CurtPalme.com Home Theater Forum

bofr · Joined: 03 Feb 2008 Posts: 28

Hi there.
I was trying to find the perfect sweet spot for my projector today and to help
me i made this lookup table holding all the (relevant) 16:9 resolutions that don't need any rounding off.

Thought it might be of use to someone else.

W lines
1280 720
1296 729
1312 738
1328 747
1344 756
1360 765
1376 774
1392 783
1408 792
1424 801
1440 810
1456 819
1472 828
1488 837
1504 846
1520 855
1536 864
1552 873
1568 882
1584 891
1600 900
1616 909
1632 918
1648 927
1664 936
1680 945
1696 954
1712 963
1728 972
1744 981
1760 990
1776 999
1792 1008
1808 1017
1824 1026
1840 1035
1856 1044
1872 1053
1888 1062
1904 1071
1920 1080

CIR Engineering · Posted: Thu Jun 04, 2009 10:36 pm Post subject:

That's actually pretty handy... I was going to do the same thing in a couple weeks.

craigr

J Kildare · Joined: 28 Sep 2006 Posts: 164

Just in time, as the vp50 can't quite do full 1080p at 72 something around 1840x1035 should work within its 165mhz bandwidth.

Thanks
Jim

David_Web · Joined: 02 May 2007 Posts: 418 Location: Sweden

There is no reasoning to decrease horizontal resolution whatsoever. You will only remove possible detail.
Forget about bandwidth as it does not apply to horizontal resolution (in the sense that each pixel needs it's own "window")
Also forget what one on one off patterns show as this rarely occurs in real content.
Think instead that we have a steady state signal that suddenly needs to change state. Like the edge if a cartoon figure or a window.
It will be as sharp regardless of where the transition occur and by deceasing the number of places it can do so you will make the rest of the object blocky. You will also remove content with added "ringing" to increase apparent sharpness. Making it less sharp.

So increasing horizontal resolution will not make the image muddier, instead it will push the projector to it's limit. Not stay below it.
It will look muddy if you have a lot of high frequency content and expect it to resolve.. But it will still be better than to remove it altogether and never have a chance to see it.

Test it for yourself. Take a high resolution picture that has lots of detail. Then experiment with different horizontal resolutions.

It's like lowpassing your speaker at 6k just because it can't resolve a square wave at -3dB any higher. Nevermind the lost detail that it could produce at 16k.

Yep it's a short rant. Something about can't do and bandwidth triggered it. No need to agree, just making a point.

Useful table you have there.

ecrabb · Posted: Fri Jun 05, 2009 3:20 pm Post subject:

David... It's funny you wrote this. I've always thought the same thing. I've never understood the mission to perfectly resolve a pixel or line by throwing away resolution in the source?!!?? I think it's a little misguided, left over from the days of scaled DVD where our graphics and even data projectors had far more resolving power than the source had resolution.

I guess if your objective is to display a computer desktop as sharp as possible, and with as much spatial resolution as possible, then it makes sense to work to find your sweet spot to maximize both. But, if your objective is to display as much of a 1920x1080p source as possible, then it makes no sense at all to throw away a portion of it in the digital domain... just so that the projector can resolves all of the subset of the original source. What's the point?

I, for one, love the "oversampled" look of sending a bit more res to my G70 that it can actually resolve. I can't see scan lines or pixels, and for the most part... It looks a damn lot like 35mm film... and isn't that the point?!?!!?

SC

Kiev Savoie · Joined: 25 Oct 2007 Posts: 432

ok but what about custom resolutions like 800 x 1920 where you get rid of the unused bars on the top and bottom of a 2.35:1 movie? wouldn't that make a difference on your G70 which as you say does not fully resolve 1080p? And wouldn't that also be less demanding of your projector?

MikeEby · Joined: 24 Jun 2007 Posts: 5237 Location: Osceola, Indiana

ecrabb · Posted: Fri Jun 05, 2009 8:34 pm Post subject:

CIR Engineering · Posted: Sun Jun 07, 2009 8:14 pm Post subject:

The resolutions listed above are useful for running 16x9 at 72Hz with a 1080x1920 source. The idea is to crop the top, bottom, left, and right down to a resolution that the video processor can handle at 72Hz (or that the projector can resolve). You loose some resolution by throwing away a little of each side, but you then have more bandwidth to run 72Hz. There is no way to run 72Hz on a Lumagen at 1080x1920 so you must reduce your total pixels enough to fit inside the clock.

Also, you match the input and output resolutions such that you get 1:1 mapping (just like cropping the black bars for 818x1920). I agree that 1:1 mapping is not important for the sake of test patterns. However, what 1:1 mapping ensures is that you are not PROCESSING the image. Avoiding scaling will make a difference in real world images and that is why 1:1 mapping is valuable.

craigr

perisoft · Joined: 29 Aug 2007 Posts: 2920 Location: Ithaca, NY

ecrabb · Posted: Tue Jun 09, 2009 5:47 pm Post subject:

Yeah, that makes sense. Sorta.

Craig, I don't think I understand what you're saying. I get the value of 1:1 scaling - that was kind of my point. But, the res table looked like a *scaling* table to me... I could see cropping 1920x1080 to maybe something like 1856x1044 or Jim's 1840x1035... That's just sort of like "overscan" in the digital domain. That's fine, but it's already a 3-4% area loss. If you're already met the <165mhz Lumagen requirement at 1840x1035, why go any further? You sure as hell wouldn't crop 1920x1080 to 1440x810... So, if the table isn't for scaling, what's that res combo (and all the lower resolutions) for? Looks like it's for scaling to me... which is what I don't see the point in.

SC

bbfarmht · Posted: Tue Jun 09, 2009 5:54 pm Post subject:

ecrabb · Posted: Tue Jun 09, 2009 7:57 pm Post subject:

Maybe it does, Adam... But, I'm not so sure. In fact, that's kind of my point...

What IS the real advantage of running "a few resolutions above 1280x720" on a 6pgxtra? Assuming we're talking about about movies on Blu-ray, you're starting with a 1920x1080p source. Tell me what benefit is provided by scaling down to something like 1328x747 or 1440x810? What's the point of throwing away nearly HALF the resolution in the original source just to stay at some arbitrary resolution where pixels and on/off lines are best resolved? I don't get it! We're not watching test patterns! This whole "sweet spot" thing is really misguided, IMHO.

To me, scaling down to the "sweet spot" is like David's example... It's like saying, "Well, my tweeters can't quite reproduce a square wave at 19 khz, so I'll run all my HD Blu-ray PCM 96khz audio through a resampler to resample it to 36khz before I send it to the amp." Huh?

I guess if I had a 6pgxtra, I'd probably run it at 1080i just like I did my 12xx. Looked awesome. I saw no benefit to finding the exact resolution at which my 12xx best resolved pixels and lines... In fact, I got into CRT so I WOULDN'T see pixels or lines!!! Smile

SC

bofr · Joined: 03 Feb 2008 Posts: 28

Jikes.

Well here is my pov on the subject.

Why would you possibly want to run at a lower resolution than the source material?

* Headaches...yes maybe 1080p48hz looks wonderfull but for some of us a lower res with higher refresh would be a better compromise.

* Tearing / Banding etc. Long cables, bad connections and many more things can behave badly at high frequencies

* Desktop. If you ever have the need to display fine computer graphics (ex. small text) going with a lower resolution than the one normally used when watching movies will give you a sharper display. Sure you might simply choose one of the normal resulolutions exposed by the graphics-card driver but most of them are 4:3 which might be a bit difficult to fit on your normal screen..

I used the table to find a tolerable resolution when connected to my main computer in another room.
Now I use a custom made htpc with the shortest cables possible (sitting directly under the projector), this made for a much cleaner signal and I can now run it at 1080p65hz.

Still..I sometimes have to connect my main computer and whenever I do that i run at 1440x810.

ecrabb · Posted: Tue Jun 09, 2009 9:25 pm Post subject:

Thanks for your reply, bofr! Those are all good points... I don't think any of them are directly related to optimizing image quality (which I think is the main reason typically given for "sweet spot" optimization), but they are all absolutely valid concerns... I totally agree RE: the computer desktop issue, which is why I mentioned it, too. Good points, though.

Just so I'm clear, I'm not intending to be argumentative or critical of anybody for choosing to set up their machines however they please... I'm just interested in having the discussion, learning, and debating to fully understand the "why" and "how" of the way we set this machines up. It seems like the "sweet spot" method is a bit of a holdover from the SD upscaling days, and I'm not sure it still makes a lot of sense... So, I'm interested in discussing it.

Carry on. Wink

SC

Mark_A_W · Posted: Wed Jun 10, 2009 2:22 am Post subject:

I'm with you SC. I run 1920x1080 as it make no sense to chuck res away before the projector.

HOWEVER, if you want to run 72hz, due to deflection deficiencies in the XG, I think you have to drop the resolution. You get ringing on the LHS if you push too far.

My compromise is occasional field line visibility at 96hz interlaced, as opposed to judder at 60hz or flicker at 48hz.

But I tried 1920x804p at 72hz and I could still see the scanlines. They were just as visible as the field lines at 96hz interlaced.