Time to define an alternative to ANSI contrast? - Page 11 - AVS Forum
Forum Jump: 
Reply
 
Thread Tools
post #301 of 341 Old 01-29-2007, 12:22 PM
AVS Special Member
 
gregr's Avatar
 
Join Date: Nov 2001
Posts: 2,127
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by GlenC View Post

If your DI on/off CR is 5000:1 ....

The DI is not going to be that low, but you wouldn't know that from the patterns and charts that were used here. That's the point.

Greg Rogers
Video Engineer/Product Designer

gregr is offline  
Sponsored Links
Advertisement
 
post #302 of 341 Old 01-29-2007, 12:31 PM
AVS Club Gold
 
GlenC's Avatar
 
Join Date: Jun 2003
Location: Los Angeles - Whittier, CA
Posts: 2,555
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by gregr View Post

Yep, those numbers suggest that Projector A probably has more than twice the full-field contrast ratio of B. So you wouldn't need any patterns to tell you that A will measure better at the lowest luma levels. Had you got those numbers with the checkerboard %luminance patterns the full-field contrast ratios would have been around 7000:1 vs 3000:1 for the two projectors. And you would know that projector B would measure better when the pattern was about 15% or higher.

At much lower levels it will provide some information about how DI projectors perform vs non-DI projectors.

But isn't this expressly clear in the results of the Sony VLP-VW50 in the spreadsheet? On/off = 4553:1, Iris 1 and 3108:1 no Iris, 46% improvement for on/off but only 16% for the 1% APL to 3% at the 5% APL.

Glen Carter
Home Theater Calibration

To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.
GlenC is offline  
post #303 of 341 Old 01-29-2007, 01:03 PM
AVS Special Member
 
gregr's Avatar
 
Join Date: Nov 2001
Posts: 2,127
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by GlenC View Post

But isn't this expressly clear in the results of the Sony VLP-VW50 in the spreadsheet? On/off = 4553:1, Iris 1 and 3108:1 no Iris, 46% improvement for on/off but only 16% for the 1% APL to 3% at the 5% APL.

I don't really understand your last sentence? But the VPL-VW50 full-field (on-off) contrast ratio in the DI iris mode was over 11,000:1 on my sample. That is its real ability to produce contrast and remove haze and veiling in dark scenes. I don't see any number close to that in your sentence. That's the problem.

Greg Rogers
Video Engineer/Product Designer

gregr is offline  
post #304 of 341 Old 01-29-2007, 02:16 PM - Thread Starter
AVS Club Gold
 
Mark Petersen's Avatar
 
Join Date: Dec 2000
Posts: 4,671
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 11
I'm finally caught up on this thread and thought that I would propose the idea to everyone that we hold off for a short time on the detailed technical discussions related to DI characterization and focus on some procedural issues that are needed to clarify the direction of the project. I think it's important that we try to summarize:
  • The goals of the project (again).
  • Review the work that has been achieved so far in the first two stages.
  • Discuss how to improve our working relationship.
  • Discuss the next step.
Please think of these points as a platform that we need for taking the next step. As the thread starter, what I would like people to do is take a few moments and read through this post and then comment on what has been posted. I'd like to get all of those who have been contributing recently to discuss these steps and when everyone is in agreement we can then move forward. If you don't agree with this platform or the need for it or don't want to go along with it then voice that opinion but be willing to defend your position while also providing an alternative platform.

Let's also agree to use the term static contrast to mean instantaneous contrast that does not involve dynamic gamma adjustments. This term applies to native projectors, fixed iris projectors and can also apply (in the correct situations) to dynamic iris projectors. Let's use the term dynamic contrast to refer to contrast that changes due to the combination of dynamic gamma and dynamic iris (typical DI behavior such as Sony's). Let's also use the term instantaneous contrast to mean mixed contrast that is available at one time (unlike sequential). I think with the possible exception of instantaneous contrast that this is Darin's terminology and it seems to describe things well so let's try and be consistent and use it from here on out.

The Goals of the Project
I think everyone is in agreement that this project is not attempting to replace the ANSI and sequential contrast standards. Rather than replacing these standards what we are attempting to do is to supplement these standards with additional tools that help to quantify contrast and hopefully also allow people to better visualize and understand these metrics. I'm going to change the thread title to better capture this title.

Review of the Work So far
There has been discussions that the work achieved so far is essentially meaningless because contrast vs 0-100% luminance is something that can be calculated from on/off and ANSI metrics. I beg to differ on this point for the following reasons:
  • As stranger89 correctly stated, static contrast with a DI isn't something that falls out of sequential on/off. Interestingly enough, it has worked out that this value that can be closely approximated with the work done so far (using the contrast at 0% APL definition). I think this alone is of value and it's also something that I didn't anticipate when going through this process and it's an example where going through a process can yield benefits that aren't initially seen.
  • This work has also provided some metrics for the relative change in white level vs luminance as well as relative change in black level vs luminance.
  • It has also provided metrics on the static contrast benefits of reducing the iris aperture.
  • Finally as Chris Wiggles pointed out early on there is always the chance that reality doesn't mesh with theory. Performing these measurements has given us a chance to compare theory with reality and it also allows us to identify and avoid problems in the future when we try to apply these techniques to other technologies.
  • A very good point was raised that this stage of the project unfairly depicted DI projectors because it doesn't account for dynamic contrast. Even though this phase of the project was not intended to capture dynamic contrast, I think that this is fair criticism and it's something that I'm in the process of correcting by adding a section defining static contrast and also changing the names of the graphs so that it's clear what is and is not being measured. I want to caution though that just because one part of the project is unclear, mistaken or completely wrong (not that it was in this case) does not invalidate the whole project and people should try to keep their criticism focused on the part that is wrong versus the whole.

There was also the issue raised on why we should use these particular test patterns generated by Wm and not a traditional 4x4 test pattern that achieves APL by varying luminance % by using grey whites rather than different sized boxes with full whites. Here are my thoughts on this:
  • The current test patterns do a fine job of characterizing static contrast which was their original purpose.
  • A lot of work has gone into these designs so far and some of the measurements, spreadhseets, data collection and methodology gets dumped if we move to new test patterns. If the current work done so far keeps us from achieving our goals them we must do what we need to (including abandoning work), but I think we should only do this if there is very compelling reasons to do so.
  • One of the initial goals of this project was to involve the AVS community in getting them to submit data on their projectors. As such keeping them as easy to use and requiring as little HW as possible is a goal. These inital test patterns require nothing more than either a PC or a HD-DVD/BD player and a light meter.
  • As pointed out it does not allow for characterization of dynamic gamma and therefore dynamic contrast with DI's, so if we decide to make DI characterization a goal then this is something we should (and have been) debating.
  • If we decide the next step is to characterize DI performance then it's clear that these 0-100% test patterns will not test dynamic gamma so they can't be used.
  • If we decide to come up with a different set of test patterns and methodology for testing dynamic contrast then it isn't clear that the initial set of test patterns need to be replaced, but rather supplemented.

Discuss how to improve our working relationship.


I want to reinforce jsaliga's comments about the need to not let attacks get personal. One way to achieve this is by avoiding sarcasm completely as this doesn't come off well in posts and it reduces the readers resolve in trying to see a persons point of view.

Another thing I would caution against is shooting down a persons complete argument when part of it is either incomplete or flawed. I saw this a few times in analogies being bandied around and it's important for people to try and not dismiss an idea because of a flaw, but to see past the flaw and try to understand the argument.

The other thing I noticed is that we could better use e-mail in some of the discussions. Hopefully people are okay with pm'ing others with their e-mail address. E-mail is a good mechanism for discussions that don't need to involve everyone and it helps to keep the forum focused on key issues. It has the potential on the otherhand of keeping people out of the loop on important issues and discussions so I wouldn't use it if something has benefit the group only if it's mostly beneficial to one (or a few).

The Next Step
There has been much discussion recently on methods for measuring DI performance even though this wasn't the initial goal of the project. If it ends up that we decide against measuring DI performance then a lot of this discussion was for naught. I for one think however that all of this discussion is uncovering a desire in everyone to measure DI performance and hopefully make this the goal in the next step of the project. So I would like to propose that we all agree to conclude the static contrast results of the first stage of the project (assuming that people see no issues with what has been done so far) and then to agree that the third step of the project is to tackle dynamic contrast.

Whatever people collectively decide to do for the next phase, I would urge people to try and stay focused on this phase and not let small issues sidetrack the discussion. Divide and conquer!


To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.
Mark Petersen is offline  
post #305 of 341 Old 02-01-2007, 03:13 PM
AVS Addicted Member
 
darinp2's Avatar
 
Join Date: Sep 2003
Location: Seattle, WA
Posts: 21,174
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 6 Post(s)
Liked: 54
Mark,

I would vote that you continue measuring and graphing the test patterns with the reduced spatial picture level (less bright area, but still 100 IRE), but also include a separate graph of the 4x4 checkerboards with the brighter rectangles at different levels like Greg mentioned. That one could just say estimated from say a 2.4 gamma and Erik's calculator could be used for each projector. While I don't think that would give you 100% accuracy, I think it would be close enough for this. This way you would have one set that tends to discriminate against DI projectors, with another that doesn't really discriminate against them and shows one of their main strengths.

--Darin

This is the AV Science Forum. Please don't be gullible and please do remember the saying, "Fool me once, shame on you. Fool me twice, shame on me."
darinp2 is offline  
post #306 of 341 Old 02-01-2007, 04:20 PM
AVS Club Gold
 
GlenC's Avatar
 
Join Date: Jun 2003
Location: Los Angeles - Whittier, CA
Posts: 2,555
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by gregr View Post

I don't really understand your last sentence? But the VPL-VW50 full-field (on-off) contrast ratio in the DI iris mode was over 11,000:1 on my sample. That is its real ability to produce contrast and remove haze and veiling in dark scenes. I don't see any number close to that in your sentence. That's the problem.

The numbers came from the spreadsheet here ,from the AVS Contrast Project v2.0 (updated 1/26/07) thread.

Glen Carter
Home Theater Calibration

To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.
GlenC is offline  
post #307 of 341 Old 02-01-2007, 04:59 PM
AVS Addicted Member
 
stanger89's Avatar
 
Join Date: Nov 2002
Location: Marion, IA
Posts: 17,407
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 119 Post(s)
Liked: 144
Quote:
Originally Posted by darinp2 View Post

Mark,

I would vote that you continue measuring and graphing the test patterns with the reduced spatial picture level (less bright area, but still 100 IRE), but also include a separate graph of the 4x4 checkerboards with the brighter rectangles at different levels like Greg mentioned.

Might I suggest, that to maintain some level of consistency within this project, that the same "style" of test patterns be used for both spacial and lum-varying patterns? For example, keep the current patterns, and augment them with one using the same layout as the 50% pattern, but vary the %lum of the boxes on the pattern.

Then, at least at 50% APL (100% lum boxes), there would be a 100% match between the two sets of measurements.

Also (and I'm not sure how practical/accurate it would be) but assuming a set of patterns were generated with a range of %lum boxes, would it not be possible to determine the rough gamma of the projector? I suppose that would fall apart for DI projectors with their dynamic gamma.

See what an anamorphoscopic lens can do,
To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.
stanger89 is offline  
post #308 of 341 Old 02-01-2007, 05:25 PM
AVS Addicted Member
 
darinp2's Avatar
 
Join Date: Sep 2003
Location: Seattle, WA
Posts: 21,174
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 6 Post(s)
Liked: 54
Quote:
Originally Posted by stanger89 View Post

Might I suggest, that to maintain some level of consistency within this project, that the same "style" of test patterns be used for both spacial and lum-varying patterns? For example, keep the current patterns, and augment them with one using the same layout as the 50% pattern, but vary the %lum of the boxes on the pattern.

I thought about suggesting the same thing, but with the other direction. Since the ANSI CR 4x4 checkerboard is a standard, start with it and vary the %stim for the brighter rectanges for one group and the size of the brighter rectangles for the other group. That would mean 8 bright rectangles in each pattern, but just of different sizes. I figured people wouldn't want to go back and start over with new spatial patterns though.

One advantage to using the 4x4 checkerboard for the test that varies luminance instead of starting with the 50% pattern here and varying luminance of the peak part is that we have ANSI CR and on/off CR for many projectors already. New measurements wouldn't have to be taken and Erik's calculator can just estimate the values. Here I am assuming that the new 50% pattern isn't using the ANSI CR 4x4 checkerboard, but I haven't looked at what that 50% pattern looks like.

BTW: I meant to answer Greg's question about where the contrast calculator is earlier. It is here:

http://home1.gte.net/res18h39/contrast.htm

--Darin

This is the AV Science Forum. Please don't be gullible and please do remember the saying, "Fool me once, shame on you. Fool me twice, shame on me."
darinp2 is offline  
post #309 of 341 Old 02-01-2007, 05:46 PM
AVS Club Gold
 
GlenC's Avatar
 
Join Date: Jun 2003
Location: Los Angeles - Whittier, CA
Posts: 2,555
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
One issue I see, when you start dealing with varying luminance levels is that, if white is say 16x then 50% = 8x, but when you bring gamma into the picture at a correlated IRE level, at a specified gamma, say 2.4, that IRE level will vary based on black-level, given the same white luminance level.

Glen Carter
Home Theater Calibration

To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.
GlenC is offline  
post #310 of 341 Old 02-01-2007, 06:58 PM
AVS Special Member
 
HoustonHoyaFan's Avatar
 
Join Date: Jan 2003
Location: Houston, TX
Posts: 3,964
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 1 Post(s)
Liked: 20
Quote:
Originally Posted by Mark Petersen View Post

...static contrast with a DI isn't something that falls out of sequential on/off...

The defination of "static" contrast seems to be the same value as the iris on (fixed) value for a DI, based on the way the benchmark is setup.
For a Ruby:
Iris off: ~3,000:1 on/off CR
Iris on: ~ 5,000:1 on/off CR

Iris DI: ~5,000:1 "static" CR.

What are we calling the CR difference that is very clear when one A/Bs on vs. DI?
HoustonHoyaFan is online now  
post #311 of 341 Old 02-03-2007, 12:11 AM
AVS Addicted Member
 
darinp2's Avatar
 
Join Date: Sep 2003
Location: Seattle, WA
Posts: 21,174
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 6 Post(s)
Liked: 54
Quote:
Originally Posted by HoustonHoyaFan View Post

The defination of "static" contrast seems to be the same value as the iris on (fixed) value for a DI, based on the way the benchmark is setup.
For a Ruby:
Iris off: ~3,000:1 on/off CR
Iris on: ~ 5,000:1 on/off CR

Iris DI: ~5,000:1 "static" CR.

What are we calling the CR difference that is very clear when one A/Bs on vs. DI?

For a DI projector I call it the Dynamic On/Off CR. If comparing to a non-DI projector then I would call them Full On/Off CR (since the second one isn't a dynamic value), but maybe there is a better name that should be used there.

--Darin

This is the AV Science Forum. Please don't be gullible and please do remember the saying, "Fool me once, shame on you. Fool me twice, shame on me."
darinp2 is offline  
post #312 of 341 Old 02-03-2007, 12:55 AM - Thread Starter
AVS Club Gold
 
Mark Petersen's Avatar
 
Join Date: Dec 2000
Posts: 4,671
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 11
Quote:
Originally Posted by darinp2 View Post

For a DI projector I call it the Dynamic On/Off CR. If comparing to a non-DI projector then I would call them Full On/Off CR (since the second one isn't a dynamic value), but maybe there is a better name that should be used there.

--Darin

I think dynamic contrast is a good term and I think we should continue to use it. To be precise we should use it whenever luminance is boosted at the same time the aperture is decreased in a DI. In the full white and full black test patterns that we've used so far, luminance can't be boosted because it's already at 100 IRE so what's being measured is static contrast (even as the iris aperture is reduced).

But getting back to HHF's point, the "Iris 1" static contrast can be different from the static contrast in the "Iris On" mode because the iris aperture may differ. So even though the definition is the same a projector like the VW50 may yield 3 different static contrast curves on the contrast benchmark (and potentially more if one considers DI settings/tweaks).


To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.
Mark Petersen is offline  
post #313 of 341 Old 02-03-2007, 01:14 AM - Thread Starter
AVS Club Gold
 
Mark Petersen's Avatar
 
Join Date: Dec 2000
Posts: 4,671
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 11
As we discussed, I changed the thread title to finding a "supplement" to ANSI CR rather than the original "alternative" terminology to clarify that we're not attempting to replace the ANSI definition. I'll ask Alan G. to push the change so that it appears on the thread title (rather than just the first post of this thread).

I also updated the graphs in the "sticky" contrast results thread so that the title clearly states that what has been measured is "Static Contrast" per the definition we've adopted. I'm also in the process of adding a section that clarifies what is meant by static and dynamic contrast.

Since only Darin has responded so far to my request to comment on finalizing the current phase of the project, I'm assuming that everyone is okay with the current test patterns and methodology for measuring static contrast. I think Wm is adding a few more points at the very bottom of the luminance range (0.1 and 0.2% APL) so I assume that once those are done and there are no objections that the test patterns/ benchmark can be released to a wider audience.

I also assume that people want to accept the challenge of characterizing dynamic contrast in a DI as the next phase of the program so I'd like to open up the discussion for this topic.


To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.
Mark Petersen is offline  
post #314 of 341 Old 02-03-2007, 12:09 PM - Thread Starter
AVS Club Gold
 
Mark Petersen's Avatar
 
Join Date: Dec 2000
Posts: 4,671
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 11
Quote:
Originally Posted by stanger89 View Post

Might I suggest, that to maintain some level of consistency within this project, that the same "style" of test patterns be used for both spacial and lum-varying patterns? For example, keep the current patterns, and augment them with one using the same layout as the 50% pattern, but vary the %lum of the boxes on the pattern.

Then, at least at 50% APL (100% lum boxes), there would be a 100% match between the two sets of measurements.

Also (and I'm not sure how practical/accurate it would be) but assuming a set of patterns were generated with a range of %lum boxes, would it not be possible to determine the rough gamma of the projector? I suppose that would fall apart for DI projectors with their dynamic gamma.

Quote:
Originally Posted by Darinp2 View Post

I thought about suggesting the same thing, but with the other direction. Since the ANSI CR 4x4 checkerboard is a standard, start with it and vary the %stim for the brighter rectanges for one group and the size of the brighter rectangles for the other group. That would mean 8 bright rectangles in each pattern, but just of different sizes. I figured people wouldn't want to go back and start over with new spatial patterns though.

One advantage to using the 4x4 checkerboard for the test that varies luminance instead of starting with the 50% pattern here and varying luminance of the peak part is that we have ANSI CR and on/off CR for many projectors already. New measurements wouldn't have to be taken and Erik's calculator can just estimate the values. Here I am assuming that the new 50% pattern isn't using the ANSI CR 4x4 checkerboard, but I haven't looked at what that 50% pattern looks like.

These are good points. Fortunately, the 50% APL test pattern is a variation of the ANSI 4x4 but modified so that the probe postion doesn't change. A person runs the test patterns by lining up the sensor with a target image and then just runing through all of the patterns without moving the probe. This is similar to gregr's "modified ANSI" test except that the probe is centered so the 4x4 boxes need to "wrap" from left side to right side. As far as measurement deviations go, both the standard ANSI and the 50% APL yield the same measured values or very close to it.

So I take it everyone is in agreement to go ahead and use the 50% APL test pattern and then vary % lum for gauging dynamic contrast?


To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.
Mark Petersen is offline  
post #315 of 341 Old 02-11-2007, 09:55 PM - Thread Starter
AVS Club Gold
 
Mark Petersen's Avatar
 
Join Date: Dec 2000
Posts: 4,671
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 11
As we discussed, I want to update the contrast project thread with DI dynamic contrast information which in combination with the static contrast information will provide a balanced feel for DI behavior. As suggested by Bob Sorel, I'm soliciting feedback, comments and criticisms from forum DI afficionados before updating the contrast project thread.

Since the dynamic contrast test patterns aren't finished yet, I decided to do the next best thing which is to use the measured VW50 full white (iris 1, iris off) and black level (iris 1, iris off) data from the static contrast benchmark and then to use this data to extrapolate a range of white levels (%stim) as specified by Erik Garci's contrast calculator (for a 2.5 gamma projector).

Since this uses a mix of measured and calculated values it's purpose is for illustration uses only, actual results may vary. This provides only one scenario for DI behavior, but I think it serves the purpose for a general illustration quite well.

The scenario chosen assumes that the overall APL is low so that the iris aperture is at it's smallest. The % stim data points correspond to pixel intensities that might exist in a mixed, but low APL scene. The assumption is that gamma is boosted enough to overcome light loss due to iris reduction if sufficient headroom is available. Plots are done in both linear and logarithmic scale as linear scale helps to show what happens with bright detail and logarithmic scale helps to show what happens with dark detail.

This graph shows a native 3100:1 projector (VW50 data iris off), this same projector with a simple DI added that only adjusts iris aperture, the same projector with a full featured DI that performs gamma boost and lastly a native 14000:1 projector which is suggested by the on/off CR of the projector with DI.



Same data plotted with a log scale to show effect on dark details



Normalized white level deviations showing the BC effects of gamma boost



Same data plotted with a log scale to show the effect of dark detail



Spreadsheet Analysis Available Here


To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.
Mark Petersen is offline  
post #316 of 341 Old 02-12-2007, 12:38 PM - Thread Starter
AVS Club Gold
 
Mark Petersen's Avatar
 
Join Date: Dec 2000
Posts: 4,671
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 11
Looking at this data a couple of points jump out:
  • The contrast benefits of a DI equipped projector vs the same without DI are very clear - the DI projector has much improved contrast performance with dark detail in low APL scenes with the iris closed although the DI is less beneficial to any bright detail in the same low APL scene.
  • The dynamic contrast for dark detail (also with a low APL scene) matches fairly well with the profile suggested by the on/off CR.
  • The dynamic contrast for bright detail (also with a low APL scene) matches fairly well with the profile suggested by the static contrast benchmark.
  • This clearly shows the shortcoming of using just on/off CR or the static contrast benchmark to gauge DI dynamic contrast as the contrast is a reflection of both.
  • Brightness compression effects are easily determined by normalizing the white levels and comparing the relationship between different stimula against a standard 2.5 gamma (iris open) plot.
  • Brightness compression is heavily reduced if bright stimula pixels (IRE > 60) aren't present in the low APL scene.
  • Gamma can be lowered to reduce BC for bright stimula pixels but at the resultant cost that dark level contrast is reduced - this also implies that the dynamic contrast shown is a reflection of the maximum achievable.


To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.
Mark Petersen is offline  
post #317 of 341 Old 02-12-2007, 06:19 PM
AVS Special Member
 
HoustonHoyaFan's Avatar
 
Join Date: Jan 2003
Location: Houston, TX
Posts: 3,964
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 1 Post(s)
Liked: 20
Mark

Great summary!
HoustonHoyaFan is online now  
post #318 of 341 Old 02-13-2007, 12:57 PM - Thread Starter
AVS Club Gold
 
Mark Petersen's Avatar
 
Join Date: Dec 2000
Posts: 4,671
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 11
Thanks HHF! One other note, the area between the yellow curve that says, "Projector A w/ full DI (incl gamma boost)" and the dark blue curve that says, "Projector A without DI" reflect the complete spectrum of contrast ranges for all iris and gamma boosting combinations in this example (which ignores Iris 2 mode for now). I think I'll shade this region to make it more clear. It'll be interesting to see what the measured data for each iris mode with an actual test pattern will yield.


To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.
Mark Petersen is offline  
post #319 of 341 Old 02-15-2007, 07:58 PM
AVS Special Member
 
HoustonHoyaFan's Avatar
 
Join Date: Jan 2003
Location: Houston, TX
Posts: 3,964
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 1 Post(s)
Liked: 20
Quote:
Originally Posted by sethk View Post

How do you arrive at a "real world" max CR range of 100:1 from the above information?

Quote:
Originally Posted by Mark Petersen View Post

...As far as the RS1 having a max CR range of <100:1 in low APL scenes goes, I can't really see how you arrived at that...

In the other thread I stated that I believed that a 15,000:1 on/off CR projector had a maximum CR (intra image dynamic range) of < 100:1 at < 10% APL. I was asked how I came up with that number. The following is my understanding;

Luminance = Power (APL, gamma).

For a gamma of 2.2 and 10% APL, luminance = Power (0.1, 2.2) = 0.006.

At 10% APL/ 2.2 gamma scenes, a projector puts out 0.6 % the amount of light as it does at 100%. Given that the off value does not change, the max CR of our pj at 10%APL/gamma 2.2 is ~0.6% of its full on/off value = (15,000 *.006) = ~90:1

An example setup 15,000:1 projector:
Max = 15 FtL, Min = 0.0001 FtL, CR = 15,000:1
For 10%APL/2.2 gamma: Max = (15*0.006) = 0.09 FtL, Min = 0.001FtL, CR = ~90:1

What am I missing?
HoustonHoyaFan is online now  
post #320 of 341 Old 02-15-2007, 08:26 PM
Advanced Member
 
tstites's Avatar
 
Join Date: Nov 1999
Location: Dawsonville, GA USA
Posts: 804
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
What are you missing? An image with a 10% APL could still have some amount of elements that are at a luminance level of 100% (15FtL) and the ratio between those elements and "off-state" is still 15,000:1 or thereabouts...subject to the amount of light scatter through the optics that would reduce CR to some extent.

Tom Stites
"Speaking only for myself...."
tstites is offline  
post #321 of 341 Old 02-15-2007, 09:09 PM
AVS Club Gold
 
GlenC's Avatar
 
Join Date: Jun 2003
Location: Los Angeles - Whittier, CA
Posts: 2,555
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by HoustonHoyaFan View Post

An example setup 15,000:1 projector:
Max = 15 FtL, Min = 0.0001 FtL, CR = 15,000:1
For 10%APL/2.2 gamma: Max = (15*0.006) = 0.09 FtL, Min = 0.001FtL, CR = ~90:1

What am I missing?

Tom is right.

With your above example, on a 100" diagonal screen (29.67 sq-ft), 10% APL (not 10 IRE) = 1.5 ftL (100% APL = 15 ftL) therefore, with black at .001 ftL, a 15 ftL area of .098 sq-ft and the remaining area, 29.572 sq-ft at .0001ftL = average 1.5 ftL and 15/.0001 = 15000:1

Glen Carter
Home Theater Calibration

To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.
GlenC is offline  
post #322 of 341 Old 02-15-2007, 09:43 PM - Thread Starter
AVS Club Gold
 
Mark Petersen's Avatar
 
Join Date: Dec 2000
Posts: 4,671
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 11
Yes, it's important to clearly differentiate between %stim and %APL. %stim refers to pixel intensity whereas APL refers to overall average image luminance. 100% stim can easily exist in the darkest of scenes with very low APL for example, which is the scenario that triggered this discussion. The static contrast benchmark and related graph is a good indicator of how contrast changes as a function of overall image luminance (APL) while the dynamic contrast example shows how the contrast changes for various pixel intensities within a fixed APL scene. At 100 IRE the dynamic contrast example yields exactly the same value as a point on the static contrast graph at the specified APL (in the scenario chosen for the DI example earlier in the thread it was close to 0% APL).

One thing to keep in mind is that the DI dynamic contrast scenario talked about earlier in this thread happens with low APL scenes where the iris is fully closed. From the static contrast benchmark we can see that the iris doesn't even begin to close until we're below the 5% test pattern APL so 10% APL using the test pattern histogram is actually pretty bright.

HHF does have a point in that for the 10% APL the contrast is fairly low because the blacks are elevated due to the luminance. From the static contrast benchmark, we can see that for the 10% APL test pattern the RS1 only has 1000:1 max CR (for a 100% stim (ie 100 IRE) pixel). Another thing to keep in mind is that the APL from the test patterns more closely matches the "projector average luminance" from Erik Garci's Intrascene calculator. In other words the space scene has a 4% image APL, but the calculated projector APL (also from the Intrascene calculator) is only 0.12% APL after figuring in gamma encoding so from a test pattern APL stand point it probably equates to the very bottom (far left) of the static contrast graph. Matching test pattern APL to image APL was lightly touched on in this thread: http://www.avsforum.com/avs-vb/showt...&&#post9485030


To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.
Mark Petersen is offline  
post #323 of 341 Old 02-15-2007, 10:11 PM
AVS Addicted Member
 
darinp2's Avatar
 
Join Date: Sep 2003
Location: Seattle, WA
Posts: 21,174
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 6 Post(s)
Liked: 54
Quote:
Originally Posted by HoustonHoyaFan View Post

In the other thread I stated that I believed that a 15,000:1 on/off CR projector had a maximum CR (intra image dynamic range) of < 100:1 at < 10% APL. I was asked how I came up with that number. The following is my understanding;

Luminance = Power (APL, gamma).

For a gamma of 2.2 and 10% APL, luminance = Power (0.1, 2.2) = 0.006.

At 10% APL/ 2.2 gamma scenes, a projector puts out 0.6 % the amount of light as it does at 100%. Given that the off value does not change, the max CR of our pj at 10%APL/gamma 2.2 is ~0.6% of its full on/off value = (15,000 *.006) = ~90:1

An example setup 15,000:1 projector:
Max = 15 FtL, Min = 0.0001 FtL, CR = 15,000:1
For 10%APL/2.2 gamma: Max = (15*0.006) = 0.09 FtL, Min = 0.001FtL, CR = ~90:1

What am I missing?

I think you are on the right track overall if you use PPL (which I made up as Peak Picture Level). Also, I would use 2.4 for the gamma and not 2.2. I believe that there is some misunderstanding there as the display should have a higher gamma than the source to get to the little bit higher than 1.0 overall that people have been found to prefer, and if the source is 2.2 then setting the display to 2.2 instead of a little higher takes away the benefits of that. 2.4 is the number I've heard recommended by Poynton for the display. It would take the simultaneous CR down to about 60:1 peak for a 15k:1 display and an image with PPL of 10%stim.

--Darin

This is the AV Science Forum. Please don't be gullible and please do remember the saying, "Fool me once, shame on you. Fool me twice, shame on me."
darinp2 is offline  
post #324 of 341 Old 02-15-2007, 10:44 PM - Thread Starter
AVS Club Gold
 
Mark Petersen's Avatar
 
Join Date: Dec 2000
Posts: 4,671
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 11
Quote:
Originally Posted by darinp2 View Post

Mark,

I would vote that you continue measuring and graphing the test patterns with the reduced spatial picture level (less bright area, but still 100 IRE), but also include a separate graph of the 4x4 checkerboards with the brighter rectangles at different levels like Greg mentioned. That one could just say estimated from say a 2.4 gamma and Erik's calculator could be used for each projector. While I don't think that would give you 100% accuracy, I think it would be close enough for this. This way you would have one set that tends to discriminate against DI projectors, with another that doesn't really discriminate against them and shows one of their main strengths.

--Darin

I've been giving more thought into the dynamic contrast test patterns and unfortunately I don't think the ANSI patten modified with various % stim will work. For gauging DI performance we need a low APL image so that the iris is closed while the image intensity (% stim) is varied. Unfortunately the ANSI pattern has much too high of an APL so it's unlikely the iris will close until the % stim gets really low and then we won't have captured the full range of % stim variability.

Rather than a ANSI style pattern, I would suggest that we use one of Wm's low APL test patterns and then vary the % stim with that pattern. From the static benchmark, the VW50 iris closes between the 2% and 5% APL patterns (100 % full white) so we could probably use the 2% pattern to see the % stim relationship from 0-100% with the iris fully closed. On the other hand, it might also be useful to see the DI behavior with various iris apertures from fully open to fully closed so we could use the 5% APL pattern and then lower % stim which will likely give us a mix of iris apertures at various % stims. I think starting with the iris fully closed (2% APL test pattern) captures the maximum benefit of the DI though so I think it's the most relevant case to use.


To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.
Mark Petersen is offline  
post #325 of 341 Old 02-16-2007, 07:10 AM
AVS Special Member
 
HoustonHoyaFan's Avatar
 
Join Date: Jan 2003
Location: Houston, TX
Posts: 3,964
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 1 Post(s)
Liked: 20
Thanks Tom, Glen, Mark, and Darin!
HoustonHoyaFan is online now  
post #326 of 341 Old 02-16-2007, 12:40 PM - Thread Starter
AVS Club Gold
 
Mark Petersen's Avatar
 
Join Date: Dec 2000
Posts: 4,671
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 11
In light of the dynamic contrast analsysis that I posted, I went back and reread some of the recent posts in this thread and thought I would summarize a few key points. These points are of course dependent on the validity of the dynamic contrast scenario that I posted earlier and I invite people to point out flaws in the treatment of that scenario.
  • I believe it was GlenC who pointed out that the static contrast benchmark should provide the upper limit (ie maximum) contrast for both native and DI projectors alike. I think this is true. The real benefit of dynamic gamma is that it allows lower % stim pixels to have the same contrast as it would with a higher contrast native projector, but by definition the lower %stim pixels have less contrast. So the static contrast benchmark is useful in providing an upper contrast bound for both DI and native projectors.
  • I struggled in trying to relate the contrast from the static contrast benchmark with the dynamic contrast example that I posted earlier (both graph simultaneous contrast after all). I found out that one easy way to think about how they interact is to think of each point along the APL axis of the static contrast graph as providing the maximum contrast possible for the dynamic contrast scenario. The dynamic contrast scenario then shows how the contrast for various IRE pixels vary within this APL constraint.

    In other words increasing luminance from the static contrast benchmark decreases contrast by reducing blacks for each APL step. At each one of these steps we can vary the pixel stim and get the range of pixel contrasts which is what is really needed for analayzing dynamic contrast.

    This approach breaks down when large numbers of pixels are varied (such as using an ANSI style pattern) because changing the %stim in such a case can cause a signficant change in APL. I think this shows another reason why we would want to keep the checkerbox for the dynamic contrast test pattern as small as possible. Varying %stim on a large checkerbox changes the APL and it becomes impossible to relate the static and dynamic data.


To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.
Mark Petersen is offline  
post #327 of 341 Old 02-16-2007, 02:34 PM
AVS Addicted Member
 
darinp2's Avatar
 
Join Date: Sep 2003
Location: Seattle, WA
Posts: 21,174
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 6 Post(s)
Liked: 54
Quote:
Originally Posted by Mark Petersen View Post

I've been giving more thought into the dynamic contrast test patterns and unfortunately I don't think the ANSI patten modified with various % stim will work. For gauging DI performance we need a low APL image so that the iris is closed while the image intensity (% stim) is varied. Unfortunately the ANSI pattern has much too high of an APL so it's unlikely the iris will close until the % stim gets really low and then we won't have captured the full range of % stim variability.

I'm not sure I understand what you are saying here. Is it that you don't think the iris would close enough if a 4x4 checkerboard of 50%stim/0%stim was used, for instance? In theory the iris could close 4x for white there pretty easily while still maintaining the 50% stim where it should be and lowering the absolute black, but I don't know what projectors like the VW50 would actually do.

The iris could do completely different things with a 25% APL image that is 50%stim and 0%stim in a 4x4 checkerboard than what it would do with a 25% APL image that contains 100%stim pixels, since the first one doesn't require any brightness compression if they close the iris down.

--Darin

This is the AV Science Forum. Please don't be gullible and please do remember the saying, "Fool me once, shame on you. Fool me twice, shame on me."
darinp2 is offline  
post #328 of 341 Old 02-16-2007, 07:48 PM - Thread Starter
AVS Club Gold
 
Mark Petersen's Avatar
 
Join Date: Dec 2000
Posts: 4,671
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 11
Quote:
Originally Posted by darinp2 View Post

I'm not sure I understand what you are saying here. Is it that you don't think the iris would close enough if a 4x4 checkerboard of 50%stim/0%stim was used, for instance? In theory the iris could close 4x for white there pretty easily while still maintaining the 50% stim where it should be and lowering the absolute black, but I don't know what projectors like the VW50 would actually do.

Yes exactly. If you think it through, using an ANSI pattern starting with 100%/0% stim and then decrementing the % stim through 50%/0% to something lower like 1%/0% stim what you'll capture is a range of iris positions and only a small range of % stim for each iris position. The curve will look like the "iris off" curve for most of the higher % stim regions and then gradually improve to reflect the benefits of the smallest iris aperture but only for the lowest %stim. So it won't capture the full range of % stim for each iris postion or even show the full benefit of a DI. It may reflect more of a "real world" example than the theoretical best case that I graphed earlier though.

Quote:


The iris could do completely different things with a 25% APL image that is 50%stim and 0%stim in a 4x4 checkerboard than what it would do with a 25% APL image that contains 100%stim pixels, since the first one doesn't require any brightness compression if they close the iris down.

Yes this is very true. There are a lot of possible variations of iris and % stim gamma boosting combinations that can be used. I don't think it's necessary to capture everyone of them though. In fact graphing the best case (smallest aperture) and the worst case (iris open) will describe the gamut of contrast ranges. I think I mentioned that the region in the graph between the yellow curve (iris 1 - smallest aperture) vs the dark blue (iris off) describes the "in between" cases of larger iris apertures and less gamma boost. That particular plot captures the case of smallest aperture, which in turn yields the high contrast advantages of the DI. A person could capture larger iris apertures but the contrast benefit from the DI will be less.


To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.
Mark Petersen is offline  
post #329 of 341 Old 02-16-2007, 08:34 PM
AVS Addicted Member
 
darinp2's Avatar
 
Join Date: Sep 2003
Location: Seattle, WA
Posts: 21,174
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 6 Post(s)
Liked: 54
Quote:
Originally Posted by Mark Petersen View Post

Yes exactly. If you think it through, using an ANSI pattern starting with 100%/0% stim and then decrementing the % stim through 50%/0% to something lower like 1%/0% stim what you'll capture is a range of iris positions and only a small range of % stim for each iris position. The curve will look like the "iris off" curve for most of the higher % stim regions and then gradually improve to reflect the benefits of the smallest iris aperture but only for the lowest %stim. So it won't capture the full range of % stim for each iris postion or even show the full benefit of a DI. It may reflect more of a "real world" example than the theoretical best case that I graphed earlier though.

Washout that is shown by ANSI CR will most likely dominate down to where the iris can close all the way anyway with a DI that closes down to 3-4x for white. That is, by the time it gets to the point that the black floor is the dominant factor in the simultaneous CR (as opposed to the washout effect being the dominant factor), the iris will probably be closed down anyway. I would expect the simultaneous CRs to be reasonably close to what a non-DI projector with the same full range on/off CR and the same ANSI CR could do from a 100/0 checkerboard all the way down to a 1/0 checkerboard. Actually, if the ANSI CR is measured like most of us measure it, then the black floor from the lower on/off CR for the current iris position (like 3000:1 for the VW50) is already factored into that ANSI CR. But even if the on/off CR was 5000:1 at that iris position, with no other changes the ANSI CR wouldn't be much different (since the ANSI CR is going to be closer to 300:1 and so have a "black" that is at least 10 times as gray as in an all black image).

--Darin

This is the AV Science Forum. Please don't be gullible and please do remember the saying, "Fool me once, shame on you. Fool me twice, shame on me."
darinp2 is offline  
post #330 of 341 Old 02-16-2007, 09:32 PM
AVS Club Gold
 
GlenC's Avatar
 
Join Date: Jun 2003
Location: Los Angeles - Whittier, CA
Posts: 2,555
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
I still don't see/understand the need for "special" test patterns for a DI projector. The patterns that Wm has developed seem very appropriate for measuring the maximum CR in low APL. If a DI projector doesn't improve their CR with these patterns, then I don't see much advantage to the DI.

I see DI as a gimmick to help a poor performing projector (poor black levels), necessatating reliance on the manufacturer to correctly modify the gamma curves to compensate for the reduced light output.

We will not be seeing moveis released that are "optimized for DI projectors"

Glen Carter
Home Theater Calibration

To view links or images in signatures your post count must be 0 or greater. You currently have 0 posts.
GlenC is offline  
Reply Digital Hi-End Projectors - $3,000+ USD MSRP

User Tag List

Thread Tools
Show Printable Version Show Printable Version
Email this Page Email this Page


Forum Jump: 

Posting Rules  
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off