View Full Version : What gives with gradients inHD optical disks?
ottscay 01-26-07, 05:29 PM Ok, so I'm flipping through the HD/Voom channels on my dish network, and I stop on the Discovery Channel, as it is showing some gorgeous footage of animals in Africa at sunset. I watch it for a few seconds, thinking it looks awfully good for satelite HD, and then it hits me: there is a perfect gradient from dark purple down to a pale orange in the sky, and there is not a HINT of banding. So...why on earth are we still getting banding (sometimes really awful banding ala Superman Returns) in movies on HD disks, when they have more space and higher bandwidth???
Can someone please explain how bitrate-starved sat HD can provide a better gradient then many movies do on either HD DVD or Blu-ray?
Well, Discovery HD actually has quite a bit more bandwidth than Superman Returns.
That said, they also use MPEG2 compression which, so we are told, is much less efficient than VC-1.
So I guess your guess is as good as mine. My guess is that banding is difficult to control in VC-1. It is there on Last Samurai (the beginning) and it is still there on Superman Returns.
I guess we had better get used to it.
Mr. Hanky 01-26-07, 06:47 PM Can it really be said that banding is absent for digital satellite programs, altogether? I would not doubt you could find it in either d-sat or hd-disc, given enough samples.
Possibly, the cbr style encoding in broadcast affords better handling of gradients in simple scenes (assuming it isn't getting bit starved from other picture details), while the vbr style encoding of offline vc-1 may naturally seek greater compression where possible, which would increase the possibility of gradients getting mangled?
The overlying theory I would apply to this scenario is that increases in compression also increases the possibility of losing the subtleties of bit-depth. The complete range of the bit-depth is still available, but the smoothness going from one digital step to the next becomes not so smooth, as the information for the "minor steps" is thrown out to achieve the data reduction.
You can certainly see banding issues in many broadcast situations.
One of the most common is on a scene change that is not at an I frame. This happens more in broadcast and offline since they use real time encoders.
randosel 01-26-07, 07:13 PM I see it often with fade ins from white.
ottscay 01-26-07, 07:14 PM Whoa, I wasn't implying there is no banding in broadcast HD. I was just struck at how a relatively low bitrate mpeg2 Hd broadcast outdid many of the VC-1 encodings I've seen (in this regard). I am not saying tha mpeg2 is better, or even solves the problem, I just want to know what is causing this obnoxious artifact in our movies, and what can be done to fix it? There's been a lot of "maybe it's in the master, or mpeg2 can't do better" kind of talk, but here is a lower bandwidth mpeg2 image with a gorgeous gradient, so I'm not taking excuses anymore for why many of my disks (including mpeg2 encoded disks) have such obnoxious banding.
One of the insider codec guys needs to see what can be done to fix this problem, asap.
rexdigital 01-26-07, 07:32 PM Whoa, I wasn't implying there is no banding in broadcast HD. I was just struck at how a relatively low bitrate mpeg2 Hd broadcast outdid many of the VC-1 encodings I've seen (in this regard). I am not saying tha mpeg2 is better, or even solves the problem, I just want to know what is causing this obnoxious artifact in our movies, and what can be done to fix it? There's been a lot of "maybe it's in the master, or mpeg2 can't do better" kind of talk, but here is a lower bandwidth mpeg2 image with a gorgeous gradient, so I'm not taking excuses anymore for why many of my disks (including mpeg2 encoded disks) have such obnoxious banding.
One of the insider codec guys needs to see what can be done to fix this problem, asap.
I have an old discreet smoke edit system that is 8-bit, but its really high precision.
So it can capture 10-bit digibeta, edit and output back and retain quality with no banding.
If there were more rounding errors in the 8-bit data you would see banding.
So what we need in VC-1 is 8-bit high precision and your banding will be gone.
I just want to know what is causing this obnoxious artifact in our movies, and what can be done to fix it?
I'm not taking excuses anymore for why many of my disks (including mpeg2 encoded disks) have such obnoxious banding.
I would be interested to know of any MPEG2 discs that have banding - if you can post time codes I would be interested to take a look.
With regard to your first question - I doubt that the VC-1 types will ever answer it. Getting them to even admit there is an issue is a battle in itself. Been there - tried that - they think their work is transparent to the master... :(
ChrisWiggles 01-26-07, 08:25 PM Banding is also easily introduced at the system level, so it can be very difficult to ascertain whether the banding is in the source unless you are certain your system is banding free. Just FYI.
darkedgex 01-26-07, 08:26 PM (sometimes really awful banding ala Superman Returns)The banding in Superman Returns is in the source AFAIK (I noticed the banding in the theater). Not sure about other titles though.
The banding in Superman Returns is in the source AFAIK (I noticed the banding in the theater). Not sure about other titles though.
That's interesting. I don't recall seeing it in the theatre, nor have I seen it on the BD which I've viewed multiple times. I use non-pixel tech (CRT projection) though.
ottscay 01-27-07, 04:16 AM I did not see SR in theaters, which is why I've refrained from arguing the point too strongly on this disk vs. the encode. Otoh, I have seen the banding on the Blu-ray version, so there's no reason to make this a format issue (the encodes are identical). To me, banding, whether in gradients or during black or white wipes, is the most egregious problem with many disks right now. YMMV.
I did not see SR in theaters, which is why I've refrained from arguing the point too strongly on this disk vs. the encode. Otoh, I have seen the banding on the Blu-ray version, so there's no reason to make this a format issue (the encodes are identical). To me, banding, whether in gradients or during black or white wipes, is the most egregious problem with many disks right now. YMMV.
Digital displays exhibit banding artifacts. Google "banding artifacts lcd", or "banding artifacts plasma", depending on which display type you have.
Is the satellite HD 60Hz? The smoother gradients could be a result of temporal noise dithering.
|
|