Noise, is it just me ????

[apfraats]

Using a good TV-set and DVD-player for playback using RGB interface (europe), I see a lot of backups having a quite noisy playback when compared to my digital cable TV connection and LIVE or recent recorded programms.

The big problem is NOISE here, I mean REAL NOISE, not blocks...

I have two questions...

1) Can NOISE be added y the CCE encoder I use or the compress proces as a whole ??? (I only thought about blocks and artifacts to be related to MPEG-2 compression....)

2) Are more peple noticing NOISE on even recent released DVD's. For example Ocean's Twelve, which really had a lot of noise ??????

3) Noise filters I don't use. That what I've tried really messed up other aspects of picture quality.


Anybody any reactions on this ?


I would expect that present DVD form recent movies, contains NO NOISE at all, but on my backup's I often notice it. The times I tried the Original, it was equally noisy.......

I quality of DVD-recording still often this bad in 2005 ????

[jptheripper]

cce doesnt introduce mosquito noise.
It is possible the RBG analog connection is the problem, but i doubt it (have you checked other sources?)
recently dvds - the ocean movies in particulr, have awful encoding. You could smooth them with an undot type of filter, the result is softer but much less noisy.

[apfraats]

cce doesnt introduce mosquito noise.
It is possible the RBG analog connection is the problem, but i doubt it (have you checked other sources?)
recently dvds - the ocean movies in particulr, have awful encoding. You could smooth them with an undot type of filter, the result is softer but much less noisy.


That just what I wanted to know. I even tried the originals.

There is nothing wrong with my equipment, and an average DVD of a not to old movie gives good to excellent results.

But yes, I was working on ocena's 11 and 12 and the noise was really annoying....

But I now know it's the quality of the source material, as I expected.

I didn't just expect that in 2005 this kind of low-quality stuff is put on the market as 'DVD' with the expectation of related quality.

So nothing to worry about, just bas source material. It's a shame people have to pay up to $49 for these kind of poor DVD's.......

And filtering I already tried a bit. It has not been to my satisfaction, as it's also affecting other aspects of the picture and in facts make things worse.

I'll have to live with the noice.... Just set the SHARPNESS setting of my TV to a lower number greatly improves overal picture quality experience.
If there just was an AVI filter doing exactly the same, but I have'n't found one. UNDOT() is no good. UNDOT().DEEN even worse...

[jptheripper]

i actually like undot().deen() on noisy sources, but to each there own. when filtering i find you need way more passes, like 6ish

as for the quality of stuff, unfortunately dvd is a format, not a quality.

you can put like 12 hours of bad vcd material from an old hi8 cam in 720x480 at 1.5mbit and format to dvd, and its a dvd. afterall, dvd is digital versatile disk, not a measure of quality.

[SAPSTAR]

CCE is well known for introducing some mosquito noise....You have to tweak the VBR Bias to make it disappear...Some people are not disturbed by it some others cant stand it...matter of taste...I suggest you to give a try to AQE (www.autoqmatenc.com), you shouldn't have it...It's slower than CCE but for low bitrates and without tweaking I think it's better from a visual quality point of view...

[apfraats]

Well looks like I have to do a POLL on it.

There's one saying that CCE isn't producing musquito noise, and one saying it's well known for that, so I'm lost here.

Let others put their experiences on this subject. And take CCE SP 2.66 in mind. I have read several test reports on encoders, and everytime CCE is at the top....

So I have to watch closely to the original and the compressed version with CCE to determine where the noise is comming from. Also it's player dependent. But the player that produces the most noise is the same player that produces the sharpest and most vivid pictures on good sources. I think the other player has less bandwith and gives smoother pictures by default.

I don't expect a digital software encoder to add extra noise to a source, but what the hell I know about CCE internals.....

I do expect any hardware encoder to add noise and are sure DA en AD converters do. But CCE just calculates a lot with MPEG2 vectoring-based compressions algorithms and I expect it not to add musquito noise.

But here the expierences of others are of use.

So if anyone knows about CCE and noise addition , I would greatly appreciate some more feedback.

And indeed DVD is a 'just' a standard and not a garantuee for quality.

But I expect the pro's (movie companies e.d.) to try to get the most out if it, and expect quality, which seems to be far from reality regarding significant material that has noise that wasn't there when I watched the movie in the theater, on a much bigger screen as I have at home.

Despite the source quality I really like to know others about CCE SP 2.66 adding musquito noise (or snowy noise) to the processed materials.

It seems unlikely but if someone is saying it's a well known topic, I get a little uncertain about it.

[apfraats]

CCE is well known for introducing some mosquito noise....You have to tweak the VBR Bias to make it disappear..


I 'tweaked' it to 0, to get FULL VBR, which I prefer strongly....

Maybe that's making things only worse, so what do you mean with tweaking ?

[Boulder]

Mosquito noise == Gibbs artifacts == a thing every MPEG2 encoder will introduce to the video in areas of high contrast.

EDIT: and it's not VBR bias which affects it, it's the quantizer characteristics setting. The more to the left the slider is, the less mosquito noise should appear. However, the flat areas will begin to suffer if it's too much to the left.

[apfraats]

and it's not VBR bias which affects it, it's the quantizer characteristics setting. The more to the left the slider is, the less mosquito noise should appear. However, the flat areas will begin to suffer if it's too much to the left.


In CCE SP 2.66 SP there are no sliders.. At least not using DVD-RB-PRO.

I only can set two parameters VBR_BIAS (which is 0) and precision (which is at its highest).

So I wonder what you exacly mean, I suppose precision. I set it to it's highest value, and cinemacraft says indeed flat areas can suffer. But that I didn't notice so I leave it there for now.

If every encoder produces 'musguito noise' more or less, I have seen a lot of movies on DVD that hasn't these == ... ARITIFACTS.

So if they are always there, they should be not so annoying as the noise I mean. I've seen perfect MPEG-2 reproduction without noticeable artifacts or noise, as long as you don't sit in front of the screen at 10 centimeters distance....., but then again you even see the red, blue, green pixels of the fosfor layer.....


The noise I mean is a kind of 'snowy noise' , like in old movies. Or which you get when TV reception is not at it's best... Many analogue cable connection have it, as too many amplifiers are used in the cable network, each contributing to noise in the analogue signal.

That's were digital TV comes in, which is very good here, due too high bitrates used... And even that's MPEG-2 but gives a perfect picture especially with LIVE transmissions and quality pre-recorded programms.

There is nothing annoying about that quality, although the noise I'm talking about is.

I suppose the noise that's implicit to MPEG-2 encoding is not the problem.

Sure it's not HDTV but reasonable high quality reproduction in my perception is possible.

So still concerned about 'my' noise. And concerned about the addition of noise of whatever kind after re-encoding with CCE SP 2.66.

[Boulder]

Gibbs artifacts appear less as you spend more bits on the image. That is why most commercial DVDs don't show the phenomenon much.

I see it all the time on one of our cable companies channel. The subtitles are burnt on the video and cause high-contrast edges and have a huge amount of mosquito noise around them. This is due to the fact that they don't use much bandwidth for the transmissions, from what I've heard, it's something like 3000-4500kbps.

You could also mistake noise for film grain.

[manono]

Hi-

You've got that good Philips HDTV, right? Does it have a DVI or HDMI port? The reason I ask is that it's well known that the Digital->Analog->Digital conversions that will take place with your setup will introduce noise to the output on your TV set. If you can get a digital connection from player to display, that'll prevent a lot of noise from showing up.

Also, have you experimented with using different matrices? CCE has been accused of introducing noise. I'm not entirely sure if it was something to do with CCE 2.50 (you're using 2.67) or the fact that the Standard Matrix is default. You may get better results with less noise, by experimenting with using different matrices for your encoding. And as Boulder mentioned, the lower the bitrate, as compared to the original, the more noise will be introduced.

...and precision (which is at its highest).

I don't know how much that setting has to do with producing noise, but I, for one, would never use that setting

[SAPSTAR]

I 'tweaked' it to 0, to get FULL VBR, which I prefer strongly....

Maybe that's making things only worse, so what do you mean with tweaking ?
VBR_BIAS (which is 0) and precision (which is at its highest). are most certainly an issue, the default values chosen by jdobbs are rather good for many movies!!!
Check http://forum.doom9.org/showthread.php?p=549211#post549211
...You have to adjust VBR Bias + Q prec to make it disappear....in VBR mode, you're always VBR, VBR Bias = 0 means you're going to allow bigger variations in the bitrates, but you may have some spikes for example....



@Boulder : You're right, I mixed Qprec and VBR Bias but the artifacts you're talking about is not what apfraats is talking about, the mosquito noise is something really well known...The "noise" encountered with cables distribution, is more like blocking due to a poor compression/bandwidth....

[Boulder]

One thing to consider is that when you have film grain (and some movies have lots!) and a lowish bitrate, you end up getting a swimming noise pattern throughout the video due to excessive quantization during encoding. It doesn't look like actual blocking but close to it and it's very noticable especially on any flat coloured area.

[apfraats]

It's kind of difficult I see to get the most out of it.

No I do not have a HDTV, the time is not ready for it here. First let's have 'the blue ray disk' for example.

I've just a 34 inch widescreen pixel+ TV, which delivers excellent picture quality by doing real signal processing. I use the RGB interfaces, which are analogue but have great result for the normal 'TV-resolution'. Somewhere in the TV the signals have to be converted to analogue signals always...

No progressive scan e.d., as we in Europe use other techniques to improve picture reproduction, such as 100 hz vertical scanning , and real video processing circuitry done bij Philips patended pixel+ chips, which are not yet availbale to the 'open market', so y'll only find them in the more expensive Philips TV's.

This improves the PAL vertical resolution by 33% and does some other things as improving sharpness and image resolution and handles motion control.

Of course what aint in the source can't be magically reproduced, but the chips just try to proces the signal as intelligent as possible and make the most of it, of course at the expense of some drawbacks. It's always one or the other. But the picture reproduced has in many tests winned the top ratings concerning conventional PAL TV system picture reproduction.

Also MPEG-2 artifacts normally not seen, are espcially annoying on these kind of TV's as they are enhanched and get more clearly visible. So far it's the best TV I've had to check MPEG-2 image reproduction, and gives the best picture possible not being HDTV.

It's way behond the older mechanism just to improve picture sharpness and resolution, which were simply based on analogue tricks and did have serious drawbacks....

But the artificacts in high contrast area's such as mentioned is well known indeed especially in case of white bright text on picture, but still that's not the noise I'm concerned about. Using low-bitrates always introduces this significantly and it's clearly seen on my TV.

I'm concerned about the 'snowy noise' that's all over the picture.

Like the effect when you enlarge a photo and get to see the grainy picture caused by the way chemicals are used. (so not JPEG !, just good old chemical optical photography)

My cable connection is a digital one, and not shows up even the slightest "GIBBS artifacts" as they use high bitrates which generates a very sharp picture and even with liitle white letters on a grey backgound, you get no gibbs atrifacts at all.

Some channels have bad source, even seeing 'blocky effects' in the picture...

But they haven't the noise, and I use the same RGB connection. So it''s no topic related to the TV or the interface.

Even a good DVD gives perfect images.

What I now want to know is the best setting for VBR_BIAS and DC_PRECION concerning introduced noise by CCE, from whatever kind.

But I also want full VBR as this makes theoratically the best distribution of the space (and bitrates) you have left for video. I don't want to wate just one bit on a picture that doens't really need it, but I want all bit's I can get for pictures that demands it. So full VBR looks like the best option. I even always check te result with bitrate viewer to make sure spikes are not going over the top (10.0 or so Mbps minus bitrate needed for audio).

I even use low-bitrate matrices for higher bitrate stuff, as they SEEM to improve bitrate distribution.

I never have 'blocky' effects, but I also are not trying to use bitrates that have an average below 4000. Then I split the DVD simply into two.

But noise being not GIBBS ATRIFACTS, is still a problem...

Actually trying one with UNDOT() and one without it comparing on TV, not my computer screen, as the TV shows up much more quality as my computer screen, strangely enough.... (Using powerd-DVD and hardware acceleration).

Looks like I have to take a reference source and doing a lot of experimenting with VBR_BIAS, PERECISION, MATRICES and FILTERS...

Oh boy, that's gonna take me 5 years of my live........

[jptheripper]

a couple suggestions

1. try the defaults on a disk you thought was noisy. see if they are better, my bet is they are
2. a major thing of note. Every item i have seen on calibrating tv's for best picture, etc, from the professionals, is to turn every image enhancement off. Especially sharpness. I set my hdtv sharpness to 0 and it has doen wonders. Also the brightness and cnotrast on your tv, combined with the huge number of image enhancing items the philips seems to have, i can only believe is doing more hard than good considering that these dvdrs are now slightly bitrate starved.

as a test, just once try a dvd you considered bad at the default settings. Then play it on your tv with contrast below 40%, brightness below 50%, and all image enhancements off, and sharpness set to 0. The colors wont be what you are used to, but just look for noise. My money says its gone. Then play with your settings until you like the image but the noise hasnt returned.

[Boulder]

Please upload a small sample of the noisy video. Otherwise it'll be impossible to determine any possible causes.

Altering VBR bias probably won't help you much, as won't DC precision. My guess is that the film grain gets quantized during encoding and that causes the noisy look. I think you should use a high-bitrate matrix with higher bitrates because they do not quantize as much as low-bitrate matrices.

A good TV will reveal the bad stuff sometimes too easily and it might be settings-related as well as jptheripper said.

[theclaus]

Yeah I notice noise on my HD set when watching the Sex and the City DVDs. This is on the originals.

[apfraats]

@jptheripper

I know what you mean, and you are right in almost every case, but pixel+ is a different case howver. You even can't turn it off !!! The image will aways be processed, only a demo mode is available that show's you the image in the left half og the screen without pixel+ processing and in the right half with pixel+ processing. And sure, without pixel+ the noise is greatly gone, but also details are lost and it's much more without details then the pixel+ half.

You can manually change sharpness setting, which is not to be compared the the commonly known 'fake' sharpness. This is a kind of different thing also concerning real image processing.

And again with shrapness at 0 , noise is 50% or more gone, of course.

Bit the point is, when the signal is good, the picture is really even better then comparing it to the screenpart not having pixel+ signal processing.

The difference between a good and less good signals are much more noticed with pixel+ then without pixel+ and having a good source it's even merely a better picture.

Picture+ emcompasses real image processing, not some 'electronic' joke, to get the well known 'fake' enhancements, as seen on many TV's today.

So Philips hasn't made it switchable. If that's just arrogance or the reasoning 'why turning something off that really enhanches your picture quality' is for people themself to decide. Just like turning off colors on a color-tv.... When I bought this model, I compared it with the general BAD cable-signal they have in TV shops, and a high quality DVD. I personally liked the Philips much more then others in picture and color reproduction, so I bought it.

I know that you can discuss this topic, but being not able to switch it off, I have to live with it, and i'm still verry happy with it too, as this is an overall improvement.

I can switch the TV to 100HZ digital scan, without pixel+, but then also I can count the horizontal scanlines from up to down.... That even isn't an option at all.

Contrast and Brightness I'm always easy with. Standard settings are way too high and infuence the geometric stability of the picture and even the contrast is too high so you'll get out of focus images and pain in youre eyes too..

Brightness is set to 44 or so, but absolute numbers are strongly TV-model dependent so they don't say much. Contrast is about 42.

Brightness is set so that black is real black and just no grey is showing up in the usual environmental lightning.

Contrast is set that even the hardest white is still perfectly in focus, so the tube is not 'overrayed' and gets out of focus.

The problem is still noise and I try to post an example.

Also the earlier spoken about Gibbs artifacts (musquito noise in contrasted areas) show up even on some original material.

But as far as I understand, noise will take a lot of bitrate to reproduce. So reducing noise will leave more 'headroom' for preventing too much Gibbs artifacts. I always notice some Gibbs artifacts on almost every DVD in some scenes if I watch very critical, and they even are there with my digital cable signal sometimes, depending on the programm.

So let try the more complex noise filters and see if that's makes a difference. If i'm right they are applied to the source BEFORE encoding takes place. If the final source for the encoder has less noise, even the Gibbs artifacts should go down.

I still use 2.66 and don't know if 2.70 is any better, though it should be in fading static scenes... (as listed under improvements by Cinemacraft).

Maybe there are filter for Gibbs artifacts already in the source, I have to do a search on that one.

First let's try AVS filtering and playing with the PRECISON setting that has something to do with it too.

All is fine, untill I see thinhs remembering me of DVDSHRINK, for examply just a blocky effect. This is really terrible to watch and must definitely not be there. That's why I set PRECISION to it's higjhest value.

In this thread sofar i've learned a lot, and see so many variable factors, that indeed I can spend 5 years finding out the most optimal combination of all variables.

Well, let's make a hobby of it... :D

[Myth!]

just a though about cables....i presume the 'RGB' connection is 21-wire SCART. even though you use the same scart cable for DVD and cableTV...one might be kicking out RGB colour and the other composite colour ie using different wires inside the cable

[manono]

Hi-

Somewhere in the TV the signals have to be converted to analogue signals always...

Maybe your TV. Maybe all PAL people's TV. Not my TV. There was a big and obvious drop in the noise when I switched from component to DVI. Have you calibrated your TV? If not professionally, then with the AVIA or DVE discs? Or even with the (what's it called?) THX Optimizer found on some DVDs, such as the Star Wars DVDs? That can help a lot, since most TV defauts suck, especially for brightness, contrast, and sharpness. Lowering the brightness will help with blocking in dark areas. Lowering the sharp setting should help with the noise, and almost all TVs have the default sharp setting way too high. I just noticed that jptheripper mentioned the same thing. I agree, and also have my sharp setting at 0, in both the player and the TV.

I'm concerned about the 'snowy noise' that's all over the picture

Do you see the same thing when playing the DVD on the computer, like with PowerDVD, or if testing on a friend's TV set? If not, then it's something to do with your TV or the connections, and not with DVD-RB or CCE.

But I also want full VBR as this makes theoratically the best distribution of the space (and bitrates) you have left for video.

I agree, although evidently that's a minority opinion around here. This has been discussed quite a bit already, though.

[apfraats]

just a though about cables....i presume the 'RGB' connection is 21-wire SCART. even though you use the same scart cable for DVD and cableTV...one might be kicking out RGB colour and the other composite colour ie using different wires inside the cable


Good thinking although, but:

Nope, I sure don not want to buy a $2000 TV and use $1 scart cables....

I'm not like people who used their cars to ride 100 miles because something is $10 dollars cheaper there, and they don't think about the costs of the car ride.....


Also my TV gives a message when switching from SCART to SCART and also reports the kind signal it's recieving for example RGB,CVBS, or even YUV (s-vhs). The cables all report RGB mode on the display.

I even shwithed cables and used the same scart plug for both just to be sure. Nothing wrong there.

I have flat scard cables that are fully wired and were wuite expensive. I hate round scart cables, especially the thin ones, as they are of improper quality.

[apfraats]

Hi-

Somewhere in the TV the signals have to be converted to analogue signals always...

Maybe your TV. Maybe all PAL people's TV. Not my TV. There was a big and obvious drop in the noise when I switched from component to DVI. Have you calibrated your TV? If not professionally, then with the AVIA or DVE discs? Or even with the (what's it called?) THX Optimizer found on some DVDs, such as the Star Wars DVDs? That can help a lot, since most TV defauts suck, especially for brightness, contrast, and sharpness. Lowering the brightness will help with blocking in dark areas. Lowering the sharp setting should help with the noise, and almost all TVs have the default sharp setting way too high. I just noticed that jptheripper mentioned the same thing. I agree, and also have my sharp setting at 0, in both the player and the TV.

I'm concerned about the 'snowy noise' that's all over the picture

Do you see the same thing when playing the DVD on the computer, like with PowerDVD, or if testing on a friend's TV set? If not, then it's something to do with your TV or the connections, and not with DVD-RB or CCE.

But I also want full VBR as this makes theoratically the best distribution of the space (and bitrates) you have left for video.

I agree, although evidently that's a minority opinion around here. This has been discussed quite a bit already, though.



My TV is just a normal NOT HDTV and even in your TV (if it has a tube) the signal goes to analogue, because a tube is an analogue element.

I agree most default settings are far to optimistic set by factory default.

Brightness and contrast are way down the factory setting, these were to high.

No I even notice soem noice with power-dvd, but the size of my computer screen isn't comparible to the size of my TV-screen.

And if you talk in pixels/inch, I think my TV screen has a much higher resolution as the PC monitor (just using 1024*768), which shoulf be enough for basic PAL resolution.

Then again it's is able to produce a high quality picture (no HTDV !!) as far as possible.

Noise is on DVD's, even on some originals and worse on DVD-RB-PRO processed DVD's with CCE.

I tried the UNDOT() and it helps on the TV concerning this noise.

So the fact is the TV is capable with it's current setting to produce a high quality image, inclusing the build in image-processor.

Even my TV has it's own STILL option and you can fetch a single picture (two frames interlaced) and look at it. (Digital processed however, it's showed in 100HZ pixel+ mode). The 100HZ option and increased vertical scanning resolution we use instead of progressive scan.

Then you can see the difference between DVD and the digital cable decoder.

Even the DVD-player I use is of infuence.

My Sony gives more clearer, more vivid but also more noise pictures.

The targa DVD-HHD DRH-5000 gives a smoother less noisier picture.

But I suppose as the Sony DVD-player was far more expensive, it's just having higher bandwith and noisy picture become more visible.

Just have a good original DVD with high bitrates can give excellent results.

Maybe it's all about bitrates. I don't know.

Just keep on experimenting with all variable factors and even try another encoder as CCE, while I still am convinced CCE is the best.

[Boulder]

Like I already said, post a sample of the noisy video.

Film grain is perfectly normal. Even noise is normal and all this even on DVDs. Sometimes they appear as if a VHS tape was used as the source. Most DVDs look like crap on my TFT monitor because it reveals errors and noise much more than my old TV.

UnDot() removes some of the film grain and that's why it appears to help.

[xebolon]

@apfraats
for the beginning try this tabel, I found it in the german doom9-forum.

AVG-Bitrate VBR_Bias Qual.prec Pass DC-Prec Matrix
>6000 25 16 2 10 Standard
5000-6000 25 16 3 10 Standard
3800-5000 25 16 3 9 Standard
3400-3800 20 16 3 9 Standard
3000-3400 15 20 3 8 Standard
2000-3000 10 24 3 8 Very Low
<2000 10 24 3 8 Jawohr

try the attachment, I think you can read it better.

And buy a quality connection cabel.

cu Xebolon

[TheSeeker]

I just think alot of the time the initial quality of the original dvd just sucks, thats what your seeing most of the time I would say. Try a dvd like Fifth Element (the new ultimate edition) and oh man that is some really really great freakin quality for dvd standard resolutions. Then you look at the Friends tv series dvd's and they absolutely suck major ass.

EDIT: I have an hdtv and an hdtv dvr box from Time warner, and most of the time even broadcast quality hd will blow many of my dvd's out of the water, so now I notice grain on dvd's that I previously didnt notice. I also see that the fine detail on many dvd's is severely lacking. I can't wait until an hd dvd standard is chosen and we can get cheap players and movies....

[apfraats]

I think you get the point here.

I now have 2 categories of problems:

-1- Main snowy Noise already on original DVD's.

-2- GIBBS ARTIFICATS, for example seeying a bright white logo on a brown backrground tou see ARTIFTOFACT around the edges of the LOGO.

Problem 1 is there and I have to try filtering of dome kind.


Problem 2 is more interesting as I just want if people have experiences with the GIBBS ATRIFIFACTS and how to prevent them. Of course on HIGH-BITRATE GOOD SOURCE DVD's this is not noticed however.

But some people say the GIBSS are introdiuced by CCE.....

Now the question is HOW TO GET GIBBS ATRIFACTS to a minimum.

As I have no HDTV, but it has probably much higher bandwith and resolution then is native for PAL. That's why a video signal processor adds scan-lines and does tricky things to try to use this over-resolution to still produce a good qaulity overal picture and that's generally nicely done by Philips.

We do not have any HDTV sources at the moment, but they it is being reintroduced in Europe again. But the content providers have to cooperate and they just now are willing to... (not all of course).

We have had an earlier introduction of HDTV which was a total waste of effort and didn't break through to the consumer-market at that time.

So what about the GIBBS artifact and the CCE precision setting ?

Anyone ?

What indroces the most GIBBS artifacts 64 the highest setting or 0, the lowest setting ? I already tried and noticed not much difference...

[Boulder]

The lower the value is, the less Gibbs you should see. A low value is supposed to direct more bits to complex parts of a frame, hence less Gibbs. I've never done any tests so I don't know if there is any real difference. Surely it'll also depend on the avg bitrate, the difference should be more apparent with lower bitrates.

Gibbs artifacts are introduced by every encoder, it is the nature of MPEG2 encoding which causes this. MPEG1 and MPEG4 are just the same.

For film grain/general noise, I would use LRemoveDust with mode 5, 2 or 17 (in order of strength) and limit 1 to avoid oversmoothing.