WinTV-PVR-250/350/500 Which one to get?

[reboot]

Yes. The Firefly is a good choice, because it uses RF, not IR, so you don't need line of sight.
I was just pointing out, that if you're buying new, gettin two 150's is probably the best option, unless you can get a 500 and a Firefly for less money, or you absolutely NEED an RF remote.

[Wilbert]

@davideck,

Thanks for the short and correct explanation!

[davideck]

You're welcome Wilbert. It's good to see some appreciation for theory.

[SatStorm]

Wilbert and Davideck,
Since you two know what is correct and what is wrong, and since you know so much the theory behind all this, I believe you both are able to point to the rest of us, the known hardware and the brands, capable to succeed the 100% of what the theory says in practice.

What I wish to see, is a buy "this", "this", "this" and "this" guide, so to be able to transfer my VHS / SVHS tapes the best possible way on DVDs, according to what the theory says.

I believe this community here need this.
We all are waiting

[davideck]

SatStorm - in your own words...

Davideck we don't "debate" on something here. We simply testify what we know about a hobbie we both love! I don't have to proove something to you, neither you have to proove something to me. We have our results, thoughts and tests and simply post them so others to read and get informed! They gonna choose, we won't tell them what is right and what is wrong!

I have tried my best to explain sampling theory as it pertains to analog video.
What is still missing is an explanation of your "canvas" as it pertains to analog video.
Perhaps we should start there...

[edDV]

I hope this helps.

In Photoshop the "canvas" is the backgound frame. Here I created a blank NTSC frame of 640x486. Square pixels would produce 640x480 but the NTSC spec calls for 486 lines (243 per field) in the active picture area so I added 3 additional lines to the top and bottom to fill the space. I made the canvas blue to make it show up. Normally this would be colored 7.5 IRE black. PAL would be 768x576 at 0,0 IRE black.

Also shown below is a Hi8 capture shown in square pixels. The edge defects will be cropped and the result pasted (composited) into the center of the canvas (NTSC frame) without scaling.



And the result


TV overscan would mask the canvas area. It would show as a black border when viewed on the computer.

[davideck]

Yes, the canvas is the sampling grid of pixels that is created after capturing an analog video signal into the digital domain. The canvas can arbitrarily be 720x480, 640x480, 352x240, etc.
There is no inherent canvas size associated with an analog video signal.

[edDV]

Yes, the canvas is the sampling grid of pixels that is created after capturing an analog video signal into the digital domain. The canvas can arbitrarily be 720x480, 640x480, 352x240, etc.
There is no inherent canvas size associated with an analog video signal.

True but there is an inherent luminance resolution to analog video and for NTSC Hi8, sampling somewhere between 3fsc (10.7 MHz.) and 4fsc (14.4 MHz.) is customary. CCIR-601 13.5 Mhz has become the standard sampling rate for both PAL and NTSC and results in the familiar 704x480/576 raster, albeit with non-square pixels.

[davideck]

True but there is an inherent luminance resolution to analog video and for NTSC Hi8, sampling somewhere between 3fsc (10.7 MHz.) and 4fsc (14.4 MHz.) is customary. CCIR-601 13.5 Mhz has become the standard sampling rate for both PAL and NTSC and results in the familiar 704x480/576 raster, albeit with non-square pixels.

I think that is the issue at hand.
What sample rates are appropriate for the various source resolutions (VHS, S-VHS, Hi-8, etc)?
BTW, CCIR-601 is 720, but who's counting?

[SatStorm]

Davideck, I didn't ask you to explain to me the sampling theory as it pertains to analog video. So, let me ask you once more, in a more straight way: Please, point me hardware to use so to capture the 100% of a signal.

Regarding the canvas.
Canvas here, is the word to determine the area of the analogue TV Screen. Inside that area, any kind of analogue video info, scretch so to fill it. This is how the analogue works: By zooming the info so to fill the area.
Capture cards don't know about this. Capture cards thinks digital, so they determine an area as 720 x 576 pieces, buffer some millisecond of the analogue info that enters, and take a "photo" of that.
So, they grabb / capture / photo the Canvas that includes the original info, not the original info only.
Now, or we keep this "as is", or we extract the real info.
It is impossible to explain more. I am not able. If you still don't get it, you never will. This was always a problem: Some people get it, some people not. Seems like the theory people are unable to visualise something like this.

The bottom line is, that me and many others that got it , are capable to produce 352 x 576 DVDs from VHS / SVHS tapes, that look like yours at 720 x 576 but in 1/3 filesize.
And the reason some people can't see it, is because most DVD standalone don't support that framesize correct.

[edDV]

BTW, CCIR-601 is 720, but who's counting?

The CCIR-601 pixel aspect ratio results in a 704x480 raster (NTSC space). The extra 8 pixels on each side were added as buffer against horizontal phase (H position) errors during A/D or during broadcast distribution. The black horizontal edges are built into 720x480 and should be expected if capturing analog sources.

DVD often uses the full 720 space for rendering video but ATSC SDTV still uses the 704x480 raster since all broadcast recording libraries reference CCIR-601 (D1). Recall that half D1 (352x480/576) multiplies to 704 not 720.

[davideck]

Please, point me hardware to use so to capture the 100% of a signal.

It depends upon the quality of your source. See below.

The bottom line is, that me and many others that got it , are capable to produce 352 x 576 DVDs from VHS / SVHS tapes, that look like yours at 720 x 576 but in 1/3 filesize.

This is where we disagree.
The maximum bandwidth that can be preserved with 352 samples_per_line is;

352 * (1/1.33) = 265

265 horizontal lines. Slightly better than VHS bandwidth. Practical limitations would reduce this figure, but for arguments sake...
If your S-VHS sources are limited to this bandwidth, then 352 should be fine.

But the S-VHS horizontal resolution can be up to 400 lines. I doubt that my S-VHS masters are as high as 400 lines, but they are beyond 265 lines. Theory therefore suggests that capturing at 720 will preseve more detail than 352. And when I try this out in practice, my 720 captures look better!

The answer to your question, IMHO, is that the optimal solution to any problem is the one that satisfies both theoretical as well as practical requirements in the easiest way.

[edDV]

Forgive the technical response. We can try to adapt the conversation next.

352 x 576 is in theory adequate to sample VHS (3 MHz) but is not adequate for full spec S-VHS. S-VHS exceeds Nyquist's Law that sample rate must exceed twice the source bandwidth. Therefore raw S-VHS playback will generate sample alias products (digital Moire noise http://www.mathematik.com/Moire/ ) unless it is analog low passed filtered back to less than 3.375 MHz bandwidth before A/D.

S-VHS and Hi8 are capable of more analog resolution (up to 5.0 MHz) so at the limit 352x576 (half D1) is not sufficient. If you low pass filter S-VHS or Hi8 and accept the lower resolution quality, then you can use 352x480/576. But here is the limitation. There are no consumer capture cards with such a low pass filter optimized for 352x480/576 capture. They all target 640 or 720.

So what to do? You capture at 720x480/576 and then rescale to 352x480/576 using software digital filtering that will filter out the excess bandwidth.

Added.
Now for the additional unfortunate observation. Even though VHS in theory has less than 3MHz luminance resolution, it also carries the color subcarrier at 3.58 (+/- 0.5 sidebands) MHz for NTSC and 4.43 (+/- 0.5 sidebands) MHz for PAL on the composite connection. Very simple bandpass filters are used to lowpass filter luminance in consumer capture cards. In most cases significant high frequency subcarrier products leak through the filters during VHS capture. If the sample rate is 13.5 MHz (720x480/576) then these subcarrier products are captured as noise but do not cause aliasing.

However if the capture card samples at 6.75 MHz (352x480/576) then the subcarrier products (4-6MHz) will produce significant sample alias products. That is unless an expensive analog filter (a sharp 3MHz low pass filter) is used before the luminance A/D. I doubt any consumer capture card is doing this because such filters can cost more than the capture card unless analog filtering technology has recently improved.

So in conclusion, a capture card optimized for 352x480/576 (6.75 MHz sample rate) will need special expensive analog low pass filters to get good results.

Ref: Nyquist sampling theorem

Definition:

A law of sampling theory stating that if a continuous bandwidth-limited signal contains no frequency components higher than half the frequency at which it is sampled, then the original signal can be recovered without distortion. If the signal is not sampled fast enough, aliasing will occur.

http://en.wikipedia.org/wiki/Analog_to_digital_converter
http://en.wikipedia.org/wiki/Shannon-Nyquist_sampling_theorem
http://en.wikipedia.org/wiki/Anti-aliasing_filter

If you skip to the Anti-aliasing filter link, you will see that oversampling (i.e. 13.5MHz) is the method used to prevent the cost of the analog low pass filter going to infinity or in our frame of reference, >$30-100.

http://www.allenavionics.com/LC_F/Pr_files/vfl_Pr.htm

[Wilbert]

@SatStorm,

You seem to tell people they can't grasp it if they don't agree with you. That's a bad habit It's a bad habit in itself, but also in this case, because it is false.

Regarding the canvas.
Canvas here, is the word to determine the area of the analogue TV Screen. Inside that area, any kind of analogue video info, scretch so to fill it. This is how the analogue works: By zooming the info so to fill the area.

I think you mean the following with Canvas. This is what a capture device does:

A capture device digitizes the analogue signal. This is done by measuring the signal at a fixed time interval and recording the value as a binary number. The rate at which this happens is called the sample rate. Video is typically sampled many millions of times a second, or MHz (= megahertz, or 1000000 samples per second).
The capture card samples at a fixed sample rate and (...)

see http://www.doom9.org/capture/introduction.html

I think you call this samplerate (or more precisely the corresponding number of samples for pal/ntsc) the canvas. Note, if this is indeed the case, it's purely a digital concept.

Also note that this samplerate depends on the capture chip, and doesn't necessarily equate 720 pixels. For SAA71xx devices it's 27 MHz (PAL and NTSC), which equates 2*720=1440 pixels. For bt8x8/cx2388x cards for example: NTSC: 28.64 MHz, and PAL: 35.48 MHz.

see http://www.doom9.org/capture/sizes_advanced.html

So, they grabb / capture / photo the Canvas that includes the original info, not the original info only.
Now, or we keep this "as is", or we extract the real info.

I think you call 'the original info' the amount of detail present in the original signal. Here, the term bandwidth comes into play (an analog term): The term bandwidth describes the theoretical maximum amount of detail that can be reproduced for a given medium. At the end you want to know what this mean for the minimal required capture size (*) and Nyquist comes into play. For SVHS for example, it has a bandwidth of 400 lines of resolution (defined by the number of side by side dots that can be reproduced within a scan line) which equates 5.0 MHz. This corresponds (using Nyquist) with a minimal horizontal size of 520 pixels.

(*) as you know, i think, the samplerate by which the capture card captures is fixed, and is downsized to the requested capture size.

see http://www.doom9.org/capture/introduction.html

Wilbert and Davideck,
Since you two know what is correct and what is wrong, and since you know so much the theory behind all this, I believe you both are able to point to the rest of us, the known hardware and the brands, capable to succeed the 100% of what the theory says in practice.

I will do that soon.
[/i]

[SatStorm]

Wilbert, a bad habit is also talking theory, when someone asks you for practice. You, and few others, talk theory and you have nothing to say about how to do this in practice and with what.
Regarding myself, since you don't understand what I said (well, edDV got most of it), how you judge me? You judge something you don't understand? Now that is something....

I want to see from you something like: "Use the "SatStorm 2K" capture card with the "SatStorm C" Cables and the "SatStorm 2005" VCR. Capture at 720 x 576 using that kind of avi (or whatever codec) and you have 100% of your source on your PC.

In my eyes,you talk theory because you wish to proove that the average technology we, the home users and semi pross, use, is capable to capture and work as the theory says. Well, you don't impress me. This is possible only with very expensive hardware, even the few pros that gathers here don't have to afford it. With what we, the typical users, have access to use, it is impossible.
This is not a technical theory forum. This is a technical practical forum. We do tricks and voodoo to succeed the anorthodox way something close to what the theory points. The cheap way. And this "cheap" is already expensive for the average user.

Davideck I never said not to capture at 720 x 576! I said to encode that 720 x 576 we capture, after filter it and resizing it, to 352 x 576! If the source is a S-VHS tape, that is enough.
And the only way to do this realtime, from what I know, is with mainconcept. With the PVR cards, we can't do this.

edDV you did a great job to put in great english most of the things I try to say in broken english. Thanks!

We capture the higher we can, because the hardware we have, can do good only this way. Then, we filter / resize those captures using various software technics, so to end up with the "true", the original video info. Then we encode to the closest match: Half D1 for both VHS and SVHS.
SVHS because it is close to it and VHS because we can't use 352 x 240 (NTSC, this framesize can't be interlace) or 352 x 288 (this framesize usually don't have good DVD standalone support during playback and looks like crap)

[Wilbert]

Wilbert, a bad habit is also talking theory, when someone asks you for practice. You, and few others, talk theory and you have nothing to say about how to do this in practice and with what.

I said i would do that soon. I guess you missed that.

Regarding myself, since you don't understand what I said (well, edDV got most of it), how you judge me? You judge something you don't understand? Now that is something....

I don't understand what you said, because it didn't make much sense when taken literally. I'm not the only one who thinks that, because everyone is trying to guess what you mean by "canvas".

Did you try to understand why i wrote? Do you agree with my interpretation of your idea of canvas, or do you still think it can be applied to the analog world?

In my eyes,you talk theory because you wish to proove that the average technology we, the home users and semi pross, use, is capable to capture and work as the theory says.

No, far from that. If you read the analog capture guide (at doom9), you would know this is not true. Yes that guide is theoretical, but it discusses also practical issues (like the crap resizer bt8x8 uses at 352x576) and gives practical recommendations. (And yes i admit the guide is too technical for people who are new to the scene, or don't care about any technicalities.)

Well, you don't impress me. This is possible only with very expensive hardware, even the few pros that gathers here don't have to afford it. With what we, the typical users, have access to use, it is impossible.

I think i never said or claimed you need expensive hardware to get good quality. Moreover, i think that using expensive hardware doesn't give you significantly more quality than using the appropriate cheap/free hardware/software.

[davideck]

Davideck I never said not to capture at 720 x 576! I said to encode that 720 x 576 we capture, after filter it and resizing it, to 352 x 576! If the source is a S-VHS tape, that is enough.

As edDV has pointed out, filtering, resizing, etc. down to 352 reduces the bandwidth to something less than 265 lines, which is not sufficient for S-VHS.

[edDV]

Davideck I never said not to capture at 720 x 576! I said to encode that 720 x 576 we capture, after filter it and resizing it, to 352 x 576! If the source is a S-VHS tape, that is enough.

As edDV has pointed out, filtering, resizing, etc. down to 352 reduces the bandwidth to something less than 265 lines, which is not sufficient for S-VHS.

Yes, SVHS and Hi8 should be saved at 720x480/576 for DVD. 352 will cause loss of quality.

Wilbert, a bad habit is also talking theory, when someone asks you for practice. You, and few others, talk theory and you have nothing to say about how to do this in practice and with what.
Regarding myself, since you don't understand what I said (well, edDV got most of it), how you judge me? You judge something you don't understand? Now that is something....

I want to see from you something like: "Use the "SatStorm 2K" capture card with the "SatStorm C" Cables and the "SatStorm 2005" VCR. Capture at 720 x 576 using that kind of avi (or whatever codec) and you have 100% of your source on your PC.

Slow down SatStorm it just isn't that easy. Everything has tradeoffs of cost, effort and quality result. Broadcasters have to go for best quality, and they want a process with minimal labor intervention and as a result their captial cost goes sky high. Home users are willing to trade their own effort and some quality to get the capital cost within reason. We could list devices that may add up to thousands of euros and you wouldn't be any better off. So now we need to talk about labor (process) vs quality issues holding costs below say 1000 Euros and only after we do that can we spec equipment.

More later, lunch hour is over.

[SatStorm]

Wilbert, many already stated that they can "follow" what I say and that they can understand me. From what I know, those that don't get it the way I said it, are much less that those that get it. It is also simply statistics: This post here has so far 2075 hits and I believe that more than 100 people have read it. Only few posted "I don't understand". And some stated "I can follow him". So, it is obviously that it is not "everyone" as you said.
And I didn't missed something. I'm waiting. And what makes me wonder, is how it is possible someone that knows so well the theory behind all this, don't have to point solutions instanty, and 2 days thinks about them. When we talk about the theory of something, we have to give something to the people to test it, judge it, play with, instantly. If we don't do this, it is simply an atempt of demonstration of knowledge. A "show of".


edDV, somehow you managed to get my point: The 99% users of this forum, are using hardware not capable to follow all this theory you all mention here. So, why you confuse them with all this? Those who wish to learn more, they gonna google and they gonna end up on forums, which are about theory. This forum here, it is not about theory! It is about practice!!!!

Why it is so difficult to understand that with the cards we all use, capturing full D1, filter / resizing to half D1 and encode to half D1, produce results equal with encoding direct to full D1, but in 1/2 of bitrate?
Because thetheory says that full D1 keeps more info? But to support what the theory says, we need hardware beyond what we have. So, since it is impossible to keep what this D1 offers, why not use half D1 which is less bitrate so smaller in filesize?
We gain nothing to keep a VHS / SVHS at full D1.
Beyond the psychological need to believe that 720 x 576 is better half D1 when the source is VHS / SVHS, this option what more has to offer, compared 1/2 D1?
Nothing! In theory, a loss of 30? 40? 70? 100? horizontal lines... In practice it might be the info of what? 10 Horizontal lines. Who gonna notice that difference? In which display? On a plasma screen that shows crap anything came VHS/SVHS? On a 36" CRT screen? On a HDTV projector?
Anyway, this conversation never ends! it is the same the last 5 - 6 years!
Personally, I convert a 4 hour VHS or SVHS tape on 1 DVD, using as my target framesize 1/2 D1. The picture is excellent, and equal what Full D1 shows on screen, but in the half filesize. And I tested that with plenty cards over the years: From bt8xx based ones, to Canopus 55ADVC and from philips 9bit to conexant 10bit chipsets.
And hauppauge PVR cards are crap compared the results you can succeed the "old" fashion way. They do their job well, but they are not the best for quality.

[edDV]

edDV, somehow you managed to get my point: The 99% users of this forum, are using hardware not capable to follow all this theory you all mention here. So, why you confuse them with all this? Those who wish to learn more, they gonna google and they gonna end up on forums, which are about theory. This forum here, it is not about theory! It is about practice!!!!

The laws of physics apply equally to the budget home user and the national TV broadcasting networks. Before I can simplify, I need to reference the physical laws of sampling theory. These physical laws are not open to opinion only matters of practice and compromise.

Why it is so difficult to understand that with the cards we all use, capturing full D1, filter / resizing to half D1 and encode to half D1, produce results equal with encoding direct to full D1, but in 1/2 of bitrate?
Because thetheory says that full D1 keeps more info? But to support what the theory says, we need hardware beyond what we have. So, since it is impossible to keep what this D1 offers, why not use half D1 which is less bitrate so smaller in filesize?
We gain nothing to keep a VHS / SVHS at full D1.

Not true for SVHS or Hi8 as I explained above. You must be willing to compromise bandwidth (aka: frequency response or luminance "detail"). SVHS can produce up to 400 analog lines of resolution (approx 530x576) and 352x576 is less than that.*

If you undersample video, your signal is subject to alias noise. The only way to prevent aliasing is to low pass filter bandwidth to less than half the the sample rate. If you do this in the analog domain, the cost of the low pass filter becomes prohibitive. So, you oversample at 720x480/576 or multiples higher and then digital filter to 352x480/576 in a second pass.

....
Anyway, this conversation never ends! it is the same the last 5 - 6 years!
Personally, I convert a 4 hour VHS or SVHS tape on 1 DVD, using as my target framesize 1/2 D1. The picture is excellent, and equal what Full D1 shows on screen, but in the half filesize. And I tested that with plenty cards over the years: From bt8xx based ones, to Canopus 55ADVC and from philips 9bit to conexant 10bit chipsets.

If you are happy, then good for you. My cursed eyes can easily see the difference between analog SVHS/Hi8, 720x480/576 and 352x480/576.
Maybe the difference is the source quality of my SVHS/Hi8 material or my monitors.

I generally agree that VHS/8mm is just as good encoded at 352, but as shown above, any attempt to capture (sample) at 352x480/576 will severely reduce quality using consumer capture cards.

And hauppauge PVR cards are crap compared the results you can succeed the "old" fashion way. They do their job well, but they are not the best for quality.

No comment because I haven't compared them.



* analog to digital comparisons are very theoretical but a simple industry rule of thumb is 80 lines of "perceived lines of resolution" as viewed from a test chart is equivalent to 1 MHz analog bandwidth. 400 lines is approx 5 MHz analog bandwidth. For 5MHz to be properly sampled, the sample rate would be >10MHz. 10 MHz sampling would produce a raster of approx. 530x576.

[davideck]

to support what the theory says, we need hardware beyond what we have.

I disagree.

We gain nothing to keep a VHS / SVHS at full D1.

I disagree.

Beyond the psychological need to believe that 720 x 576 is better half D1 when the source is VHS / SVHS, this option what more has to offer, compared 1/2 D1?
Nothing!

I disagree.

And hauppauge PVR cards are crap compared the results you can succeed the "old" fashion way.

I disagree.

-------

Detailed explanations that dispute these claims already exist in this thread and in the links provided by edDV and Wilbert.

[davideck]

SatStorm -

Earlier in this thread you and I agreed that our two approaches to this hobby are equally worthwhile. That hasn't changed, has it? I'm sure that you are very pleased with the DVDs that you create, and I am very pleased with mine. That's all that really matters...

[FulciLives]

SatStorm -

Earlier in this thread you and I agreed that our two approaches to this hobby are equally worthwhile. That hasn't changed, has it? I'm sure that you are very pleased with the DVDs that you create, and I am very pleased with mine. That's all that really matters...

SatStorm got a bit frustrated at times with you because quite frankly you really did not treat him with any respect and now you have the balls to tease him for getting a bit agitated at you and your poor attitude.

So not kewl.

- John "FulciLives" Coleman

[davideck]

Then I apologize to SatStorm. It was not my intention to show disrespect or to tease anyone.

My previous post was a sincere attempt to put things in perspective and find common ground.
I also apologize to anyone who interpreted it otherwise.

[SatStorm]

Thanks FulciLives

I still wait a post here, with the hardware you all know that goes according the theory. It's about a week now....

edDV, you don't have cursed eyes: You probably have a DVD standalone (or more than one) uncapable to playback 100% what the 352 x 480/576 is (I already point out this issue). Also, you may have a TV with a bad internal scaler (Those LCD/Plasma TVs (and LCD / DLP projectors) don't care to do a great job on this, they care to be cheap)

Another parameter, is NTSC vs PAL. With NTSC, any slight piece of signal counts. On the other hand, visually, most PAL users can't even notice a difference between VHS / SVHS, so imagine what happens when they have to notice a difference between a 352 x 576 and a 720 x 576 mpeg 2, converted from a full CCIR SVHS capture. Also, my experience with NTSC VHS/ SVHS is low and most VCRs I have playback the NTSC tapes as PAL60. So I loose something there too.

BTW, I didn't notice any PAL user post in this subject at all. And imagine: the PAL users in this forum, is at least half...