[Bloodflowerz]

Originally Posted by Mrb1972

Philips Statement regarding HDNM problems/complaints

Thanks for this. I expected this too. However, as I mentioned, some things have been blown way out of proportion. People should be happy they have a consistent, uniform panel, not many dead pixels, and no buzzing/whining from the panel's capacitors! All TV's have some issues to a degree. With this knowledge, just enjoy the show.

[John Pickles]

Originally Posted by Bloodflowerz

Thanks for this. I expected this too. However, as I mentioned, some things have been blown way out of proportion. People should be happy they have a consistent, uniform panel, not many dead pixels, and no buzzing/whining from the panel's capacitors! All TV's have some issues to a degree. With this knowledge, just enjoy the show.

Here here.

[Ninjafoot]

Bloodflowerz it was unfortunate having too faultly tvs hopefully you will eventually get third time lucky.

[eikka]

Originally Posted by Mrb1972

Philips Statement regarding HDNM problems/complaints

Hmmh. Sounds like the damage control department is kicking in. I think people need to extend the tests of these clips, plus the "urban acrobat", which is horrible, to other manufacturers (Sony, Samsung, LG) 2012 models to verify Philips claims. It may be that the source video is corrupt as they claim. But the way things have progressed with this and other problems with the new Philips TVs should teach us to take this sort of explanation with a pile of salt.

The "Urban acrobat" problem on youtube:
http://www.youtube.com/watch?v=GhkPi57v9Zo

[John Pickles]

Originally Posted by Mrb1972

Backlight controls how dim or bright the picture is, video controls the white level, around 90 is correct

So in effect, is Backlight Contrast the same as what other manufacturers refer to as "backlight control" ( sorry if that's an obvious question) and people are advised to "lower the backlight" ?

[Mrb1972]

Originally Posted by John Pickles

So in effect, is Backlight Contrast the same as what other manufacturers refer to as "backlight control" ( sorry if that's an obvious question) and people are advised to "lower the backlight" ?

Contrast = white level
Brightness = black level
Backlight contrast is overall brightness, around 65 is fine but i leave light sensor on so the automatically backlight will adjust depending on room lighting

[eikka]

2. Also the scene with the jail bars in "White Collar" at 6m07, show exactly the same behaviour on all tested models: there are artefacts visible in the jail bars while they pass by.

As explained above, the noted effects are normal and expected behaviour. They are a result of the incoming material, which has judder inside (24/25/30p film), and defined by the used picture processing technologies and choices made.The fact that some of you experience this as a fault, cannot be acknowledged, and is really a matter of TV settings, source material, and personal perception.There is nothing wrong with our television equipment and, compared to competition, Philips Natural Motion has certainly clear advantages in many scenes.

Our final conclusion is that our 2012 models fully behave as specified, and are at least on par with (and in some cases better than) competition on Picture Quality.

The bolded parts stick to my eyes. Majorusa has already proven that the older Philips TV 9603D can keep up with the jail scene perfectly with PNM implementation:
Philips 9603D HDNM test - YouTube

This means that the artifacts are being introduced by the newer models processing. How can Philips claim that noted effects are normal and expected behaviour when their older video processing HW can cope with it? It sounds pure spinning to me.

[Mrb1972]

Originally Posted by eikka

The bolded parts stick to my eyes. Majorusa has already proven that the older Philips TV 9603D can keep up with the jail scene perfectly with PNM implementation:
Philips 9603D HDNM test - YouTube

This means that the artifacts are being introduced by the newer models processing. How can Philips claim that noted effects are normal and expected behaviour when their older video processing HW can cope with it? It sounds pure spinning to me.

Dont forget you can have HDNM set to min and that clip is fine, med and max are the problems - The only way to know for sure if its spin or not is to take the clip to JL or something and test it on various models/makes

[eikka]

Originally Posted by Mrb1972

Dont forget you can have HDNM set to min and that clip is fine, med and max are the problems - The only way to know for sure if its spin or not is to take the clip to JL or something and test it on various models/makes

Well, this makes the Philips explanation even more awkward, doesn't it? They say "everything is fine" and "there's absolutely nothing wrong with the frame interpolation techniques we are using" and yet there are modes of operation where HDNM does not produce artifacts and modes that produces them.

I suppose at the minimum setting the interpolation chip is in less aggressive mode and does not compute as many in-between frames for the most complex scenes than the med/high settings, reverting back to lower framerates when scenes become too complex for it. Now, this trick could be also be used for med/high modes if they are able to detect the stereotypical structure that creates these artifacts.

It is hard for me (with background in video coding) to believe in source coding problems in this particular case, because the artifacts are clearly showing up systematically with similar edge orientations in different sources. High frequency (=high contrast) static/moving vertical edges in scenes containing horizontally oriented optical flow is where these artifacts always seem appear with HDNM mode active (med/high modes). They just won't admit that the challenge to interpolate edge motion in these situations is too much for their algorithm and are practically saying people should disable the function entirely if they are annoyed with the problem, ahem, sorry, non-issue. And this does not align with the way they advertise HDNM/PNM with the TVs: they say it is working perfectly with any video source.

I think some tech sites should take this issue under scrutiny to see a truthful evaluation whether philips motion system is actually no worse than any other frame interpolation in the market with this type of motion characteristics.

Any tech journalists reading this thread?

[Mrb1972]

Originally Posted by eikka

Well, this makes the Philips explanation even more awkward, doesn't it? They say "everything is fine" and "there's absolutely nothing wrong with the frame interpolation techniques we are using" and yet there are modes of operation where HDNM does not produce artifacts and modes that produces them.

I suppose at the minimum setting the interpolation chip is in less aggressive mode and does not compute as many in-between frames for the most complex scenes than the med/high settings, reverting back to lower framerates when scenes become too complex for it. Now, this trick could be also be used for med/high modes if they are able to detect the stereotypical structure that creates these artifacts.

It is hard for me (with background in video coding) to believe in source coding problems in this particular case, because the artifacts are clearly showing up systematically with similar edge orientations in different sources. High frequency (=high contrast) static/moving vertical edges in scenes containing horizontally oriented optical flow is where these artifacts always seem appear with HDNM mode active (med/high modes). They just won't admit that the challenge to interpolate edge motion in these situations is too much for their algorithm and are practically saying people should disable the function entirely if they are annoyed with the problem, ahem, sorry, non-issue. And this does not align with the way they advertise HDNM/PNM with the TVs: they say it is working perfectly with any video source.

I think some tech sites should take this issue under scrutiny to see a truthful evaluation whether philips motion system is actually no worse than any other frame interpolation in the market with this type of motion characteristics.

Any tech journalists reading this thread?

Also Philips said they tested on the medium setting on various brands and the results were the same. You make it sound like Its just a philips issue.

[eikka]

Originally Posted by Mrb1972

Also Philips said they tested on the medium setting on various brands and the results were the same. You make it sound like Its just a philips issue.

And yet it doesn't show on 9603D from Philips with HDNM? This is how majorusa commented on the issue in Philips forums:

I truly love when people are making allegations without bothering to verify them. For your viewing pleasure I will post a video of my 42PFL9603D - a four years old TV with the same scene that I posted to show the issues 6007 series have. Look closely at the bars on the right hand side of the picture. Then look at the other video. Both have HDNM on maximum.

Philips 9603D HDNM test - YouTube

The picture on the 9603D is not perfect, It has some blur but this is not what we asked from Philips. We asked to fix the HDNM/PNM of the 6xx7 series in order to handle motion in the way my 4 years old TV was handling. Is this unreasonable?

The other video is here:
PHILIPS TV 6007 series faulty picture, HDNM issue 2 - YouTube

47pfl6877t - Page 16

So the source video works with age old Philips HDNM but not with the new HDNM/PNM. With this example alone it is hard to swallow that the interpolation chip works as intended with the new tv sets and there's no other TV on the market that could fare better than Philips (that's what they are saying now).

Of course the possibility exists that the newest versions of the chipset are using more complex algorithms than 9603D and thus having more complications with complex motion scenes but then we come into this question: has the progress in frame interpolation regressed from what it used to be, quality-wise?

I am not owning this TV but was planning to. Now I am not sure anymore. I am trying to get a grip on this issue. Is HDNM min completely OK for a non-juddery artifact-free motion experience for SD/HD broadcast and HD movies in these types of vertical line motion scenes or not? Many, many people seem to notice this problem to appear in various occasions, some say they have HDNM off but some other settings on etc. Some complain that artifacts appear when there's graphical overlays on the video, such as menus and subtitles. Then Philips comes up and says that their tech works with no issues. Whatever they say evidence for these problems exists on the web and many seem to be unhappy about it. What to believe? Well, I believe evidence when I see it.

[aise]

on my 8007 i have tested all the clips PNM MINIMUM and all the crap is visible, this talk about on minimum not causing any problems is BS.

[Mrb1972]

Originally Posted by aise

on my 8007 i have tested all the clips PNM MINIMUM and all the crap is visible, this talk about on minimum not causing any problems is BS.

No one said min was problem free, but it does help in some situations, example the jail scene is ok on min

[eikka]

Originally Posted by Mrb1972

No one said min was problem free, but it does help in some situations, example the jail scene is ok on min

So what do you normally do with the tv? Do you keep HDNM on minimum and bite teeth when you recognize a problem, keep the entire feature disabled or what? Honestly, are you content with the situation as it stands? (I may have asked you this before but what's your status now?)

[Mrb1972]

Originally Posted by eikka

So what do you normally do with the tv? Do you keep HDNM on minimum and bite teeth when you recognize a problem, keep the entire feature disabled or what? Honestly, are you content with the situation as it stands? (I may have asked you this before but what's your status now?)

Well all I can say is that this is my 4th LED TV in around 18 month, LG LW550, LG LW980, Samsung D8000 (all returned for various issues) , and the philips 6007 is the only one I actually keep the motion system on, especially for sport.

You will notice most professional review sites always turn off these motion system because of the issues they can bring.

To answer your question I am content but only because I dont think changing brand or model would make much difference and the actual picture quality I have now is very good compared to other TV's I have had.

[aise]

Originally Posted by Mrb1972

No one said min was problem free, but it does help in some situations, example the jail scene is ok on min

it is not ok to me.

[eikka]

Originally Posted by Mrb1972

You will notice most professional review sites always turn off these motion system because of the issues they can bring.

Yeah, I find this odd sometimes. These TVs are being advertised with their image enhancement features and consumers actually try to use these features, yet many reviewers disregard them or mention them briefly. It would be fresh to see a review section where these image enhancement features are being compared against each other when viewing SD/HD, sports, soap opera, documentaries, movies... Only way to compare them would certainly be subjective, but qualitative consideration of strengths and weaknesses of these proprietary features would be enlightening read for many people.

[Mrb1972]

Originally Posted by aise

it is not ok to me.

What have you got? I forget

[zlatko]

I used to have a 9603 and now I own a 9606. For my eyes HDNM have been improved in these 3 years - I see less halos and artefacts, and even when I see them they are a way more subtle then what was on 9603. I always keep HDNM at minimum setting.
I am sure there isn't a perfect motion interpolation system, but for me benefits of having one outweighs its shortcomings.

PS. I have searched thoroughly, but can anyone point me where I can download the jailbars clip to test on my 9606?

[eikka]

Originally Posted by zlatko

I used to have a 9603 and now I own a 9606. For my eyes HDNM have been improved in these 3 years - I see less halos and artefacts, and even when I see them they are a way more subtle then what was on 9603. I always keep HDNM at minimum setting.
I am sure there isn't a perfect motion interpolation system, but for me benefits of having one outweighs its shortcomings.

PS. I have searched thoroughly, but can anyone point me where I can download the jailbars clip to test on my 9606?

I think they are using some re-encoded (or at least re-encapsulated) episode, based on the name of the file ...
White.Collar.S01E01.720p.HDTV.x264.mkv

[eikka]

Hey all,

Out of interest to this HDNM issue, I decided to do some experimentation myself. Since I don't have any LCD display that would have frame interpolation function, I chose to harness my quad core PC to playback frame interpolated video with my Dell 2709W monitor running at 60Hz.

To achieve motion compensated frame interpolation on PC video playback, one can download SmoothVideo Project software
SmoothVideo Project (SVP) - motion estimated frame interpolation with any video player and frame doubling in real-time

This software integrates itself into Media Player Classic Home Cinema video player and lets you configure various settings to achieve frame interpolation e.g. from 24/25 fps video source to display frame rate (60Hz) or any other frame rate (2x or 3x frames). SVP uses AviSynth and ffdshow codecs for the actual frame processing.

The estimation of complex object motion is very demanding for a CPU, and even my Quad core system cannot do it in Full HD, not even if I force my 6950 AMD GPU to support in computation. Granted, this might improve as SVP developers manage to optimize their software, but right now with v3.1.2, it is very demanding task for CPU and GPU.

However, I managed to configure the software to produce smooth playback of 1080p24 content with my system by forcing the rendering frame size to 800p. This downscales the video and therefore reduces the amount of motion computation so that it is manageable with different quality settings.

Then I downloaded the reference video that manuj has reported as being difficult for Philips HDNM. The video is called hd_other_philips_urban_acrobat.m2ts and can be found from the web by searching with the filename. Video is 1080p24 h264 encoded clip about fast motion urban acrobatics. Rather challenging setting for any frame interpolation technique.

Then the tests. For performance reasons I configured the SVP to use 800px framesize and cranked all other quality settings to the maximum (or near maximum). There are nine different quality settings to choose from. I played back the above video using MPC-HC repeatedly, stopped playback and progressed frame by frame to examine the interpolated frame quality.

I took screenshots from the most challenging shots, in which the acrobat made different jumps, spins and flips with camera following the actor. These shots produced lots of artifacts which you can view in the image gallery:
Philips Urban Acrobat Frame Interpolation w SVP - Imgur

The gallery shows how challenging it is to interpolate (i.e. create) visual reality using computational methods. The complex object motion in the high contrast edges creates artifacts in the intermediate frames that people tend to perceive as abnormalities in the playback flow. However, these artifacts are different than the ones that are being produced by Philips' HDNM chip. In Philips system limbs disappear in the intermediate frames, whereas SVP produces mostly malformations of the objects. To compare, you can see Philips system's performance in this youtube clip:
Philips TV 6007 6877 8007 6907 faulty picture HDNM another example / Artifacts / HD Natural Motion - YouTube

SVP also integrates to your other video players that use FFDSHOW, for example I use it to decode DVB-C SD broadcasting. Therefore I also tested the frame interpolation with SD broadcasting. It works much better, but not completely artifact free. Likely explanation is that SD content as being MPEG2 576i50 has less complicated motion flow matrix, which is easier to estimate computationally. Right now I am watching My Darling Clementine (1946) which is really smooth but also loses some of the "magic" of the movies... Whether this is a good thing or not, is a matter of taste of course.

The issue of higher framerate playback is very hot topic at the moment, since Peter Jackson decided to film and release two versions of "The Hobbit", one with classic 24FPS, another with 48FPS (High Frame Rate). There are lots of polarizing opinions on how adding frames affects to the experience. Some people hate it whereas others see it adding immersion...

Some conclusions from this little experiment:

* creating realistic in-between frames by estimating them is a challenging task for complex object motion.
* frame interpolation works best with scenes that contain sufficiently calm motion.
* Interpolating frames with CPU is very, very demanding. You need a really high-end PC system to interpolate full 1080p frames.
* Philips HDNM artifacts are different type than the ones with SVP algorithms. Philips object estimation makes some parts of complex motion objects disappear whereas SVP creates malformations.
* For this video, I would disable SVP and enjoy the movie with the orginal 24FPS format instead of suffering from anomalities with the perceived playback flow.
* SD frames seem easier to interpolate without artifacts than HD images, but the artifacts were still there when both background and foreground object were under complex motion. SVP algorithm is not perfect, but it's free.
* Peter Jackson is trying to push the framerates beyond classical 24FPS and is using Hobbit HFR version as a guinea pig. Perhaps in the near future we start having video sources that are filmed with 48FPS.

[eikka]

More tests with the above video.

This time I used a commercial video player Splash PRO that uses proprietary Motion2 technique to interpolate frames. They have 30 day trial version of the player, so you can try it yourselves. It is a straightforward install:
Splash PRO - The Next Generation Player

Overall the results with the urban acrobat test video were much better than with SVP interpolation. There were less frames with noticeable artifacts and the type of arfifact was not a malformation of the object. I have collected some of the artifacts here:
Philips Urban Acrobat Frame Interpolation w Splash PRO - Imgur

This player seems to be able to interpolate 1080p frames without overburdening my PC setup, so it proves that the frame interpolation is feasible to implement if the software is optimized. The results are not perfect here either, but overall the interpolation technique doesn't produce disappearing or malformed body parts. Most of the artifacts are ghosting remnants from previous frames, which do not break the actual object motion flow when viewed in real time. Also the biker scene at the end of the video works fine unlike with SVP.

Conclusion: Splash Pro PC video player software does indeed produce interpolation that is always tolerable and often pleasant experience in this challenging clip. There are some shots where the technique breaks down and produces artifacts, but these frames do not break the flow of motion during viewing.

Based on what manuj posted on Youtube, I get the feeling that Splash Pro Motion2 does produce more tolerable artifacts in the backflip scene than Philips HDNM technique. This is due to fact that Motion2 does not make limbs disappear entirely in the estimated/interpolated frames.

The assumption is that manuj used HDNM med/high in his test, whereas setting HDNM to min might improve situation for this clip. Maybe someone could make a video or take pictures of their TV running the backflip scene with HDNM set to min?

[Mrb1972]

Originally Posted by eikka

More tests with the above video.

This time I used a commercial video player Splash PRO that uses proprietary Motion2 technique to interpolate frames. They have 30 day trial version of the player, so you can try it yourselves. It is a straightforward install:
Splash PRO - The Next Generation Player

Overall the results with the urban acrobat test video were much better than with SVP interpolation. There were less frames with noticeable artifacts and the type of arfifact was not a malformation of the object. I have collected some of the artifacts here:
Philips Urban Acrobat Frame Interpolation w Splash PRO - Imgur

This player seems to be able to interpolate 1080p frames without overburdening my PC setup, so it proves that the frame interpolation is feasible to implement if the software is optimized. The results are not perfect here either, but overall the interpolation technique doesn't produce disappearing or malformed body parts. Most of the artifacts are ghosting remnants from previous frames, which do not break the actual object motion flow when viewed in real time. Also the biker scene at the end of the video works fine unlike with SVP.

Conclusion: Splash Pro PC video player software does indeed produce interpolation that is always tolerable and often pleasant experience in this challenging clip. There are some shots where the technique breaks down and produces artifacts, but these frames do not break the flow of motion during viewing.

Based on what manuj posted on Youtube, I get the feeling that Splash Pro Motion2 does produce more tolerable artifacts in the backflip scene than Philips HDNM technique. This is due to fact that Motion2 does not make limbs disappear entirely in the estimated/interpolated frames.

The assumption is that manuj used HDNM med/high in his test, whereas setting HDNM to min might improve situation for this clip. Maybe someone could make a video or take pictures of their TV running the backflip scene with HDNM set to min?

That clip is just as bad on min setting

[eikka]

Originally Posted by Mrb1972

That clip is just as bad on min setting

I have now seen "Urban Acrobat" on Sony 2012 HX855 (Motionflow), Philips PFL6097 (Perfect Natural Motion) and LG LW980 (Trumotion). All of them produced artifacts in different situations. The legs and hands of the actor were blocky during the fastest spins and jumps. However, Philips PNM differed from LGs and Sony's as it had the smoothest background motion (=the urban scenery: walls of the houses) of them all. The notorious backflip jump was the low point of Philips PNM as parts of the body disappeared for some of the interpolated frames. This effect is visible in manuj youtube video. Overall, every TV produced different artifacts in different situations but Philips motion experience was the smoothest.

This sample video is very, if not the most challenging naturally created action movie (no special effects) and it was not clear if I preferred such amount of motion and action in 24p or with some interpolated frames with artifacts..

[Bloodflowerz]

Originally Posted by eikka

I have now seen "Urban Acrobat" on Sony 2012 HX855 (Motionflow), Philips PFL6097 (Perfect Natural Motion) and LG LW980 (Trumotion). All of them produced artifacts in different situations. The legs and hands of the actor were blocky during the fastest spins and jumps. However, Philips PNM differed from LGs and Sony's as it had the smoothest background motion (=the urban scenery: walls of the houses) of them all. The notorious backflip jump was the low point of Philips PNM as parts of the body disappeared for some of the interpolated frames. This effect is visible in manuj youtube video. Overall, every TV produced different artifacts in different situations.

This sample video is very, if not the most challenging naturally created action movie (no special effects) and it was not clear if I preferred such amount of motion and action in 24p or with some interpolated frames with artifacts..

Hello Eikka

Thank you very much for taking the time to compare the motion processing of both the Sony and LG, in relation to the Philips.

I've been wanting to test the HX855 (HX853 in UK) including it's motionflow but have not had the chance to do so. It's comforting to know that the Sony too produces artefacts to some extent with the same, complex scene.

As you have found, all brands/motion systems produce artefacts during a complex scene. This is the very nature of the technology being utilised. Luckily, I was so impressed with 8007's picture quality, sharpness, motion etc with HDNM Off, that I think it's not that much of an issue; for me anyway.

Once I return to Germany, I'll buy the 8007 for the third (and hopefully final) time, knowing that HDNM is comparable to Motionflow and other high end TVs.

Finally, something positive to read.

[eikka]

Originally Posted by Bloodflowerz

Luckily, I was so impressed with 8007's picture quality, sharpness, motion etc with HDNM Off, that I think it's not that much of an issue; for me anyway.

When I did side-by-side comparison with Sony, I noticed that PFL6097 produced sharper image (but not over sharp) with the Urban Acrobat clip, so I consider that as a plus for Philips, at least for the PFL6xx7 series.

I talked with the salesman at the store and he said that in practical situations these artifacts show their face when watching HD Sports, e.g. football, if at all. I recall that mrb said that with Philips HDNM can be kept on (at min setting) when watching sports. So that's a definite plus for the technology.

I believe there is a demand for a review that would show and describe subjective image differences with various image enhancement features for different brands. Hope that some tech journalist would take a hint for a thorough review.

Anyways, HTH!

[eikka]

Originally Posted by eikka

When I did side-by-side comparison with Sony, I noticed that PFL6097 produced sharper image (but not over sharp) with the Urban Acrobat clip, so I consider that as a plus for Philips, at least for the PFL6xx7 series.

I talked with the salesman at the store and he said that in practical situations these artifacts show their face when watching HD Sports, e.g. football, if at all. I recall that mrb said that with Philips HDNM can be kept on (at min setting) when watching sports. So that's a definite plus for the technology.

I believe there is a demand for a review that would show and describe subjective image differences with various image enhancement features for different brands. Hope that some tech journalist would take a hint for a thorough review.

Anyways, HTH!

In the same breath I have to say that there are some examples that might be alarming in case of sports:
https://www.youtube.com/watch?v=kKVeKjXESvw
https://www.youtube.com/watch?v=YrbE...o&feature=plcp


Read more here:
Your experiences with FW v 150.48.13 for 6, 7, 8 and 9xx7 series (2012 models) - Page 15

I think it relates to specific type of motion problem where complex horizontal motion messes up with vertical edges among the motion field. You can see that from the football clip where the artifacting becomes stronger when camera's parabolic trajectory is at peak position (ie. producing mostly horizontal background movement)

Every manufacturer has their weak spots and this will be the beauty spot on Philips' otherwise fine system.

[Bloodflowerz]

Originally Posted by eikka

Every manufacturer has their weak spots and this will be the beauty spot on Philips' otherwise fine system.

Very nicely said.

[Mrb1972]

Originally Posted by eikka

When I did side-by-side comparison with Sony, I noticed that PFL6097 produced sharper image (but not over sharp) with the Urban Acrobat clip, so I consider that as a plus for Philips, at least for the PFL6xx7 series.

I talked with the salesman at the store and he said that in practical situations these artifacts show their face when watching HD Sports, e.g. football, if at all. I recall that mrb said that with Philips HDNM can be kept on (at min setting) when watching sports. So that's a definite plus for the technology.

I believe there is a demand for a review that would show and describe subjective image differences with various image enhancement features for different brands. Hope that some tech journalist would take a hint for a thorough review.

Anyways, HTH!

I actually prefer it on min for football and motor racing, I don't really see this issue with the logo break up as the sky sports logo doesn't seem too bad, but I can see why it would be annoying

[Bloodflowerz]

Originally Posted by Mrb1972

I actually prefer it on min for football and motor racing, I don't really see this issue with the logo break up as the sky sports logo doesn't seem too bad, but I can see why it would be annoying

Hello Mrb1972

Do you have Sky Sports HD? Can you confirm during football that the logo does not break up and artefact as shown in the video a few posts ago?

Cheers.