Sony VPL-vw1000 - Page 39

It would help if we knew exactly what the the two RC sliders did. I would think that they vary terms in the scaling and associated noise filtering algorithms used by Sony.

The issue is how to observe exactly what effect these controls have on picture quality without knowing the forumulas and what these sliders vary with respect to those formulas. I suspect the formulas are very proprietary and will not be disclosed by Sony.

Now I suspect that there is no absolute correlation between how one should be set with respect to the other. Sony could have supplied only one slider and taken away user choice with respect to the other. It didn't but an inkling as to the relationship between the two is the defaults Sony uses for normal and high brightness operating modes (20/20 for normal and 50/50 for bright). An implied linearity in the setting of the two and the same sloping of the two controls with the same crossing of the axis. Yea Yea. Its probably not a straight line.

But what do these controls actually do? How should one adjust them for the input signal one is feeding the projector. One would think noise associated with the input signal should be dealt with the noise etc controls in the advanced operating menu and that the resolution and noise controls under Reality Creation are not dealing with the resolution of the input signal and not how noisy it might be. Of course I could be wrong but then again I could be right. I hope that covers the possibilities as to my thesis.

If I am correct, we would need several test patterns to properly adjust the controls, patterns for each input resolution such as 720p, 1080i, 1080p 24 and 1080p 60. Assuming I am right, then what would be an appropriate pattern or patterns?

Looking at content, it is hard to discern significant changes but selecting certain let's call them torture patterns, the controls have significant visable effects and some quite negative if they are what I will assume are not correctly set or rathe set incorrectly. A engineer/lawyer quibbling with language because I sure can tell what is a wrong setting but I can't determine what settings are optimally correct. There is another possibility that there is no correct setting and that either of the two controls depend on user taste but once again the proactive control would be resolution with no guidance as to how to set it and the reactive control, noise filtering, to make watcheable what havoc operation of the proactive control might cause. A noise control normally removes noise appearing as high frequencies (I feel this statement will probably attract varying degrees of criticism but I am trying to generalize and think this through). High frequencies are detail and throwing out some noise will undoubtedly throw out some detail. Theory versus what you can see. If you can't see the high frequency detail but you see the noise then probably it is OK to remove the noise and the high frequency detail as a consequence. Also some may not be as sensitive to noise as others. Its almost like the varying tolerances possessed by some AVS members to read posts containing significants amounts of drival such as this one on the slim outside chance that something of significance may be inadvertently stumbled on.

I still do not know what is an appropriate test pattern and how to use it to set the RC controls. A traditional resolution pattern such as Wolfgang was projecting with the Isco thingy in front of his lens? The one Randall posted?. I don't think so because they don't tell you much it.

I think on off pixel patterns might be more appropriate including ones that presented diagonal lines rather than horizontal or vertical. I don't have such a pettern but I can tell you horizontal and vertical pixel patterns are DRASTICALLY affected by operation of the controls, however the extreme havoc the let's call it improper RC settings have on these patterns doesn't always translate to havoc on normal content. In my experience operation of the sliders causes pink coloration in the patterns and in the extreme pink banding, however by operation of the noise filtering control together with the resolution control most the pink coloration can be eliminated. Also extreme operation of the noise control can cause severe blotching on the pattern. One click too much, but the blotching is not visable on content. At present, I use the pattern to set the cotrols to smooth the pattern slightly and set the noise control to give me good gray coloration of the pattern. Even here there is considerable range available and I have no means of determing whether noise at 10 or 20 is the better setting, but I can make all hell breaks lose at certain settings. I am being illustrative with my numbers, not reporting them exactly.

The cowards way out. Just leave them at the Sony defaults. Why the 50/50 for bright modes instead of 20/20 for less bright modes? Simply, the higher the RC sliders are set, the brighter the picture appears.

One thing I did notice is that the picture got measurably brighter when turning RC on with resolution test patterns displayed. With a 100 percent white screen there was no measurable difference.

I read the manual and seems like there is a confusion as to what these things are. The projector has two distinct processing features relative to the topic at hand:

1. Noise Reduction. This has a setting of low, med, high and off. This is the classic noise reduction which will get rid of high-frequency noise but with it, will also soften the image. You should have this set to off unless you are watching something specific where noise is objectionable and it is better to have a softer image than one that is noisy.

2. Reality creation. The purpose of this feature is to boost the contrast of the image as to make you think the image has higher resolution than it really does. It is essentially the sharpness control but with an addition called noise reduction.

The first control essentially does what a sharpness control does by boosting high-frequency/high-texture areas of your image. What this is doing is increasing contrast of the areas in high resolution areas. This is why the image looks brighter because the highlight areas are made even brighter (and lowlight areas even darker).

If the above contrast enhancement is applied to color channel it has a side effect of creating color shift in areas where one of the color components is much weaker than the other. Typically this shows up at the sharp edges. So this explains the pink color shift. The right solution is to NOT sharpen the color component at all. The eye is not very sensitive to color resolution anyway. If you just sharpen the luminance (black and white) then you get the increased subjective effect with no chance of false color. This would have been easy to do but seems like Sony opted to not do that and you have no control over it.

The other drawback to the above is that it then exaggerates everything that is high frequency. This means that both film grain and compression artifacts become more visible in addition to creation of halos/ringing. This is why DTV is looking bad btw. Sony's solution to the former is to give you a second control which prefilters the high-frequencies, possibly with some smarts to only remove what is random, and then sharpening what is left. As a result you get a complex mix of benefits and artifacts from both features.

The solution is to leave RC OFF. In photography we do use these features but that is because we have control of the one image we are applying it to. Not here. You will be getting an artificial image with it and artifacts.

These are consumer type features much like Vivid mode of TVs. They will appear to be an improvement but just like using tone controls to heavily boost the highs in your stereo, it is not something you don't want to do.

So when you set it to off, what do you observe?

Amir- Thanks for checking into that. I think most of us were under the impression RC actually upscaled the image to 4k resolution. Sounds like that's not the case. Wondering if that is under another setting or the Sony does it with all content by default?

Answering the last part first, yes, you have to have the upscalar always on or you won't get a picture. The imager has 4K pixels so it must always get that many pixels to fill the screen. So you can't turn that feature off.

Sony does call the scalar Reality Creation. This could mean one of two things:

1. It is just a market phrase. Sony has done this for a long time calling even mundane things other people do as "reality creation." I think they did this with de-interlacing and 3:2 pull down.

2. The two sliders overlay the features I mentioned as the new pixels are being interpolator. An interpolator can easily be designed to sharpen the image as it is doing its thing. After all, it is just a passband filter and its response can be modified to not be flat as you get to higher frequencies.

Either way, my recommendation is to leave it off.

Hi. Tony. Your question is simple, the Sony can simply upscale 1080p (1920 x 1080) by doubling it (3840 x 2160). The selection is made by selecting NORMAL aspect in the machine. The additional pixels on the sides will be lighted black but they are not scaled from anything. If you select an aspect of 2.35 Zoom, you will trigger the full scale scaling on of the Sony and it will scale 1920 x 1080 to 4096 x 2160.

Now with respect to Amirs post, I am answering this under my own name and not in my job capacity.

There is very little info provided by Sony as to what RC is and what the controls do. However, based on that info, I think it is fair to say that the Resolution control increases contrast and detail while the noise filtering control reduces texture or roughness which I think arises due to the scaling of HD images. RC is there to deal with the artifacts of the scaling involved and not to the artifacts introduced by the original signals. This scaling is different for each different input resolution and whether the machine is scaling to 3840 x 2160 or 4096 x 2160. Without looking at the PQ under each set of conditions, I do not think one can make a judgment as to what setting is best. Off, on with the sliders set to minimum, or on with the sliders set somewhere.

Scaling inherently causes a loss of sharpness. All scaling does and scaling algorithms incude provisions to resharpen the image after scaling. There is nothing inherently wrong or undesirable in Sony providing for user adjustments here. Obviously, the resolution control also enhances contrast. Perhaps the scaling reduces some contrast. I do not know but I don't think anyone can conclude that the best picture would be to have RC off. Noted observers have concluded that On with the sliders set to minimum is the best, Sony seems to think On with 20/20 or 50/50 is desirable, and you conclude that off is best with no opportunity to use your eyes to reach that conclusion.

I do understand being a purist and not using tradional sharpening, edge modulation etc. But here because of the scaling and the magnitude of same and the difficulty particulars with a scale to 4096 x 2160, it may be very necessary to use RC to maximize the net results.

Thanks Amir very informative post

And yet, with Super 8, it was turning Reality Creation on (with sliders at zero) that got rid of the excessive film grain. I'll try to arrange to try that test again just to confirm, but I'm quite sure about my recollection of the sequence of events. Maybe somebody else who has that disc but doesn't have to go annoy a local dealer to try it out could give it a whirl?

The other issue with turning RC off, of course, is that it seems that you may lose the benefits of the 4K panel. Somebody please correct me if I'm wrong, but I think with RC off, the 1000 just doubles the pixels in both axes to fill the 4K panel. Don't you have to turn RC on in order to get the improved details that were shown on sites like Cine4Home?

Thanks Mark. Makes perfect sense.

Amir - Thank you for your point of view as well.

It will do that if the minimum setting is not zero.


Oh, that would be pretty significant bit of information to have!


I showed that there was no improved resolution in their demos and the processing is as I just explained. See http://www.avsforum.com/avs-vb/showt...3#post21366543

Boy, it would be good to just have one thread on this topic . I can't keep track of where I have post what....

Agreed. Thanks for dong the homework for us.

So my BIG question is what is (if anything at all) the OVERSCAN feature doing in the 1000ES?

It does what it does elsewhere. It enlarges the image as to cut off the edges. We did this in the old days because there was usually junk at the extreme ends of the image. You definitely want to set that to off.

If doing so shows some artifacts (such as dancing dashes at the top which is the VITC timecode) then sadly you will need to turn it on. This should never happen with Blu-ray but may with some set-top boxes.

Thanks again Amir.

I just wasn't sure if it had anything to do with 4K scaling. Man oh man there is a lot to discover still with our 1000s.

I still like the RC. I guess it must be my less than perfect vision.

I don't think anyone has said "improved resolution". Most have mentioned improved detail. One way images of the same resolution can show increased detail is if one of the images has higher mid frequency MTF (local contrast). Another way an image can be made to have greater perceived detail is with the introduction of advanced anti aliasing after scaling. RC includes those 2 processing algorithms along with many others as part of its scaling to 4k.

I got a chance to see a VW1000 for a couple of hours last night at a colleague's home against his SIM2 Lumis Host. I will echo Art's sentiment, sitting close to the screen at < 1.5 screen width the increase in detail of the VW1000 is very clear.

At the end of the day it is processing. RC is introducing elements into the upscaled image that was not in the original. Is it effectivE processing? Well my friend is replacing his Lumis Host with the VW1000!

One in the same. There is no improved detail either. Detail = Resolution. These transformations increase contrast. Contrast makes detail more visible but doesn't increase it. You say it correctly below as increased *perceived* detail.


Your source is not suffering from MTF issues. If the projector is not able to resolve 1080p with high MTF, then boosting the local contrast will subjectively improve things. But you better be ready for its resulting artifacts which is ringing, halos, color shift, more visible compression artifacts, etc. This is the reason sliders are given -- it is recognized that they damage things subjectively on some images.


If these RC sliders were on with the Sony and not with Lumis, then that explains this observation. You are not comparing apples to apples. Turn up the sharpness on Lumis and then compare. Or turn off RC on Sony.


Let's hope he is doing it for the right reason per above .

Zero seemed to be the minimum setting on those sliders. It was all the way to the left.

I may be totally wrong on how the upscaling works with or without Reality Creation. Most of my understanding of this projector comes from reading this forum and the various reviews. I've only had the one 90-minute visit to my local dealer, which was great, but I didn't have the opportunity to really delve into settings, and I wouldn't have wanted to do that with their projector anyway, without knowing for sure that we could put it back to how they had it!


I'm not 100% sure that I agree with the totality of your analysis of those images, or perhaps I'm not fully understanding you. I agree that there's some sharpening and some boosting of contrast going on, but there's also some smoothing of pixel aliasing.

If we look at the first side-by-side image that you posted, and look at the woman in the foreground who is facing away, in the left image you can see stair-stepping along her left side, as her hair slopes down to her shoulder. If we look at the same area of the image on the right, you can see that the edge of her hair is smooth. The same is true of the arm of the man carrying a briefcase.

It's a little tough to be totally sure on these, since we're also looking at captures from a YouTube video. But, if you go to the actual review on their site, you can see for example that there's a test pattern with diagonal lines. With reality creation off, there's obvious stair-stepping. With it on, the edges are much smoother.

So, it seems like Reality Creation is doing several things. Some desirable, and some perhaps less so. I happened to notice in re-looking at Cine4Home's images, for example, that in one of the images that shows a before/after closeup of the plaza scene's domed roofs, the shot with Reality Creation on does show more film grain in the blue sky.

I do tend to be a purist about a lot of things, but I think it's worth keeping in mind that a 1080p Blu-ray transfer of a film is an approximation of what was in the real film frame, so we're already dealing with adulterated data. If processing in the projector can help compensate for some of the degradation of the original film image and result in an image that looks more similar to the original source, then I'm in favor of it.


You ain't kidding! It's really beyond time to close a few threads and/or consolidate threads.

You simply can't sit as close to the image displayed by the SIM2 1080 x1920 projector as you can to the Sony. Pixel structure becomes obtrusive enough to be distracting and overwhelms the appearance of detail in the image.

As I said it really allows one to take advantage of great BDs to a greater extent, IMO.

Art

We should just change this thread into the Owner's thread or just start a new one dedicated to settings...

Exactly how we feel here. Last night we had a few couples over watching a movie and they had been here before. Right away I heard comments like, "Doesn't matter who sits up front now! Look at the picture. It's awesome all over the room." That is the 1000 in a nutshell. I always sit in my "seat" in the back but now I may be moving up front for a change.

Say Art are you using an acoustically transparent screen? If so how does it interact with the smaller pixels from the Sony?
Thanks

And I should mention I have two here and plan to do some measurements on the second as well this weekend. Game mode seems like the real torch mode...
And yes, I am keeping a 1000 for myself!

Would be nice to understand why that is.

Mike. If one merely doubles the 1080p source in each direction one gets 3840 x 2160 pixels. I could be wrong, but this will always be the case if the Sony Aspect control is set to normal. I believe RC will act on this panel resolution just like it will act on a full lit panel of 4096 x 2160 when the aspect is set to 2.35 zoom. This machine gets confusing because it is a true 4K panel of 4096 x 2160 (17/9) that nevertheless can be treated as a 3840 x 2160 (16/9) panel with the extra side pixels lit black but not containing any info from the source frame. Scaling is scaling but the scale necessary to get 4096 from 1980 is much more difficult and number crunching than the 2.0 scale needed to get 3840 from 1920.

Obviously, one has the choice of not using the processing that RC would otherwise apply to the scaling.

My thoughts here are logic based rather than I know based. But once again, I question any statement saying whether and how much it should be used without observing and testing.

Art uses a perforated Studeotec 130. I don't believe that Art made any comment about Moire inteference and I assume he would have had that occured.

Mine is even worse, trying to read the tea leaves from what you all are saying and extrapolating .


The purpose of that side-by-side comparison I did was to show clearly that no detail or resolution was created at all. I think we all agree that is the case. And on top of that, the processing has clearly created new artifacts Here is the image again:




I took those as compression artifacts which would vary from one video to another. Based on what we know relative to mandatory noise reduction which seems to also turn on with this mode, it would make sense that some of that was in the source and filtered.


You are correct on your observation there. Question is, was that because of pixel doubling or is the noise reduction softening the edges? There is fair bit of convergence error with red and green bleeding that makes it hard to see the pixel structure clearly to determine if there are indeed duplicated pixels. This is what I wanted to see earlier when I asked for test patterns. We should be able to determine interpolation being on/off pretty easily. Well, easily if one of your sets doesn't bleed as much as theirs does.


It does. It is good to look at them again:

Without RC on:


With RC on:


I don't know about you but I find the second image harsh and artificial. Don't see how the first image is the result of pixel doubling yet it looks smoother than when RC is on! The note under the second actually says that a "purist" may not like that effect. I guess this is proof perfect that I am a purist!

Same is true of the next two images:





And we have the pink color shift to boot. Yes at first blush you might say the second looks better but let it sink in for a bit and you should realize that it is harsher.


Unfortunately it is not doing that. It is taking the "adulterated" image and making it more adulterated . What is above the 1080p response is gone forever.


I guess the stranger bit is that we are in $3K forum discussing a $25K projector!

You can't see them ! You have to use a piece of white paper to see that they are there at a few feet from the screen. There is no moire though with my screen with the HT 5000 or the Sony.. In my room the AT visibility goes away at about eight feet back. The pixels on a 1080 projector, on the other hand, continue to be visible to about 12' at times.


Off to play with my new I Rule.


Art

Mark, sorry, I wasn't being clear. My point was that with simple upscaling to 3840x2160, you can just take a single pixel and turn it into four smaller pixels, and you'll still see the same jagged edges that you would with a 1920x1080 panel, so you're not getting any benefit from having a 4K panel in the machine.

My understanding was that with Reality Creation off, and in the standard aspect ratio, that's exactly what you got.

But, with RC on, the projector (among other things, apparently) applies anti-aliasing to the scaling, and you get the smoother edges that I commented on in my reply to Amir's posting above. So, you're getting a benefit from that 4K panel, even if you leave the "extra" pixels on the side dark.

Gotcha. I think we should just gvebup on all projectors and just try to view things in real life. This may be impracticable but the choice is to watch on a capture, transmission, and display chain designed to fool one into believing that one is there. Some want to limit the fooling, me I like to fool aroundand the test is does iot foole more? Sure its got to be color accurate, the geometry has to be accurate, whites can't be clipped or blacks crushed other things can be used to make me like it better. I want to be fooled.