Home Theater Forum and Systems banner

21 - 40 of 80 Posts

·
Registered
Joined
·
2,604 Posts
So I've been trying to find the time to play around with some of my TVs settings and my colorimeter to try and answer some of my own questions regarding exactly how some of the "classic" calibration graphs are affected by each of the different controls... suffice to say my life just keeps getting busier and I can't find the time, so I thought I'd begin picking the brains of those who already know. Not as much fun, but certainly easier, and then others can see and learn as well.

So this might be the first of many, but I'll space them out to give you guys a break...

Let's take the normalized luminance curve (% luminance vs. % stim) and how it's affected by changing the gamma of a display... I've often thought of this curve as a rope strung between two points, those being black (0% stim) and white (100% stim)... The analogy goes that as you increase gamma, it's like paying out more string, the curve hangs lower, as you take in more string, you decrease gamma, the curve hangs higher...

Taken by itself, is there any particular flaw in thinking about it in this way?
 

·
Plain ole user
Joined
·
11,121 Posts
That would be a good rough analogy. The important thing to get about gamma, however is that there are many variables at play that can be very different with different sets. Gamma needs to be thought of as a complete curve that can be complex and may not be described by a single average value. The whole rope is important, and the stiffness of the rope may vary along its lenght. Gamma may interact with other adjustments, and the exact nature of those interactions may be very different on your set than others.
 

·
Registered
Joined
·
44 Posts
Discussion Starter #25
As lcaillo points out, it's a good rough analogy. However, depending upon where you are at in the stimulus level, you may or may not want to make different types of adjustments. In some cases, you may want to elevate the "average" gamma value to increase shadow detail. Other people may want to lower it to improve black level. The interplay of shadow detail and black level may differ entirely from what you want to do in the mid-tones and highlights. It's far better to get a reasonably granular view of gamma to correlate what you are seeing, what you want to see and the effects of specific changes to converge the two.
 

·
Registered
Joined
·
2,604 Posts
OK. So your responses made me think so hard it tok way longer to ask the nedxt question than I had intended... Thanks!
I appreciate the detail of not all machines operating the same way, and some being "stiff" at certain levels and the like, I hadn't thought about that... it's my hope to better understand how things should work in the theoretical world, as well as how often/not that applies in the real world...
To that end, I think you both said, yes, that's a good theoretical model even though it holds up only to some degree in the real world...
So the next question is, I believe I've read somewhere that on some displays that don't have gamma adjustments per se, that sometimes to can slightly tweak gamma by adjusting brightness and contrast, so I ask, using the rope analogy, if you increase contrast, looking at a luminance curve (not yet normalized) as you increase contrast you would expect the high end of the curve to rise....
The question is, as it rises, what happens to the length of the rope, does it stay the same, giving less contour to the curve? Or does it elongate, but not proportionately, so there's less contour to the curve? Or does it elongate ot the point where the contour stays the same?
I'm sure you can guess the next question, which is what then happens when looking at the normalized luminance curve, but I think I've answered that one already... I think the answer will be that in the ideal world, the since you're not ajdjusting it specifically, the gamma would stay the same, and the before/after curves would overlap, assuming you weren't driving one into run-out or the like...?
 

·
Registered
Joined
·
44 Posts
Discussion Starter #27
OK. So your responses made me think so hard it tok way longer to ask the nedxt question than I had intended... Thanks!
I appreciate the detail of not all machines operating the same way, and some being "stiff" at certain levels and the like, I hadn't thought about that... it's my hope to better understand how things should work in the theoretical world, as well as how often/not that applies in the real world...
To that end, I think you both said, yes, that's a good theoretical model even though it holds up only to some degree in the real world...
So the next question is, I believe I've read somewhere that on some displays that don't have gamma adjustments per se, that sometimes to can slightly tweak gamma by adjusting brightness and contrast, so I ask, using the rope analogy, if you increase contrast, looking at a luminance curve (not yet normalized) as you increase contrast you would expect the high end of the curve to rise....
The question is, as it rises, what happens to the length of the rope, does it stay the same, giving less contour to the curve? Or does it elongate, but not proportionately, so there's less contour to the curve? Or does it elongate ot the point where the contour stays the same?
I'm sure you can guess the next question, which is what then happens when looking at the normalized luminance curve, but I think I've answered that one already... I think the answer will be that in the ideal world, the since you're not ajdjusting it specifically, the gamma would stay the same, and the before/after curves would overlap, assuming you weren't driving one into run-out or the like...?
Since in theory changing either the brightness or the contrast control ought not to change the gamma for reasonable values of all three, there really isn't a hard-and-fast rule. A properly engineered display will, if you compress the dynamic range too much, lower the effective gamma to maintain evenness between black and white. In other words, the gamma will fall as black and white get closer together (e.g., brightness too high, contrast too low). That being said, with current technologies, a well-engineered display ought never to get that far short of "stupid human tricks".

Ideally, you would want the converse to be true, as well. As the dynamic range increases, the gamma value increases -- at least up to a point. If you have a "bat cave", then you can appreciate and use an average gamma that is quite a bit higher than one meant for a living room with a fair bit of ambient light. Ambient light requires an elevated black level to maintain a visually desirable amount of shadow detail, and given a fixed maximum brightness, this means that you end up losing dynamic range as a result.
 

·
Registered
Joined
·
2,604 Posts
So then, on a properly engineered display, for "reasonable" values of contrast and brightness, you might expect increasing contrast or decreasing brightness to increase the gamma slightly, but probably by very little? If so...
Would both affect the whole range equally, or would contrast affect more the gamma at the high end and brightness more at the low end?
And how many of modern displays are "properly engineered" in this way, as in, how high end do you have to go to find this behavior vs how low end do you have to go to find these controls affecting gamma more than they should, or in ways they shouldn't?
 

·
Plain ole user
Joined
·
11,121 Posts
I would consider a properly engineered display to be one that accounts for non-linear luma in the contrast and brightness controls and has no effect on gamma. In practice that relationship is hard to get right, as there are multiple variables at play, not just the video relationships, but also the display technology and its implementation.

In practice, it just depends on the display. It is very hard to generalize. The effects vary from one manufacturer to another and even within brands considerably.
 

·
Registered
Joined
·
44 Posts
Discussion Starter #31
And how many of modern displays are "properly engineered" in this way, as in, how high end do you have to go to find this behavior vs how low end do you have to go to find these controls affecting gamma more than they should, or in ways they shouldn't?
Good video engineering isn't necessarily cost-prohibitive on the BOM, AFAIK. What might push it higher into the food chain is the cost of the people and giving them the time to do it right.
 

·
Registered
Joined
·
2,604 Posts
Perhaps, but I always assume that increase time = increased development costs = increase production price... ROI and all... no?
 

·
Registered
Joined
·
44 Posts
Discussion Starter #33
Selling price is whatever the market agrees to pay, and time-to-market is a critical piece of that puzzle. As a result, many manufacturers may not be able to spend the incremental cycles trying to get the engineering right while they are busy pushing next year's model through the development pipe.

As I said, the increased cost usually isn't reflected in the BOM (i.e., the components and assembly costs), but in the cost of development itself.
 

·
Registered
Joined
·
2,604 Posts
Selling price is whatever the market agrees to pay, and time-to-market is a critical piece of that puzzle. As a result, many manufacturers may not be able to spend the incremental cycles trying to get the engineering right while they are busy pushing next year's model through the development pipe.

As I said, the increased cost usually isn't reflected in the BOM (i.e., the components and assembly costs), but in the cost of development itself.
Well, this is getting off-topic, but if MFRs can't get the volume they need at the price they need to recoup the cost of the development itself, the business model isn't viable. The return must recoup the investment. Whether it's reflected in the BOM is immaterial. The BOM is just the beginning of costing a product.
 

·
Registered
Joined
·
44 Posts
Discussion Starter #35
Well, this is getting off-topic, but if MFRs can't get the volume they need at the price they need to recoup the cost of the development itself, the business model isn't viable. The return must recoup the investment. Whether it's reflected in the BOM is immaterial. The BOM is just the beginning of costing a product.
It's very much material, pardon the pun. BOM costs are roughly equivalent to variable costs, whereas development engineering costs are fixed (insensitive to volume). Given that a lot of display development is done as assembly and integration, especially at the low-end of the market, if you can trace the components back to their source, you can probably find a vendor who is "not doing it right" and then extrapolate to other, similar displays using the same chipsets.

After all, this tangent was started as a theoretical question that kept trying to get back into the specifics of what's available in the market. ;)
 

·
Registered
Joined
·
2,604 Posts
Very punny. You do make some good points. But even so, I was asking more about the higher end of the market. Also, would the chipsets from companies "doing it right" not cost more, as in why else would chipsets "doing it wrong" get bought? Unless the lower cost MFRs don't bother testing whether it's right or wrong, but that again implies the ones who do would have higher development costs...

I suppose at some point, to avoid this going further OT, I'll have to simply accept what you're saying even though it doesn't make sense to me... :huh:

Maybe I should start a new thread to discuss this further. I really don't want to cloud the otherwise-very-useful "calibration" thread with this talk...
 

·
Registered
Joined
·
44 Posts
Discussion Starter #37
Very punny. You do make some good points. But even so, I was asking more about the higher end of the market. Also, would the chipsets from companies "doing it right" not cost more, as in why else would chipsets "doing it wrong" get bought? Unless the lower cost MFRs don't bother testing whether it's right or wrong, but that again implies the ones who do would have higher development costs...

I suppose at some point, to avoid this going further OT, I'll have to simply accept what you're saying even though it doesn't make sense to me... :huh:

Maybe I should start a new thread to discuss this further. I really don't want to cloud the otherwise-very-useful "calibration" thread with this talk...
This thread has been relatively lightly traveled, so I don't mind going into a bit more detail. The higher-end companies have, often, done their own in-house engineering work, while the lower-end companies basically buy "kits" and assemble them (this is gross oversimplification, but work with me...). If I'm a Sanyo or Westinghouse or Vizio, then I buy a package from someone like Sigma, who then gets to spread its own engineering talent around a lot of different models from a lot of different OEMs. Conversely, if I'm someone like Sharp or Panasonic, then I (might) use internally-developed sub-assemblies (e.g., Uniphier) and spread those over large swathes of my product range. At that point, developing a specific model is about setting options in a firmware platform and tying it to the hardware.

Who gets caught out in the above business/development models were companies like Fujitsu or Pioneer who had low volume AND did a lot of custom development. Those manufacturers are basically all gone now.
 

·
Registered
Joined
·
2,604 Posts
Ah. So what makes the difference (sometimes at least) is the MFR making a conscious decision to disable or not use certain things they've developed as an excuse to offer a lower grade product at a lower price point, or basically positioning. I have no problem buying that (pardon my own pun). But don't they tend to roll out improved processing/features in the higher end sets first, and after a while (year or more) bring those down to the lower end models once they have something better for their high end models, in effect charging more for the best or most up to date stuff, and isn't that equivalent to the "early adopters" of said improvements paying for the proprietary costs?
 

·
Registered
Joined
·
44 Posts
Discussion Starter #39
Ah. So what makes the difference (sometimes at least) is the MFR making a conscious decision to disable or not use certain things they've developed as an excuse to offer a lower grade product at a lower price point, or basically positioning. I have no problem buying that (pardon my own pun). But don't they tend to roll out improved processing/features in the higher end sets first, and after a while (year or more) bring those down to the lower end models once they have something better for their high end models, in effect charging more for the best or most up to date stuff, and isn't that equivalent to the "early adopters" of said improvements paying for the proprietary costs?
Now we really do start going off the reservation. Don't necessarily mistake a generalized firmware process as either a) the only way to develop a TV set, or b) necessarily being solely a software process. In many cases, you not only "turn off" something in software, but entire sub-assemblies are then not included in hardware. If you pull the boards out of a TV, you can often find "blank" soldering pads and even connectors where something might have gone to create a higher-end TV.

As for early adopters paying higher prices, that's basically a given. The question is how much for what features. I am distinctly not a fan of the 240Hz LCDs, but for other people they are a must-have item. The way semiconductor manufacturing works, those panels might undergo exactly the same manufacturing steps to create "the glass" as a panel that only targets a 60Hz refresh rate, but since it passes a tighter quality check at the end, it gets "binned" into the 240Hz stack. The economic cost is the same, but because of the miracle of the bell curve and supply/demand, people pay more for it. The company producing it may then load more of the "costs" of production onto that panel, rather than another one, but the economic processing through a certain point would be similar.

The above is a GROSS oversimplification of semiconductor (and LCD panel) production processes, but you get the idea. The biggest misconception that many people have of "digital" electronics is that binary math rules everything -- that all things electronic are exactly the same (1 or 0) when they are built. Such a concept doesn't hold up anywhere along the production chain, which is why you have "binning" and why, wait for it, copying settings from one display to another is more likely to be wrong than it is to be right.

How's that for getting us back on topic? :help::sneeky::paddle::T
 

·
Registered
Joined
·
2,604 Posts
In many cases, you not only "turn off" something in software, but entire sub-assemblies are then not included in hardware. If you pull the boards out of a TV, you can often find "blank" soldering pads and even connectors where something might have gone to create a higher-end TV.
But that would bring us back to a real BOM difference. :bigsmile:
As for early adopters paying higher prices, that's basically a given. The question is how much for what features. I am distinctly not a fan of the 240Hz LCDs, but for other people they are a must-have item. The way semiconductor manufacturing works, those panels might undergo exactly the same manufacturing steps to create "the glass" as a panel that only targets a 60Hz refresh rate, but since it passes a tighter quality check at the end, it gets "binned" into the 240Hz stack. The economic cost is the same, but because of the miracle of the bell curve and supply/demand, people pay more for it. The company producing it may then load more of the "costs" of production onto that panel, rather than another one, but the economic processing through a certain point would be similar.
Ah, similar to the precision resistors I try to avoid designing in where I can avoid it...
Such a concept doesn't hold up anywhere along the production chain, which is why you have "binning" and why, wait for it, copying settings from one display to another is more likely to be wrong than it is to be right.

How's that for getting us back on topic? :help::sneeky::paddle::T
Bravo, Maestro! :T I actually saw that coming just a few seconds early, but kudos nonetheless!
 
21 - 40 of 80 Posts
Top