Some recent posts in the Edge thread and other comments got me thinking, and I'm posting this for input/suggestions/comments.

I'm trying to determine the best possible output formats from my Motorola STB, Blu-ray player, and DVDO Edge video processor into my 32" Samsung LN32B640 display to ensure the best processing pipeline and maintain as much transparency as possible from source along the way to the display.

I understand a lot of this is try and see (i.e. evaluate with material, test patterns, etc.), and I do, but I would like to ensure I'm understanding things correctly and what's the best method/practice in theory, even if I may not notice it on this particular display/set up. Also to satisfy my curiosity and understanding of how things work. A bit OCD and anal, but please bear with me ;)

Some details first:

Source 1: STB Box (Comcast/Motorola DCH3416)

It's set to output 1080i, and I've enabled 480i pass-through for SD stations. That's straightforward as it connects over HDMI to the DVDO Edge, and the Edge will do the de-interlacing to 1080p (the resolution I'm outputting to the display). It can also be set to output either RGB or YCbCr 4:4:4. I don't want my devices to do YCbCr -> RGB conversions, so the goal is to feed YCbCr all the way to the display and let it convert to RGB. This is the cleanest IMO. Anyhow, my only option is then to output YCbCr 4:4:4 from the STB to the Edge.

Questions. Is this YCbCr 4:4:4 *8bit*? Is the source/programming 4:2:0 8bit? Is the STB upsampling chroma to 4:4:4 using 10bit processing and then back to 8bit for output? Curious. I assume yes, yes, and no, but would like to know.

Source 2: Blu-ray player (Sony BDP-S350)

It's set to output 1080p to the Edge over HDMI. Blu-ray 1080/24p titles go straight through the Edge unprocessed and to the display (120Hz and 5:5 pulldown with Samsung's AMP off). SD 480/60i DVDs (or the rare Blu-ray 1080/60i) get de-interlaced/scaled and sent to the Edge as 1080/60p, and the Edge applies PReP (progressive re-processing) for much better de-interlacing results than the S350 did and then outputs a "new" 1080/60p to the display. PReP really works nicely I have to say - especially on video and badly flagged DVDs.

Anyhow, on the S350 the YCbCr output options are 4:2:2 and 4:4:4. Are these both 8bit? Also, is the S350 using 10bit processing for the chroma upsampling and then back to 8bit for output (same question I had for the STB above)? It's not yet clear if it's better to output 4:2:2 or 4:4:4 - more below.

DVDO Edge:

This does all its processing/scaling in the YCbCr 4:2:2 *10bit* domain and can output either 4:2:2 10bit or 4:4:4 8bit to the display. So in essence, it's downsampling chroma from any 4:4:4 it will get (i.e. STB since that's the only choice and S350 if I should choose that for output). Because of this, to me, it would make more sense to output 4:2:2 from the S350 player since it will avoid an unnecessary chroma downsampling in the Edge. With the STB at 4:4:4 I have no choice. Not sure how big a deal this is or not, but again, bear with me.

Now, the final part is what should the Edge ouput? 4:2:2 10bit or 4:4:4 8bit? What is better for the display to receive before it converts from YCbCr to RGB? I don't think my display has a 10bit panel or 10bit processor, but even if it did, it may not matter. Samsung doesn't specify and I can't tell. I've looked at grayscale/color ramps and don't notice any differences in smoother gradations, banding, etc., but there may be more to look at, etc. Plus, it's for my understanding as I say.

So thanks for reading this long post. Again, this is to mainly satisfy my understanding of best practices, even as I said, I may not notice any differences unless viewing at 1" ;)

In essence I just want to ensure I'm choosing the best output options from each device in the chain and doing things correctly. So I'd like to know, with all I've described above in detail, what is the best output from the STB, Blu-ray S350, and Edge? And (for bonus information) if anyone knows, as I've pointed out above, what bits are used? 8bit or 10bit for the STB's 4:4:4 and S350's 4:2:2 or 4:4:4 outputs.

Looking forward to your comments!

Wow. This is actually a fairly complex topic, as asked, that has at least two major components. First would be the general "rules" to follow to ensure that you get the cleanest signal. The second would be all of the device-dependent exceptions that make general rules more like simple suggestions.

Since I am at LGA waiting to jump on a plane, let me answer the EDGE questions first, since they are easy. The simple answer is that if you can't see it, don't worry about it. The intermediate answer is that if you can't see it, then check your calibration. EDGE outputs 10-bit 4:2:2, and that is the preferred output if your display will accept that format (all HDMI 1.1 compliant devices ought to).

I'll try to write more later, though others should feel free to jump in.

Bill

Wow. This is actually a fairly complex topic, as asked, that has at least two major components. First would be the general "rules" to follow to ensure that you get the cleanest signal. The second would be all of the device-dependent exceptions that make general rules more like simple suggestions.

Well put. It is complex, but shouldn't be. The first component you mention (the rules to follow) are perhaps what should be addressed first. Once I (and I'm sure plenty other folks who seem confused) understand the rules, then we can get into why this device does x instead of y, exceptions, broken rules, etc. I know exactly what you mean. We need to understand the rules before we can understand why devices may deviate from them or not even follow them.

Bogdan

Before any intelligent discussion can begin a through understanding of what all this decoding means. I would suggest starting with Stacey Spears and Don Munsil's article on Chroma Upsampling Error (http://www.hometheaterhifi.com/the-dvd-benchmark/179-the-chroma-upsampling-error-and-the-420-interlaced-chroma-problem.html). It has a fairly good explanation embedded in it. As far as which piece of equipment should do the upsampling and what format it should output would require an analysis of decoder algorithms in each piece of eqipment. In addition we would need to consider the limitations of output for each piece of equipment (sometimes you can't stop a piece of hardware from decoding). I don't think we can reach a definitive conclusion for you. The best way is to just test as many options as you can think of and pick the best.

Before any intelligent discussion can begin a through understanding of what all this decoding means. I would suggest starting with Stacey Spears and Don Munsil's article on Chroma Upsampling Error (http://www.hometheaterhifi.com/the-dvd-benchmark/179-the-chroma-upsampling-error-and-the-420-interlaced-chroma-problem.html). It has a fairly good explanation embedded in it. As far as which piece of equipment should do the upsampling and what format it should output would require an analysis of decoder algorithms in each piece of eqipment. In addition we would need to consider the limitations of output for each piece of equipment (sometimes you can't stop a piece of hardware from decoding). I don't think we can reach a definitive conclusion for you. The best way is to just test as many options as you can think of and pick the best.

I read Stacy's article r.e. the CUE a while back. Years ago in fact when I had to dump my then top of the line Sony DVP-S9000 DVD player for a Panasonic RP-91 which didn't have the bug. I re-read it and it really doesn't answer my questions...

My questions in fact are pretty basic:

1. What is cable/broadcast material encoded in? (I assume 4:2:0 8bit)
2. Does anyone who has the same STB box know how it works? (i.e. the 10bit/8bit processing). After all, posting to this forum is to get info from people who have the same equipment and possible insight into that equipment, no?
3. What is the best output for the Sony S350? 4:2:2 or 4:4:4, or it doesn't matter? Is it 8bit? Again, someone who has an S350 can know this if it's connected to a receiver that tells it, etc. - Edge doesn't tell you the bits on inputs.
4. What is the best Edge output? 4:4:4 8bit or 4:2:2 10bit and why?

Those are the basics questions, and I guess I can post them as direct and separate items in individual forums. But I was also primarily looking at the correct way to obtain the most *transparency* along the way from source to display and a best practices. To say that the way to go is to test all options is missing the point. One should understand how things work, not just try this and that and say, "OK that looks best." Because what is best and what is correct/transparent are often two very different things. I don't want subjectivity. I want to know how it SHOULD work, even if it may not. For some person what looks best may be an uncalibrated TV image. But is that correct? Is that what was on the source? See what I mean?

I was hoping to see discussion on whether upsampling chroma to 4:2:2 or 4:4:4 is no different as you don't lose information, but you want to avoid downsampling if you can because you DO/can lose information there. That having more than 8bit precision when dealing with 8bit material is a good thing during processing (i.e. the Edge's 10bit 4:2:2 processing space), but perhaps not important when the data emerges at the other end. Stuff like this.THIS is what I would like to know, not what Stacey's article is about..I get that - it's straightforward.

But I guess the subject is perhaps too vast and complicated for someone to explain.

Here's a question. I keep seeing that it's best to let the Edge output 4:2:2 10bit. Maybe, even if I personally don't see a difference between that and 4:4:4 8bit output on my display. But if it IS the best, WHY is it the best - even in theory? The display will have to take the 4:2:2 10bit and go to 4:4:4 RGB 8bit. All with 8bit processing and 8bit panel in my case. Even if it had a 10bit processor I don't see any gain if the panel is 8bit. 10bit->10bit processing is same, not better. But maybe the Edge does the better job from 4:2:2 10bit -> 4:4:4 8bit on output, and then your display doesn't need to work much (a good thing). In which case 4:4:4 would be better Edge output. And I care about WHY and in theory. Not if I can see it or not. I already said I can't.

Bottom line is that this stuff should have clear answers. And the answers should not have to come from subjective interpretations of various permutations of the available options of devices. Is that how we calibrate our display devices? Devices hardly stick to standards unfortunately.... Just as we know that it's bad to turn on motion control on sets that have it if we want to see a film as it was at 24fps (i.e. no interpolated frames), just as we know that we should calibrate our TVs to certain standards, we should also know the proper way to pass chroma through to the display in the most *transparent* manner that stays as true to the source as possible. Once we know, then we can tackle individual components and how they are engineered - good or bad. Just as we know how TV sets are engineered good or bad. This was the general intent of my post. Bill divided it up nicely, but I want to know the background. If this stuff is written somewhere, please point me to it.

Bear5k gave you the answer, If your Displays native colour space is YcBcR 4:2:2 10bit then this is what you should set the Edge/VP50pro to output. provided your Source/Blu-Ray Player can output YcBcR 4:2:2 to take full advantage such as the Panasonic BD30, if not then you won't see A difference in the picture.....

I see A big difference switching the VP50pro's HDMI output between YcBcR 4:4:4 8bit & YcBcR 4:2:2 10bit but my projector excepts YcBcR 4:2:2 & my BD30 outputs YcBcR 4:2:2 so I am able to take advantage of the VP50pro's 10bit YcBcR 4:2:2 in / out capability....


Cheers....

I have a BDPS-350, a Lumagen HDQ and a JVC HD350 (RS10) and I was looking for this exact information myself yesterday. I had my system setup so that the S350 output RGB as I have another SD source that can only output 480/576i in RGB so I had mistakenly set the whole chain up to use this (including setting the HD350 to RGB for some reason, instead of Auto). Last night I set the S350 to output 4:2:2 into the Lumagen, but I've now realised that the Lumagen must have been converting this to RGB (due to the HD350 setting above) as I got the 'wrong' colours unless I left the Lumagen set to 'RGB 444'.

I thought the picture quality on both films (The Notebook and Terminator 2) was really good, though without rewatching them with the Lumagen outputing 4:2:2 and the HD350 doing the conversion I can't say which is better or if I'd notice. To be honest I'm not sure if my original setup was any worse, even if technically it might be. :o

I've found a placibo effect whereby if I think I've got something adjusted correctly (like greyscale for example) even though it might later turn out that the sensor was inaccurate or similar. I'll watch something convinced it looks better and enjoy it all the more, so I think I'll set 4:2:2 all the way through anyway and just enjoy. :)

Interesting reading from the link above BTW. I was aware of CUE as there is a setting for it on my HDQ. I had (correctly?) understood that it shouldn't be 'ON' for high definition sources, so that's how I have it configured, but it's nice to know the whys and wherefores. I don't actually know if my DVD player is one with the error as it's not listed (actually a DVD recorder, but it's a Sony HXD870). If the player doesn't have the error is it better to leave the CUE mode 'Off'? I suppose I could check using the Monster's Inc DVD that we have somewhere to check for the error.....

Bear5k gave you the answer, If your Displays native colour space is YcBcR 4:2:2 10bit then this is what you should set the Edge/VP50pro to output. provided your Source/Blu-Ray Player can output YcBcR 4:2:2 to take full advantage such as the Panasonic BD30, if not then you won't see A difference in the picture.....

I see A big difference switching the VP50pro's HDMI output between YcBcR 4:4:4 8bit & YcBcR 4:2:2 10bit but my projector excepts YcBcR 4:2:2 & my BD30 outputs YcBcR 4:2:2 so I am able to take advantage of the VP50pro's 10bit YcBcR 4:2:2 in / out capability....


Cheers....

Not really. I have already stated that the Sony S350 can output 4:2:2. I don't know if it's 8 or 10bit. I've also said my display can accept a 4:2:2 10bit from the Edge, but I don't see a difference. A display doesn't have to be 10bit processor or panel to accept 4:2:2 10bit. The question is still out there...

I know the display does'nt need to be 10bit to Except YcBcR 4:2:2 same as the source..

The point being IF you have 10 bit YcBcR 4:2:2 all the way so long as the source is 10bit the VP is 10bit & the display is 10bit then you will see A differance, if you cant see Any differance then one of the components in the video chain is'nt 10 bit...

Cheers...

My questions in fact are pretty basic:

Your questions are basic the answers are not!

1. What is cable/broadcast material encoded in? (I assume 4:2:0 8bit)
HD feeds can be 4:2:0 or 4:2:2 but most are 4:2:0, there are lots of SD 4:2:2 feeds. It's up to your cable company what they send their cable box.

2. Does anyone who has the same STB box know how it works? (i.e. the 10bit/8bit processing). After all, posting to this forum is to get info from people who have the same equipment and possible insight into that equipment, no?
Ask that question in the HDTV Recorders forum and you'll get a better response.



3. What is the best output for the Sony S350? 4:2:2 or 4:4:4, or it doesn't matter? Is it 8bit? Again, someone who has an S350 can know this if it's connected to a receiver that tells it, etc. - Edge doesn't tell you the bits on inputs.
A quick scan on the internet show no obvious problems with the decoding algorithms used by the Sony S350 on interlaced or progressive video. To quote Stacey Spears:
It's not terribly important whether the output device converts the 4:2:0 information to 4:2:2 or 4:4:4. Many video encoders can accept either format. But it has to be in one of those two formats. The video encoder needs to get a line of chroma for every line of luma, at the same time, because it's converting the digital video to analog video on the fly. It can't "remember" what the chroma information was for the previous scan line, so it can't interpolate missing chroma information. Only 4:2:2 and 4:4:4 have a 1:1 ratio of chroma and luma scan lines, so the MPEG decoder needs to output one or the other.


4. What is the best Edge output? 4:4:4 8bit or 4:2:2 10bit and why?
It doesn't matter as long as all the previous up sampling was done correctly and your display accepts the Edge output and correctly converts it in to RGB pixels. Now, if one of your components does something lame brain like up sampling using a 8 bit processor or use massive smoothing to cover up a simple minded up sample algorithm then it would matter. The Sony S350 probably doesn't, What your cable company does and how your STB handles it is anybodys guess.

We watched The Tale of Desperaux last night on Blu-Ray. In the movie, there are several scenes where rays of light come shining through windows. At the ends of the rays, I saw some mild banding. This is the type of thing you are looking for when trying to see what good the extra bits do you.

There are two ways to show fine gradation in color. One is with a higher resolution (lower pixel pitch) and dithering, and the other is with a higher actual bit-depth (bits per pixel). Extra resolution is also a Good Thing in its own right, and there is something of a tradeoff in perceived detail if you use extra pixels for dithering vs. actual detail. Banding is the most visible, and objectionable (IMHO), artifact of this tradeoff.

As promised, here are some general rules:
1) Scale once, and do it with the best scaler you have.
2) Deinterlace once, and do it with the best deinterlacer you have.
3) If you need to make small changes, consider seriously whether you should make them at all (e.g., slight zoom/overscan).
4) Use the highest real bit-depth you can.
5) Use more output bits than you have input bits.
6) Convert from YCbCr to RGB only once, and make sure it is done using the correct matrix.
7) The answer to too much noise reduction is not too much sharpening.
8) If you don't see the difference, leave the control the way it shipped from the factory.
9) Algorithms matter. Not all bits are created equal.
10) Your most cost-effective upgrade is better content.

The logic behind almost all of the above has to do with the integer math and limited bit-depths used in video processing circuitry. Any time you do division, and this includes multiplying by a fraction or a decimal value, you lose information. The goal is to minimize this as much as possible. You want to apply the least amount of processing to a signal as is practical before it gets to your eyeballs. On some controls, if you don't see the difference, then the control in question may be ineffective, ineffective in your environment, or you may have a PIBKAC, as they say in tech support circles.

To the specifics of the EDGE, let's look at rule #4. The EDGE uses 10-bit 4:2:2 processing. As a result, everything upstream from it will have the content converted to this format. If your Blu Ray player outputs 12-bits, guess what? You still have a 10-bit 4:2:2 system. In reality, your Blu-Ray player is probably simply padding the data, since what is on-disc is 8-bit 4:2:0, but that's a discussion specific to your Blu Ray player. MPEG2 from OTA or Cable? 4:2:0 off the wire, though it comes down from the satellite at 45Mbps 4:2:2, and gets re-compressed.

In either case, you want EDGE to see as close to that 4:2:0 (typically a 4:2:2) signal as you can get, and then have it work its magic, and then don't touch it after that. If EDGE is not your best video processor, then you have a more fundamental problem.

Now, let's see what questions come. As for device-specific questions, you are basically on your own. Manuals and marketing tear-sheets help, as well as reviews from people who have the gear to tell the difference and the ear of the people who can answer those questions.

Bill

I have a BDPS-350, a Lumagen HDQ and a JVC HD350 (RS10) and I was looking for this exact information myself yesterday. I had my system setup so that the S350 output RGB as I have another SD source that can only output 480/576i in RGB so I had mistakenly set the whole chain up to use this (including setting the HD350 to RGB for some reason, instead of Auto). Last night I set the S350 to output 4:2:2 into the Lumagen, but I've now realised that the Lumagen must have been converting this to RGB (due to the HD350 setting above) as I got the 'wrong' colours unless I left the Lumagen set to 'RGB 444'.


Just bear in mind the Lumagen Vision products use DVI not HDMI so the output is RGB only.

D

Just bear in mind the Lumagen Vision products use DVI not HDMI so the output is RGB only.

D

I didn't realise that difference between DVI and HDMI, so that means I'm reliant on the HDQ doing the conversion to RGB anyway (rather than my HD350 PJ), which I would hope does a better job than the player generally speaking? Given the HDQ is about 5 or 6 times the cost of the player, I'd expect it to anyway. I have seen banding occasionally, but that was from an SD TV source which is output in RGB at the source, though it is probably more due to compression and low bit rate of UK digital TV.

We watched The Tale of Desperaux last night on Blu-Ray. In the movie, there are several scenes where rays of light come shining through windows. At the ends of the rays, I saw some mild banding. This is the type of thing you are looking for when trying to see what good the extra bits do you.

There are two ways to show fine gradation in color. One is with a higher resolution (lower pixel pitch) and dithering, and the other is with a higher actual bit-depth (bits per pixel). Extra resolution is also a Good Thing in its own right, and there is something of a tradeoff in perceived detail if you use extra pixels for dithering vs. actual detail. Banding is the most visible, and objectionable (IMHO), artifact of this tradeoff.

As promised, here are some general rules:
1) Scale once, and do it with the best scaler you have.
2) Deinterlace once, and do it with the best deinterlacer you have.
3) If you need to make small changes, consider seriously whether you should make them at all (e.g., slight zoom/overscan).
4) Use the highest real bit-depth you can.
5) Use more output bits than you have input bits.
6) Convert from YCbCr to RGB only once, and make sure it is done using the correct matrix.
7) The answer to too much noise reduction is not too much sharpening.
8) If you don't see the difference, leave the control the way it shipped from the factory.
9) Algorithms matter. Not all bits are created equal.
10) Your most cost-effective upgrade is better content.

The logic behind almost all of the above has to do with the integer math and limited bit-depths used in video processing circuitry. Any time you do division, and this includes multiplying by a fraction or a decimal value, you lose information. The goal is to minimize this as much as possible. You want to apply the least amount of processing to a signal as is practical before it gets to your eyeballs. On some controls, if you don't see the difference, then the control in question may be ineffective, ineffective in your environment, or you may have a PIBKAC, as they say in tech support circles.

To the specifics of the EDGE, let's look at rule #4. The EDGE uses 10-bit 4:2:2 processing. As a result, everything upstream from it will have the content converted to this format. If your Blu Ray player outputs 12-bits, guess what? You still have a 10-bit 4:2:2 system. In reality, your Blu-Ray player is probably simply padding the data, since what is on-disc is 8-bit 4:2:0, but that's a discussion specific to your Blu Ray player. MPEG2 from OTA or Cable? 4:2:0 off the wire, though it comes down from the satellite at 45Mbps 4:2:2, and gets re-compressed.

In either case, you want EDGE to see as close to that 4:2:0 (typically a 4:2:2) signal as you can get, and then have it work its magic, and then don't touch it after that. If EDGE is not your best video processor, then you have a more fundamental problem.

Now, let's see what questions come. As for device-specific questions, you are basically on your own. Manuals and marketing tear-sheets help, as well as reviews from people who have the gear to tell the difference and the ear of the people who can answer those questions.

BillYour much more patient than I am bill :rolleyes:...

Cheers mate...

Just bear in mind the Lumagen Vision products use DVI not HDMI so the output is RGB only.

D I see the new High end version Lumagen "XG" radiance with HDMI 1.3 offers 10-bit processing, is this only if/ when sending YcBcR colour space or is RGB included, and what about when you include the CMS still 10bit all the way with no bottle neck??...

Cheers....

I see the new High end version Lumagen "XG" radiance with HDMI 1.3 offers 10-bit processing, is this only if/ when sending YcBcR colour space or is RGB included, and what about when you include the CMS still 10bit all the way with no bottle neck??...
RGB is converted to YCbCr for processing, and is left that way for almost all steps in the pipeline. I had thought that they were using 12-bit, but that may just be in the output stage.

Bear5k gave you the answer, If your Displays native colour space is YcBcR 4:2:2 10bit then this is what you should set the Edge/VP50pro to output. provided your Source/Blu-Ray Player can output YcBcR 4:2:2 to take full advantage such as the Panasonic BD30, if not then you won't see A difference in the picture.....

I see A big difference switching the VP50pro's HDMI output between YcBcR 4:4:4 8bit & YcBcR 4:2:2 10bit but my projector excepts YcBcR 4:2:2 & my BD30 outputs YcBcR 4:2:2 so I am able to take advantage of the VP50pro's 10bit YcBcR 4:2:2 in / out capability....


Cheers....

Yes, but I am interested in transparency as I said. I don't want colors that aren't in the source. I don't want extended colorspaces to display those colors for the point of this conversation. And I don't know of any Blu-ray that's not 4:2:0 8bit and in the 709 space...So while your player may pad bits, pass it through, and the display also use that with a larger colorspace, it's hardly transparent, which is my point. You may prefer the look, but I want only what is in the source. Until Blu-ray goes 10bit, that will never be the case.

We watched The Tale of Desperaux last night on Blu-Ray. In the movie, there are several scenes where rays of light come shining through windows. At the ends of the rays, I saw some mild banding. This is the type of thing you are looking for when trying to see what good the extra bits do you.

There are two ways to show fine gradation in color. One is with a higher resolution (lower pixel pitch) and dithering, and the other is with a higher actual bit-depth (bits per pixel). Extra resolution is also a Good Thing in its own right, and there is something of a tradeoff in perceived detail if you use extra pixels for dithering vs. actual detail. Banding is the most visible, and objectionable (IMHO), artifact of this tradeoff.

As promised, here are some general rules:
1) Scale once, and do it with the best scaler you have.
2) Deinterlace once, and do it with the best deinterlacer you have.
3) If you need to make small changes, consider seriously whether you should make them at all (e.g., slight zoom/overscan).
4) Use the highest real bit-depth you can.
5) Use more output bits than you have input bits.
6) Convert from YCbCr to RGB only once, and make sure it is done using the correct matrix.
7) The answer to too much noise reduction is not too much sharpening.
8) If you don't see the difference, leave the control the way it shipped from the factory.
9) Algorithms matter. Not all bits are created equal.
10) Your most cost-effective upgrade is better content.

The logic behind almost all of the above has to do with the integer math and limited bit-depths used in video processing circuitry. Any time you do division, and this includes multiplying by a fraction or a decimal value, you lose information. The goal is to minimize this as much as possible. You want to apply the least amount of processing to a signal as is practical before it gets to your eyeballs. On some controls, if you don't see the difference, then the control in question may be ineffective, ineffective in your environment, or you may have a PIBKAC, as they say in tech support circles.

To the specifics of the EDGE, let's look at rule #4. The EDGE uses 10-bit 4:2:2 processing. As a result, everything upstream from it will have the content converted to this format. If your Blu Ray player outputs 12-bits, guess what? You still have a 10-bit 4:2:2 system. In reality, your Blu-Ray player is probably simply padding the data, since what is on-disc is 8-bit 4:2:0, but that's a discussion specific to your Blu Ray player. MPEG2 from OTA or Cable? 4:2:0 off the wire, though it comes down from the satellite at 45Mbps 4:2:2, and gets re-compressed.

In either case, you want EDGE to see as close to that 4:2:0 (typically a 4:2:2) signal as you can get, and then have it work its magic, and then don't touch it after that. If EDGE is not your best video processor, then you have a more fundamental problem.

Now, let's see what questions come. As for device-specific questions, you are basically on your own. Manuals and marketing tear-sheets help, as well as reviews from people who have the gear to tell the difference and the ear of the people who can answer those questions.

Bill

Thanks Bill. 4&5 is what I'm "struggling" with. The rest is straightforward in my head. In reality, I don't have many options. The STB can only output 4:4:4. My option is really the Sony S350 - 4:2:2 or 4:4:4, and the Edge - 4:2:2 or 4:4:4.

Anyhow, searching around, it seems the Sony S350 4:2:2 and 4:4:4 are both 8bit. Why Sony provides these two options when both are 8bit I don't get. What display device can't deal with 4:2:2 or 4:4:4 component when going to RGB???:confused: Weird..I understand the Edge, since its 4:2:2 is 10bit and its 4:4:4 8bit. But the Sony makes no sense to me with both being 8bit. Anyhow..

An interesting thing is that my guess is that it's best to keep chroma DOWNsampling to a minimum. UPsampling is lossless from 4:2:0 -> 4:2:2 or 4:4:4. But when you downsample 4:4:4 to 4:2:2 you may lose. So in my case, the thing to do is set the Sony S350 to output 4:2:2 to the Edge, since the Edge works in 4:2:2. This prevents an unnecessary chroma downsampling, no?

As far as the final output, yes in theory it's best to have 4:2:2 10bit out from the Edge. I get it now. It's just that in my case it doesn't make a bit (no pun intended) of difference. My display will 8bit process and display that 4:2:2 10bit to RGB 8bit. If I had set the Edge to output 4:4:4 8bit, the result is the same. As I said in my post above, if one has a 10bit processor/panel and prefers to use a larger colorspace, etc., then that's an individual call. I am talking transparency.

Thanks Bill. 4&5 is what I'm "struggling" with. The rest is straightforward in my head. In reality, I don't have many options. The STB can only output 4:4:4. My option is really the Sony S350 - 4:2:2 or 4:4:4, and the Edge - 4:2:2 or 4:4:4.

Anyhow, searching around, it seems the Sony S350 4:2:2 and 4:4:4 are both 8bit. Why Sony provides these two options when both are 8bit I don't get. What display device can't deal with 4:2:2 or 4:4:4 component when going to RGB???:confused: Weird..I understand the Edge, since its 4:2:2 is 10bit and its 4:4:4 8bit. But the Sony makes no sense to me with both being 8bit. Anyhow..
I expect that Sony is trying to provide a videophile-friendly mode where they only do the intermediate upsampling (i.e., what most likely comes out of the decoder chip), rather than the full conversion fron 4:2:0 -> 4:4:4. In many cases, the conversion from 4:2:0 to 4:2:2 is a separate stage in the processing pipeline inside the source component than the conversion from 4:2:2 to 4:4:4.

An interesting thing is that my guess is that it's best to keep chroma DOWNsampling to a minimum. UPsampling is lossless from 4:2:0 -> 4:2:2 or 4:4:4. But when you downsample 4:4:4 to 4:2:2 you may lose. So in my case, the thing to do is set the Sony S350 to output 4:2:2 to the Edge, since the Edge works in 4:2:2. This prevents an unnecessary chroma downsampling, no?
Yes, this is yet another type of processing operation that not only has limitations due to the type of math being used, but has its own engineering challenges (e.g., CUE).

As far as the final output, yes in theory it's best to have 4:2:2 10bit out from the Edge. I get it now. It's just that in my case it doesn't make a bit (no pun intended) of difference. My display will 8bit process and display that 4:2:2 10bit to RGB 8bit. If I had set the Edge to output 4:4:4 8bit, the result is the same. As I said in my post above, if one has a 10bit processor/panel and prefers to use a larger colorspace, etc., then that's an individual call. I am talking transparency.
There is a reason why these general rules often fail in the face of specific implementations. However, even 8-bit panels can have higher bit-depths leading into the imager. In such cases, it is still a good idea to minimize the amount of processing going into the display. I would check the calibration of the set and/or check at what distance you are watching it.

If the set encompases a limited field-of-view (FOV), then you are probably getting a bit of a dithering effect at your seating distance. There are various seating distance calculators around, and you generally want to be within the cone where you eye can theoretically make out an individual pixel to have a chance to avoid a dithering effect (not necessarily a good or bad thing, merely an optical/psychovisual thing). To make this last point more plain, in my HT, I can see a lot more artifacts on my 77" wide projector screen driven by a 1080p projector at 11' than I can see on the 768p 50" plasma that is 12.5' from my usual seating position.

Bill

I expect that Sony is trying to provide a videophile-friendly mode where they only do the intermediate upsampling (i.e., what most likely comes out of the decoder chip), rather than the full conversion fron 4:2:0 -> 4:4:4. In many cases, the conversion from 4:2:0 to 4:2:2 is a separate stage in the processing pipeline inside the source component than the conversion from 4:2:2 to 4:4:4.

I'm still waiting confirmation that they are both 8bit, but basically what you're saying is that if so, 4:2:2 would be the way to go out of the Sony Blu-ray and into the Edge, correct? Especially if we're avoiding a chroma downsampling in the Edge by sending 4:2:2. Really the options are:

Sony 4:2:2 8bit -> Edge 4:2:2 10bit
Sony 4:4:4 8bit -> Edge 4:2:2 10bit

And I guess, Sony 4:2:2 makes the most sense.


There is a reason why these general rules often fail in the face of specific implementations. However, even 8-bit panels can have higher bit-depths leading into the imager. In such cases, it is still a good idea to minimize the amount of processing going into the display. I would check the calibration of the set and/or check at what distance you are watching it.

If the set encompases a limited field-of-view (FOV), then you are probably getting a bit of a dithering effect at your seating distance. There are various seating distance calculators around, and you generally want to be within the cone where you eye can theoretically make out an individual pixel to have a chance to avoid a dithering effect (not necessarily a good or bad thing, merely an optical/psychovisual thing). To make this last point more plain, in my HT, I can see a lot more artifacts on my 77" wide projector screen driven by a 1080p projector at 11' than I can see on the 768p 50" plasma that is 12.5' from my usual seating position.

Bill

My set is properly calibrated, and I can correctly see the Pixel Phase 1080x1920 pattern from DVE from my normal viewing distance. I'm well within the viewing distance where I can see every detail. I'm not saying I don't see occasional artifacts or banding, posterization, etc., just that there is hardly a difference no matter the Edge's output. That's all. I guess what you're saying is pass it all through to the set, and let the set do the conversion to RGB (which I am), regardless if it's coming in as component 4:2:2 10bit or 4:4:4 8bit. Just let the set do that final processsing. I understand, but always thought that the video processor would do a better job than the display device.

I understand, but always thought that the video processor would do a better job than the display device.
This gets back to the implied aphorism in the general rules: "a chain is only as strong as its weakest link". An external VP can alleviate issues in the upper-level processing stages (e.g., deinterlacing, scaling), but once the signal makes it to the formatter and imager (e.g., de-gamma), you are strictly at the mercy of the display and the room surrounding it.

As an aside: the benefit to having a "clean" signal path is that you know with reasonable confidence when you hit a limitation of the source or the display, rather than something introduced in-between.

Bill

PS: Post 7,001. Yikes, I need another hobby.

This gets back to the implied aphorism in the general rules: "a chain is only as strong as its weakest link". An external VP can alleviate issues in the upper-level processing stages (e.g., deinterlacing, scaling), but once the signal makes it to the formatter and imager (e.g., de-gamma), you are strictly at the mercy of the display and the room surrounding it.

As an aside: the benefit to having a "clean" signal path is that you know with reasonable confidence when you hit a limitation of the source or the display, rather than something introduced in-between.

Bill

Quite right. The problem is that displays seem to be the weakest links, since they don't stick to standards. The only display I've seen that I can say is reference quality (picture wise and I also looked at the #s while my friend was calibrating it and it's dead on accurate), is the Samsung DLP Projector 800B. That is a truly amazing display device, and no doubt because of Joe Kane.

Anyway, I know my display is the weakest link, and that the Edge can convert YCbCr to RGB better, but the problem is that the display will take RGB and go back to YCbCr, muck with it, and then to RGB again. And in the end that's worse - I've seen it.

Anyhow, thanks for all your input!

PS Was I correct above in that it's best to feed 4:2:2 from the Sony to the Edge? One because it's less processing in the Sony (as you said - a separate stage), and second because it prevents an unnecessary chroma downsampling in the Edge had I fed it 4:4:4. I very much doubt the 4:4:4 is 10bit, maybe then it would be a better choice. But I just can't imagine it is.

PS Was I correct above in that it's best to feed 4:2:2 from the Sony to the Edge? One because it's less processing in the Sony (as you said - a separate stage), and second because it prevents an unnecessary chroma downsampling in the Edge had I fed it 4:4:4. I very much doubt the 4:4:4 is 10bit, maybe then it would be a better choice. But I just can't imagine it is.
From a signal integrity standpoint, you want to output 4:2:2 from the Sony.