scaling of 2.5K/5K to 2K/4K has to happen on the post production side, prior to DCP authoring in case there are any nasty looking scaling artifacts present in the imagery. If the theater were to receive a 'scope DCP in 2.5K/5K resolution and leave it to hardware to do the scaling chances would be greater to have nasty reduction artifacts hitting the screen.
Exactly my point for this whole "Super4K" thing or whatever you want to call it.
quote: Bobby Henderson
No. You can't state that at all. Both the hardware and software from production to exhibition, and not to mention simple laws of geometry, is defining those pixels to be square. The imagery hitting the d-cinema projector and its 2048 X 1080 chips has to have nothing but square pixels to map to it 1:1. Nothing in the digital realm is making any of it non-square. It is only after the digital image has been converted into analog light that the pixels are getting stretched out of normal, square proportions.
It doesn't matter at all. It's only the end result that counts. Those pixels are not meant to be presented as squares. Their aspect ratio is therefore not 1:1, so those pixels are not square in the DCP. If you present it without stretching it first, you will get very stretchy people... If you stretch it optically, digitally or whatever way you want, doesn't matter for what it's called. Is it so hard to get this?
The reason why you want to optically stretch those non-square pixels, rendered as square pixels on the fully square imager is because you want to avoid nasty digital scaling on the presentation side.
| IP: Logged
|
|
|
|
|
|
|
|
|
|
All times are Central (GMT -6:00)
|
This topic comprises 9 pages: 1 2 3 4 5 6 7 8 9
|
Powered by Infopop Corporation
UBB.classicTM
6.3.1.2
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.
|
Home
Printer-friendly view of this topic