Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Bob Sorel]

If the title question is not clear, let me explain...The "normal" way that video is handled is to maintain constant image width (CIW) as different aspect ratios are presented, though it is never labeled as such. That is, display manufacturers sell tv sets in the 16:9 aspect ratio and maintain CIW so that other aspect ratios will always fit within the borders of the 16:9 screen. Very seldom do we see sets sold in a wider AR, though I am sure that there are some.

Front projector owners, however, VERY often use screens of aspect ratios other than 16:9, with 2.35:1 being the most common (also known as a CIH screen) since 2.35:1  (or 2.40:1) are used and is very near the widest AR normally used in these setups. My question is, is there a way to tell MC that I wish to maintain CIH rather than CIW....If not, this would be an incredible feature for us videoholics who use MC, and though I have no knowledge of coding/programming, it intuitively would seem like a relatively simple idea to implement.

So, is there a way to do this already that I have not found, or alternatively, would the JR team be willing to consider adding this feature? I would think that such a feature would almost be a necessity these days since there are a growing number of titles that change AR on the fly in the course of the movie - easy to do with current CIW, but a real PITA with CIH screens.

[Hendrik]

The behavior typically is neither pure CIH or CIW by default, it's just "use most of the screen area without discarding any image, or distortions". If you were to play old 4:3 video content, the width would get reduced and you would get pillar boxing.

Essentially, if the image is wider then your screen, you get CIW, and if its less wide then your screen, you get CIH. So if you have a 2.35:1 screen and play 16:9 content, what I would expect to happen is the height to fill the entire image, and the width to get adjusted accordingly. Isn't that what you want?
The screen is CIH in such a setup because all content is generally as-wide or narrower then your 2.35:1 screen. But that wouldn't need any special handling and already do just the right thing.

Maybe it would help if you explain the expected behavior.

[mattkhan]

it sounds like you want to use a scope screen and have JRVR zoom black bars away (which is already supported)

[Bob Sorel]

The behavior typically is neither pure CIH or CIW by default, it's just "use most of the screen area without discarding any image". If you were to play old 4:3 video content, the width would get reduced and you would get pillar boxing.

Essentially, if the image is wider then your screen, you get CIW, and if its less wide then your screen, you get CIH. So if you have a 2.35:1 screen and play 16:9 content, what I would expect to happen is the height to fill the entire image, and the width to get adjusted accordingly. Isn't that what you want?
The screen is CIH in such a setup because all content is generally as-wide or narrower then your 2.35:1 screen. But that wouldn't need any special handling and already do just the right thing.

Thanks for the quick reply, Hendrik!

To address your post, no, that is not how a 2.35:1 CIH screen behaves with front projection. If I use the projector's lens zoom to utilize the full height of a 2.35:1 image, then a 16:9 image spills over the top and bottom of the screen projecting content that is clearly off the screen. If I adjust the image for CIH of a 16:9 image, then I get pillar bars on the 2 sides, as expected, BUT if a 2.35:1 image is projected with that mechanical lens zoom, then I get a 2.35:1 image within the 16:9 frame, giving me black bars on all 4 sides (2 pillar bars from the 16:9 frame and top and bottom black bars from the content within that frame). This results in a 2.35:1 image that is way too small and leaves a LOT of screen real estate unused.

After thinking about this more overnight, I realized that the problem is that the frame AR is very often different than the content AR, which is probably just a different way of saying what you just told me. For a CIH algorithm (if "algorithm" is the correct term to use here) to work, then it must be able to detect the content AR and ignore the frame AR, maintaining constant height of the content. I don't know if this is possible, but it certainly would not be as simple as I first thought.

Then there is the matter of subtitles and their placement. Very often subs are partially or wholly in the lower black bar of 2.35:1 content in a 16:9 frame. Totally ignoring 4:3 content, this works fine for CIW setups that have a 16:9 AR, though personally I find it annoying. Once again, for the CIH algorithm to work properly with subs, the subs would need to be moved upwards from the lower black bar to reside at the bottom of the content, as the top and bottom black bars would not exist in such a setup.

This is definitely not as easy as I first thought.

As far as how I deal with things right now, with my current 2.35:1 CIH screen I am forced to create precise mechanical lens zoom presets for each AR in order to obtain the largest images possible with each AR, which in most every case will be to use the entire height and then allow the width to fall where it may. I have come across very lttle content any wider than 2.35:1 (like 2.39:1 or 2.40:1), and when I do I just let the little extra width fall off the sides of the screen into the black velvet border. The biggest problem is with content that changes AR on the fly. I always have to use the lens preset that accomodates the highest AR (usually 16:9). If I use the lens preset for 2.35:1 when the content AR is 2.35:1 then I get to use my entire screen, but when the content AR changes to 16:9 or 2:1, then content gets displayed above and below my screen...very annoying. The best way I have found to deal with this type of AR that changes on the fly (like IMAX titles) is to use the entire screen for 2.35:1 and then blank out everything above and below the screen so that there is no content being displayed when the content expands to 16:9, all performed by the projector. I lose content above and below the screen, but it is the best compromise so far.

Or, I could replace my screen with a 16:9 screen and totally ditch the idea of CIH. The problem there is that my room is much wider than it is tall, so my current 158" 2.35:1 screen would need to be replaced by a 167" 16:9 screen in order to see 2.35:1 content at the same size as I currently see it. The height needed for a 167" 16:9 screen is 87.5" (including frame) and I only have 86" from floor to ceiling.... 

[Hendrik]

That sounds like you are wasting a lot of potential presentation resolution outside of your visible area. I guess what other people do is get an anamorphic lens to solve this, which is a mode that MC can already support now with black bar cropping, with the added benefit that it uses all your potential light in viewable areas.

We could support such a mode where the image is constrained within the center area relatively easily, but you would waste so much presentation space doing that...

[Hendrik]

I made a demonstration image. Is this what you are looking for?

You have the 16:9 normal frame dimensions, but you only see the center 2.35 crop, everything else is outside of your screen (you see green, blue is projected beyond your screen), so you want the active 16:9 image to be in the center of that crop, eg. orange area?

The key for this to work is also that you require black bar detection/cropping to be reliable, similar to using anamorphic lenses. And variable aspect ratio is still not covered until such a time when we have dynamic black bar detection.

[Bob Sorel]

I made a demonstration image. Is this what you are looking for?

You have the 16:9 normal frame dimensions, but you only see the center 2.35 crop, everything else is outside of your screen (you see green, blue is projected beyond your screen), so you want the active 16:9 image to be in the center of that crop, eg. orange area?

Ah, I understand the problem now. When I physically zoom in for 2.35:1 content, things are fine, as I am not losing any resolution from the content, but simply making the pixels larger, right? And then when I physically zoom out to fit the 16:9 image on my 2.35:1 screen, again I am not losing any resolution, as the image is not being cropped at all, but instead I am reducing pixel size. Accomplishing the same physical sizes through the renderer would need to done through cropping rather than physical resizing, again, am I correct?

For my request to be realized, 2.35:1 should be the canvas size with a resolution of 3840X1604 (if I am correct), and the 16:9 image would need to have the full image presented at 3840X2160, but within the canvas borders which is already sized at 3840X1604, an impossibility to do with a renderer as far as I know, as there is no way to change physical pixel size digitally...just ridiculous.

The key for this to work is also that you require black bar detection/cropping to be reliable, similar to using anamorphic lenses. And variable aspect ratio is still not covered until such a time when we have dynamic black bar detection.

Yes, but anamorphic lenses change pixel shape and size to accomplish this, once again something that is probably impossible to do digitally.

Thanks again for engaging me on this...I think I am starting to understand now...

[Bob Sorel]

BTW, not that it matters, but I checked the resolution of the 3 most common aspect ratios and here are the results:

16:9 = 3840X2160
2:1 = 3840X1920
2.35:1 = 3840X1634

so as far as I can tell, all modern content seems to be based on 16:9 CIW, though I would assume that the old 4:3 content (and any other ARs that were less wide than 16:9) would most likely use 2880X2160, thus utilizing the full vertical resolution of the 16:9 AR. Now I understand why you said "The behavior typically is neither pure CIH or CIW by default, it's just "use most of the screen area without discarding any image, or distortions". If you were to play old 4:3 video content, the width would get reduced and you would get pillar boxing."

 I just could not find any 4:3 content in my library, but now I see your point.

[Hendrik]

Ah, I understand the problem now. When I physically zoom in for 2.35:1 content, things are fine, as I am not losing any resolution from the content, but simply making the pixels larger, right? And then when I physically zoom out to fit the 16:9 image on my 2.35:1 screen, again I am not losing any resolution, as the image is not being cropped at all, but instead I am reducing pixel size. Accomplishing the same physical sizes through the renderer would need to done through cropping rather than physical resizing, again, am I correct?

If you physically zoom with the lens settings, you would not lose any resolution, but if the renderer does it, you would, as you literally have to keep parts of the image empty. But if thats the only option available to you its probably fine.

Can you confirm the layout of the image matches about what you have setup?
Blue would be where the black bars are for 2.35 content, and projected outside of your screen. Green is the 2.35 area, and orange is where you want a 16:9 image to be? Basically we're making a virtual screen thats only the green area, and are rendering the image into that as if your screen was 3840X1604 natively.

[Bob Sorel]

If you physically zoom with the lens settings, you would not lose any resolution, but if the renderer does it, you would, as you literally have to keep parts of the image empty. But if thats the only option available to you its probably fine.

Can you confirm the layout of the image matches about what you have setup?
Blue would be where the black bars are for 2.35 content, and projected outside of your screen. Green is the 2.35 area, and orange is where you want a 16:9 image to be? Basically we're making a virtual screen thats only the green area, and are rendering the image into that as if your screen was 3840X1604 natively.

Yes, Hendrik, the positioning is exactly correct as you described it.

The only difference with my physical zoom method is that the 16:9 orange area is full resolution - 3840X2160, while your diagram suggests to me that the 16:9 orange area is 2905X1634 which is, as you noted, reduced resolution. The 2.35:1 green area is 3840X1634 (full resolution), AND, as in your diagram much larger physically than the 16:9 orange area in the center due to the position of the zoom lens.

Please note that I am correcting an error my previous numbers. I will go back and correct the numbers in my previous posts. The only number that is wrong is the vertical resolution of a 2.35:1 image. I had it listed as 1604 but the correct number is 1634.

[Bob Sorel]

For my idea to work in CIH mode, to replace the digital "enlargement" that I get using the zoom lens, the way I see it 16:9 would be shown at full resolution, while the wider formats would need to be upscaled to provide the needed added pixels to increase horizontal size, something that I assume is impossible. The resolutions would be like this:

16:9 = 3840X2160
2:1 = 4320X2160
2.35:1 = 5076X2160

in order to maintain constant image height. First of all, can an Nvidia GPU even do those resolutions? And then, even if the GPU could handle them, how would they get displayed on a 3840X2160 native panel?
Madshi is doing CIH in the Envy, but I have no idea how he is doing it or at what resolutions each AR is being displayed at.

[Hendrik]

You can't increase the resolution, as your projector is very likely not going to support it. The GPU would actually be capable of doing so, there are screens that support 8K resolution afterall.

One thing to remember is that you are also reducing the physical size of the image, so a 16:9 image won't actually render at a lower resolution per inch then a 2.35 video would. You just also have less inches due to the lower width. But overall the image quality wouldn't be substantially impacted.

[nathan_h]

This would be a cool feature, for sure.

Take the recent example of Asteroid City.

In the theaters, it was in two main aspect ratios:

Mostly in 2.35:1.  And then there were short passages of 1.33:1 content.  During the 1.33:1 content, the image was the same height as the 2.35:1 content, but obviously with pillar boxing on the left and right of the image.  (And it makes artistic sense in the movie.  The widescreen stuff is supposed to look like a classic Hollywood windscreen movie from the 1950s.  The 1.33:1 stuff is supposed to look like an old b/w TV show from the 1950s.)

The home video release loses this effect:

They dump the whole thing in a 16:9 container, probably because people would complain about seeing pillar boxing inside of letterboxing, if they were to render the the 1.33:1 content inside of the 2.35 farming.

But for those of us with 2.35:1 screens, we are in a position to render the content the way it was seen in a theater.....if we had an automated way for the software to do that.  Yes, the resolution of the 1.33:1 content would be lower versus what is on the disk. But it would also be smaller in the field of vision, which means I would not be too worried about that.

---

I know some other software does this kind of things, like MadVR, but it would be great to see it available in more places.

[htnut]

Bob,

I have a “Constant Image Area” setup where my screen is actually a 2.00:1 ratio and I employ 4-way masking. Not a project for the impatient LOL.

Anyway I’m not sure if this helps but I found the optimal solution to be (will also work with your CIH setup).

** Always run my PC (desktop) resolution at the display’s native resolution.  Put up a test pattern grid.
** set a projector lens memory at full screen width (image spills above and below the actual screen).  Name:  SCOPE
** set a second lens memory zoomed smaller so the desktop fills the full vertical screen height. Name: 16x9

For variable AR movies (IMAX), you need to use the 16x9 lens memory to not be cropping anything, and endure the “wide” parts having pillarboxing+letterboxing (black bars around all 4 sides).

There are two problems, one is solved:

1.  When I’m “scope” mode, if I use a DVD or BD menu (or watch special features) the video is 16x9 and the top and bottom are cropped (spill over).  I don’t choose to re-set the projector lens memory (it takes too long and in my case I also need to move masking) so I employ the JRiver MC command (/MCC 28038) mapped to a keyboard shortcut - which causes JRiver to scale down the image to fit the height.  As Hendrik pointed out, this method throws away resolution (and brightness!) but that’s not important for navigating menus or watching special features which are usually poor image quality anyway. 

2.  I haven’t yet solved this:  if I want to use the JRiver options (up arrow) the menu bar appears at the bottom - which in “scope” mode is below my screen in the masking.  I can just barely make it out, but it would be better if I could cause that menu to “pop up” higher in the video area.  Note that the MCC command referenced above affects the video area but not the menu bar.

Hope this helps.

[Bob Sorel]

Thanks for the posts, guys!

My lens memories:
1. - 16:9
2. - 2:1
3. - 2.35:1
4. - IMAX

Your solution is very close to what I am doing now, except that I use a fourth lens preset which I labeled "IMAX". It is the "SCOPE" lens memory, but with top and bottom blanking enagaged so that there is no spillover when the AR changes to 16:9 (or 2:1). I fully realize that I am losing content at the top and bottom during the 16:9 scenes, but I find it preferable (most of the time) to having a scope image that is small and boxed in on all 4 sides with black bars. Neither of our solutions is optimal, so I guess we have to pick our poison....    Even if Hendrik were to do what I had asked in my first post, it really isn't the way IMAX material is meant to be presented, as IMAX wants to expand from 2.35:1 to 16:9, while any method that would maintain CIH would need to shrink to 16:9. Your method, while maintaining the IMAX intention of expanding to 16:9, results in a much smaller 2.35:1 image. Like I said, you have to pick the lesser of 2 evils, as there is no perfect solution. In reality, the best solution is to buy a new screen in 16:9 format that is the same width as my current one, but I simply do not have the available height for that. My 158" 2.35:1 screen is the equivalent of a 167" 16:9 screen, and that size of Silver Ticket screen (I am using a Silver Ticket WAB material acoustically transparent screen right now, and the closest they carry is a 165" 16:9 screen). That screen is 87.5" high, including frame, and I only have 86" total height available. I would have to go with a smaller 150" 16:9 screen in order to fit it in my theater, and I am not sure that I want to go that small. I know that I could do DIY screen in order to make one that fits, but at 69 years of age I just don't have the energy to do these projects any more...I'd rather just press the "EASY" button and get something more like a kit, like the Silver Ticket screens. I am considering a custom sized screen from XY Screens (they will make any screen for me for a price), but I wanted to explore the possibility of a renderer maintaining constant height first, as that would be the best thing short of a new screen.

Oh yeah, and for subtitles, I simply move them up 10% using the onscreen MC menu, and then MC remembers that setting the next time I play the movie.

I understand Hendrik's points and I can see that this is nowhere near as easy to do as I first thought, but I still have to wonder how madshi is doing it with his $16k Envy processor. If I understand correctly, the user simply inputs the AR of his screen during the initial setup, and then the renderer automatically fits any and all content ARs so that they fit "best" within the screen AR. I assume that there must be a lot of scaling and black bar detection  going on within the Envy to accomplish this, but it seems to me that even that method would not be perfect, as there would always be some resolution being thrown away.

I think having a custom 16:9 screen built will be the best solution, but it will have to wait until I have the funds.

[Hendrik]

I'll probably add an option to let you limit the parts of the screen you use, including for overlays like subtitles and the OSD, that should at least give you the ability to deal with more cases in your current setup. Of course you are very much required to use black bar detection with this, and for variable aspect ratio movies, it'll likely be quite a bit until dynamic detection would be available.

[Bob Sorel]

Sounds great, Hendrik! Many thanks!! 

[FenceMan]

Unless you do an anamorphic lens on a sled you are always giving up something.

Envy or Lumagen without an anamorphic lens would simply have you overscan your projector so 2.35 fits and then they would downscale 16:9 and other similar AR's to fit within the 2.35 AR which already fitted inside the projectors 16:9 (or 17:9) ratio.  This is good for ease of use but really bad for quality sake.

I use anamorphic lenses and move them out of the way for 1.78 / 1.85 content.

[Bob Sorel]

Unless you do an anamorphic lens on a sled you are always giving up something.

Envy or Lumagen without an anamorphic lens would simply have you overscan your projector so 2.35 fits and then they would downscale 16:9 and other similar AR's to fit within the 2.35 AR which already fitted inside the projectors 16:9 (or 17:9) ratio.  This is good for ease of use but really bad for quality sake.

I use anamorphic lenses and move them out of the way for 1.78 / 1.85 content.

I can't afford an anamorphic lens, but assuming I could, what would happen on IMAX titles? Do you have to keep sliding the lens in and out of place every time that the AR changes? Or more specifically, how do you deal with IMAX titles?

[FenceMan]

I can't afford an anamorphic lens, but assuming I could, what would happen on IMAX titles? Do you have to keep sliding the lens in and out of place every time that the AR changes? Or more specifically, how do you deal with IMAX titles?

AR changes during the movie are the worst - MadVR will switch seamlessly on the fly but with a scope screen the picture gets smaller when it changes to 16:9 (opposite of what you get in an IMAX theatre).  From what I can tell JRVR detects AR shifters as scope and leaves them in that AR (same as what you would get in a non IMAX theatre).

[Bob Sorel]

AR changes during the movie are the worst - MadVR will switch seamlessly on the fly but with a scope screen the picture gets smaller when it changes to 16:9 (opposite of what you get in an IMAX theatre).  From what I can tell JRVR detects AR shifters as scope and leaves them in that AR (same as what you would get in a non IMAX theatre).

Thanks, Fenceman, that's how I am handling those titles now. A few posts back, in answer to HTNut, I said:

Your solution is very close to what I am doing now, except that I use a fourth lens preset which I labeled "IMAX". It is the "SCOPE" lens memory, but with top and bottom blanking enagaged so that there is no spillover when the AR changes to 16:9 (or 2:1). I fully realize that I am losing content at the top and bottom during the 16:9 scenes, but I find it preferable (most of the time) to having a scope image that is small and boxed in on all 4 sides with black bars. Neither of our solutions is optimal, so I guess we have to pick our poison....    Even if Hendrik were to do what I had asked in my first post, it really isn't the way IMAX material is meant to be presented, as IMAX wants to expand from 2.35:1 to 16:9, while any method that would maintain CIH would need to shrink to 16:9.

The best solution with the least amount of concessions, is to use a 16:9 AR screen, assuming you have the room height for it...I don't...