This is topic DCP pixel dimensions for aspect ratios narrower than 1.85 in forum Digital Cinema Forum at Film-Tech Forum ARCHIVE.


To visit this topic, use this URL:
https://ft-forum.com/ft/cgi-bin/ubb/ultimatebb.cgi?ubb=get_topic;f=16;t=000900

Posted by Scott Norwood (Member # 30) on 09-23-2011, 12:29 PM:
 
Doing math gives the following:

1260x1080 - 1.17:1 (Movietone)
1440x1080 - 1.33:1 (silent)
1480x1080 - 1.37:1 (Academy)
1793x1080 - 1.66:1 (European widescreen)
1920x1080 - 1.78:1 (HDTV)

Are these values correct? Are they standardized anywhere? Mention of the narrower aspect ratios is curiously absent in the DCI spec. Actually, the spec is looking more and more half-baked every day.

This is all rather important to anyone who is attempting to set masking stops.
 
Posted by Monte L Fullmer (Member # 2797) on 09-23-2011, 12:41 PM:
 
That's what is nice with DCI projectors is the zoom option does when going from flat to scope.

The lens just zoom the image in to fit the masking tabs since the image on the chip is smaller.
 
Posted by Scott Norwood (Member # 30) on 09-23-2011, 12:44 PM:
 
Right, but you would still need to set masking stops for each of the formats listed above, and there is a reasonable chance that they would not be the same as the masking stops for 35mm.

Is there a DCP version of RP40 (err...35PA) that could be used for this purpose?

Edit: this is sort of what I had in mind:

 -

The smallest white rectangle represents Movietone, the next larger rectangle (black) represents 1.33, then 1.37, etc.

Is there a place to get an official version of something like this? Or at least specifications to make it?
 
Posted by Monte L Fullmer (Member # 2797) on 09-23-2011, 01:58 PM:
 
In some DCI projectors, there are actual screen test programs within the unit that one can run via the menu and one program is like the 35PA, but not the same that the end user can run to set and test masking tabs with.

Sometimes I run these programs to check the alignments, along with focus, white light test, et.al.
 
Posted by Brian Guckian (Member # 1678) on 09-23-2011, 02:37 PM:
 
Very good points Scott, and I know that that from day one (on the European side at least) there were attempts to have these ratios incorporated into the DCI Specification.

It may be that since DCI seemed to be framed with the purely commercial theatres in mind, it was felt that just defining 1.85 and 'Scope would suffice.

I believe that there have been moves within the Archival community to have DCI extended to incorporate the other ratios.

Also, it's be interesting to know what work has been done lately within SMPTE for example, in terms of defining new Standards in this area. If anyone could give a general overview on that, it would be very helpful.
 
Posted by Mark Hajducki (Member # 1732) on 09-23-2011, 03:33 PM:
 
Arts Alliance Media made a test card frame for systems they installed.

It had the boundaries of the various aspect ratios shown. It didn't include movietone, and it only had one out of of 1.33 and 1.37.

quote:
Edit: this is sort of what I had in mind:
You would also want to include the boundaries of 2.39 in the frame as well.

quote: Monte L Fullmer
going from flat to scope
With film I always took flat to refer to the non-anamorphic formats, so all (DCI spec) digital ratios are flat.
 
Posted by Scott Norwood (Member # 30) on 09-23-2011, 05:47 PM:
 
quote: Mark Hajducki
Arts Alliance Media made a test card frame for systems they installed.
Any idea what pixel dimensions they used?

quote: Mark Hajducki
You would also want to include the boundaries of 2.39 in the frame as well.
That would be interesting, but not useful, as the DCI specification for "scope" only provides for a letterboxed image within (well, not quite) the 1.85 image area. In most cases a lens shift or zoom (or combination of the two) would be necessary to change formats, anyway. I have no idea why they did this, but I suspect that it will prove to be a mistake.
 
Posted by Monte L Fullmer (Member # 2797) on 09-23-2011, 05:57 PM:
 
deleted...
 
Posted by Mark Hajducki (Member # 1732) on 09-23-2011, 06:00 PM:
 
It has been a while since I saw the test card, but I think it was the full 2048:1080 frame, with 1.85 being one of the ratios.

2.39 was just added as horizontal lines.

I have shown 1.66 and 1.37 films digitally in past employment with preset format commands for the common ratios.

I suspect they went for the multiple aspect ratios as they were supplying the UK Film Council network, which mainly covered art cinemas.
 
Posted by Ron Funderburg (Member # 4561) on 09-23-2011, 06:12 PM:
 
quote: Scott Norwood
That would be interesting, but not useful, as the DCI specification for "scope" only provides for a letterboxed image within (well, not quite) the 1.85 image area.
It was done to have a single lens solution to reduce that particular cost!
 
Posted by Scott Norwood (Member # 30) on 09-23-2011, 07:54 PM:
 
The scope image could still be full height/width anamorphic. A single-lens theatre could then scale it down to the current dimensions. Theatres with anamorphic lenses could show it properly. And, at some point in the future, wider chips could be developed to allow scope DCPs to be shown without scaling or anamorphics. Since a scope image is typically projected larger than a 1.85 image, it needs to have more resolution, not less.
 
Posted by Ron Funderburg (Member # 4561) on 09-23-2011, 08:23 PM:
 
Yup it does that was why it was a poor choice of a place to save cost on a lens or processor chip. then again we aren't the high paid nerds that developed this stuff so who knows!
 
Posted by Monte L Fullmer (Member # 2797) on 09-23-2011, 09:01 PM:
 
and since these units might be replaced with a period of time due to manufacturer cycles and upgrades, why make the effort?

quote: Scott Norwood
it needs to have more resolution, not less.

This is where SONY wins with their 4K LCoS Chipset - with a HD 1713x4096 resolution in scope-being above the 1080P requirements. And Ti will be the same for their DLP 4K chipsets also.

Now, we just need more content in 4K than all of this 2K stuff that doesn't use all of the imagery of the DMD chip.
 
Posted by Steve Guttag (Member # 268) on 09-23-2011, 09:41 PM:
 
Except those fools that leave their 3D lenses in the Sony and only get 2048x858...

Scott, in the Series 1 machines with the .TGA test patterns, there was one called "Framing2.TGA" that had many of the dimensions in it...I don't know if anyone has ported it over to .PNG format...which is the format series 2 machines run with (at least in the DLP world).

However, anyone with a graphics program that can save in .PNG can make their own test chart now.

-Steve
 
Posted by Ron Funderburg (Member # 4561) on 09-23-2011, 09:46 PM:
 
So there is separate lens for the 3d format what about scope and flat on those is it a single lens solution on them?
 
Posted by Brian Guckian (Member # 1678) on 09-24-2011, 02:41 PM:
 
Steve, just tying with Scott's question on standardisation of ratios, can you comment on any recent SMPTE activity in this area? Just curious - thanks
 
Posted by Monte L Fullmer (Member # 2797) on 09-24-2011, 04:23 PM:
 
quote: Ron Funderburg
So there is separate lens for the 3d format. What about scope and flat on those is it a single lens solution on them?
3D basic, basic primer: single lens - each of the frames is flashed three times in a LRLRLR (L-left, R-right) eye through the single lens and polarizer. Sony does the "over/under" trick in having two separate images on the chip, one above the other, that is dedicated per each lens.

Who wins ... prob the viewer who likes 3D that can't tell the difference between the two production formats..or even cares to know how it works.
 
Posted by Brian Guckian (Member # 1678) on 09-26-2011, 05:11 PM:
 
quote: Scott Norwood
Doing math gives the following:

1260x1080 - 1.17:1 (Movietone)
1440x1080 - 1.33:1 (silent)
1480x1080 - 1.37:1 (Academy)
1793x1080 - 1.66:1 (European widescreen)
1920x1080 - 1.78:1 (HDTV)

Are these values correct? Are they standardized anywhere?

Here is some updated information, Scott.

The Commission Supérieure Technique de l’Image et du Son / CST and the Fédération Nationale des Cinémas Français / FNCF published this document on May 5th, 2011:

Technical Guide for the Projection Booth in Digital Cinema

On Page 21 they quote the following values (2K and 4K) for these ratios:

1.33 ........ 1440 x 1080 ......... 2872 x 2160
1.37 ........ 1480 x 1080 ......... 2959 x 2160
1.66 ........ 1792 x 1080 ......... 3585 x 2160
1.78 ........ 1920 x 1080 ......... 3844 x 2160

Perhaps this could presently be treated as a "Standard" in the (to be confirmed) absence of any other relevant Standards documentation (e.g. SMPTE, ISO).

BTW, the website of the European Digital Cinema Forum / EDCF from which the CST / FNCF publication is sourced, contains several other very useful publications (usual disclaimer).

The Numerical SMPTE Standards List gives a range of recent D-Cinema Standards (ST 428-1 thru ST 433; not including related documents), but it is difficult to tell which ones might contain Aspect Ratio information (ST 428-1:2006 D-Cinema Distribution Master – Image Characteristics looks most promising), and more to the point, any values for the narrower ratios.

Unfortunately these are somewhat costly for the average user to obtain.
 
Posted by Scott Norwood (Member # 30) on 09-26-2011, 06:46 PM:
 
Many thanks, Brian!

So, how does this chart look?

 -

If anyone is interested, I'll supply a full-res and/or DCP version.

This was created with the Imagemagick tools and should be simple enough for anyone to modify at will. Here were the commands used to create it:

convert -background black -fill white -font Helvetica -gravity center -size 800 label:'2K aspect ratio test\n\n1.33 (1440x1080)\n1.37 (1480x1080)\n1.66 (1792x1080)\n1.78 (1920x1080)\n1.85 (1998x1080)\n2.39 (2048x858)' -extent 1260x1080 artest.jpg

mogrify -background black -gravity center -extent 1440x1080 artest.jpg

mogrify -background white -gravity center -extent 1480x1080 artest.jpg

mogrify -background black -gravity center -extent 1792x1080 artest.jpg

mogrify -background white -gravity center -extent 1920x1080 artest.jpg

mogrify -background black -gravity center -extent 1998x1080 artest.jpg

mogrify -background white -gravity center -extent 2048x1080 artest.jpg
mogrify -fill white -stroke blue -strokewidth 3 -draw "line 0,111 2048,111" artest.jpg
mogrify -fill white -stroke blue -strokewidth 3 -draw "line 0,969 2048,969" artest.jpg
 
Posted by Steve Guttag (Member # 268) on 09-27-2011, 06:14 AM:
 
Get arrows pointing to the actual line one is trying to hit so you know how close to the "cliff" you are.

Color coordinate said arrows too.
 
Posted by Scott Norwood (Member # 30) on 09-27-2011, 10:51 AM:
 
Good idea on the arrows. How does this look?

 -

Any other suggestions?
 
Posted by Brian Guckian (Member # 1678) on 09-27-2011, 11:24 AM:
 
Great work!

How about colouring the number values on the chart to match the colours of the arrows for easy visual reference?

So - "1.33 (1440 x 1080)" in red, "1.37 (1480 x 1080)" in green, "1.66 (1792 x 1080)" in yellow, and so on.

Numerals that do not stand out well on a black background (blue for example) could be edged in white perhaps.

All you need now is the F-T logo! [Wink]
 
Posted by Scott Norwood (Member # 30) on 09-27-2011, 12:47 PM:
 
Like this?

 -

I tried outlining the blue text with white, but it looks odd when compared to the other colors, If it's not sufficiently visible, that could be added.

Edit: fixed position of corners for scope image.
 
Posted by Brian Guckian (Member # 1678) on 09-27-2011, 02:53 PM:
 
Yes, IMO that works really well, Scott, and noted the fix to the corner arrows for the 'Scope ratio.

It's probably the purple 1.78 ratio numerals that may look a little dark, but of course how the test pattern appears on a computer monitor will be significantly different to how it appears on a cinema screen.

Some field tests of course will iron out any issues that may not be apparent here.

Great work again!
 
Posted by Scott Norwood (Member # 30) on 09-27-2011, 06:23 PM:
 
Here is the DCP. (edit: don't use this one; use the new and improved version below; original version is gone)

File is just under 350MB, so it will take most people a while to download it. It should fit easily on a CD or USB flash drive. It will need to be unzipped and the files will need to be in the root directory of whatever removable media are being used to transfer it to a server.

I have not actually tested this, but the same methods used to create it have created DCPs that work fine on a Dolby server (which is the only one that I have access to). If anyone does test this, please post here and let me know if there are any issues.

Due to large DCP file sizes, it only shows the chart for about eight seconds; presumably, most people will want to freeze-frame it to do whatever needs to be done in the way of masking, etc.

Does anyone want a 4K version? It should be fairly trivial to throw that together if there is any demand.

As always, many thanks to those who offered help and suggestions for this.

[ 10-06-2011, 11:53 AM: Message edited by: Scott Norwood ]
 
Posted by Paul H. Rayton (Member # 1817) on 10-04-2011, 08:23 PM:
 
Hi, Scott -- Downloaded it, and successfully ingested it into my "antique" Quvis Cinema Player. And it plays OK!

Many thanks for this -- I'm not sure exactly when we'll need to use it, but as a venue that plays lots of older movies, both on film and digitally, I'm sure it will come in handy.

FWIW, I regard the Quvis as a really good test of the architecture of any DCP. If things are not formatted **exactly** right, it won't see it and won't load it. This worked perfectly. So, to expropriate one famous line from West Side Story, "Ya done good!".
 
Posted by Scott Norwood (Member # 30) on 10-05-2011, 10:27 AM:
 
Thanks, Paul.

It turns out that the arrows and text were not as clear as they should have been in the first version.

Revised DCP is here

The size is now down to about 200MB.

Let me know if there are any other comments or suggestions. If not, we'll call this the final version and I will ask Brad to add it to the downloads page.

edit: I have tested this on the Dolby server. It works.

[ 10-06-2011, 11:53 AM: Message edited by: Scott Norwood ]
 
Posted by Scott Jentsch (Member # 1681) on 10-06-2011, 09:53 AM:
 
If the blue is too low of a contrast, you may want to try a cyan (RGB: 0, 204, 255). Against black, that's going to be much more readable than pure blue.

Would it help to place a perfect circle in the center to make it easy to judge the aspect ratio visually as well?
 
Posted by System Notices (Member # 2357) on 10-10-2012, 02:11 PM:
 

It has been 370 days since the last post.


 
Posted by Richard May (Member # 2627) on 10-10-2012, 02:11 PM:
 
Great job Scott. Just what we were looking for. One more thing though regarding ratios. We have been getting a lot of 1.89 (2048X1080) movies in. Why is this ratio special? It sucks because our trailers are all 1.85 and you get the few people that bitch about the TINY bit of black on either side.
 
Posted by Jock Blakley (Member # 6552) on 10-10-2012, 05:13 PM:
 
1.89:1 (or 1.90:1 depending on how you chose to round 1.896...) is the "Full Container" ratio, which has appeared for no reason other than that it uses the full resolution of the projector imager.

Are you sure the movies are actually 1.90:1 though? While I often see DCPs with 2048x1080 frame sizes listed, they're actually just 1996x1080 files that have been pillarboxed.

I suspect that this comes from DCP-creation software that refuses to accept that the DCI standards only need one side of the image to meet a specified resolution.
 
Posted by Richard May (Member # 2627) on 10-10-2012, 05:19 PM:
 
You're probably right. They may have come up as 2048, but actually cropped to 1998.
 
Posted by Jock Blakley (Member # 6552) on 10-11-2012, 02:03 AM:
 
I'd suggest that black has been added to the side of the image rather than that the image has been cropped.
 
Posted by Monte L Fullmer (Member # 2797) on 10-11-2012, 03:32 AM:
 
Just zoom up the image to fit the side maskings....if you got people that anal on presentations.....
 




Powered by Infopop Corporation
UBB.classicTM 6.3.1.2