How do you determine "up scaling or transcoding" receivers?

[Dapper Dan]

When you look at the spec sheet (literature) for most equipment, most suggest that the receiver “up converts” to component, and some also state up conversion to HDMI.

How do you determine if the up conversion is in fact a true up conversion “scaled” rather than just transcoding the signal? (Without popping the lid and identifying the chip being used).

Example is the Yamaha RXV-4600 and RXV-2600. Whilst I haven’t seen the literature on the 2600 yet, the 4600 transcodes from what I can gather and the 2600 will up covert using a scaler, how does a consumer get the facts?

Cheers...

Dapper Dan

 

[w.e-coyote]

I don't think you'll ever be able to tell what's really happening inside the receiver.

On the topic of scaling and transcoding one needs to be aware of the difference between the two. Transcoding is basically changing the "format" of the video usually in the digital domain. Typical example would be the transcoding of an MPEG2 stream to MPEG4. My suspicion here is that the term transcoding it is also used to describe conversion of composite video to eg. component or S-Video. In either case the goal is not to improve the video quality just change to format to be able to present it to another device like a TV display.

Scaling does mess around with the video quality (within a certain "format"). It either adds or deletes pixels and/or lines. Scaling ought to be done at the very end of the chain and preferably only once when scaling the video to the target resolution of the display device (eg. plasma screen). Upscaling and downscaling at various points in the video chain never improves the quality. It is not uncommon to have several scaling operations in an A/V set-up.

To cut a long story short; I do NOT believe that the receiver can do a better job of scaling than the TV. To each his/her strength; receivers for audio; TV's for video.

 

[Dapper Dan]

Thanks W.E.Coyote.

In order to change from 480 to 720 (either interlaced or progressive), can that be transcoded or does that need to be scaled as the physical resolution is changed. I guess that means that lines will either be added or deleted as required to scale the image. In this case, picture quality will be modified from the original.

In the case of transcoding, the resolution would remain the same (480 lines say) and then just formatted to a different output. Does this include to HDMI? Is that the reason that most receivers only transcode for simplicity sake?

What about 100MHz component video inputs/output (with component video up-conversion and monitor out via component video overlay)?

What’s the benefit of 100 MHz as opposed to the current standard being used?

Dapper Dan…

 

[w.e-coyote]

Hi Dapper Dan,
Let me try to answer your questions. I don't pretend to be an expert in this field so others can chime in as they like, but this is how I was given to understand it all works.

Changing the vertical resolution from 480 to 720 is primarily a matter of scaling/de-interlacing. The upconversion will always take place in the digital domain (hence the importance of good quality ADC's). So the first step is to digitize the input in case of an analog video stream (can be CVBS {= composite}, Y/C {S-Video}, RGB {Scart}, or YPrPb {also called composite or YCrCb}. You could say the video stream is transcoded at this point but that is not correct. The video stream is then de-interlaced to eg. 240p in case of 480i input and then tripled to 720p. In this situation certain algorithms are applied in the temporal (time) domain by accounting for the slight difference in the time between frames (odd and even lines) and the spatial domain accounting for the added number of lines. Many, many ways are developed to do this ranging from very simple line repetition to motion compensated de-interlacing techniques. You're quite right in saying that the picture quality is changed. After all, picture information is being added based on educated guesses or cheap and cheerful displaying the same thing twice or three times. In my opion most TV's do a better job of this than receivers.

Technically speaking transcoding is re-formatting (the video) to another "language". The resolution does not necessarily change. What does change is the compression. An example is a DVD. In MPEG2 the resolution can be 480ix720 and occupy a space of 4.7 GB. The same DVD with the same resolution in MPEG4 (H.264 or whatever you want to call it) can take only 800 MB. The conversion from one to the other is called transcoding.

My guess is that what is being reffered to here as transcoding is in fact converting the video from analog to digital. The output can then be presented over the HDMI interface. Note that HDMI is only a delivery system and does not place any requirements on resolution. HDMI is a digital interface like DVI while all the others are analog.

The 100MHz component video is actually the sample rate with which the video is being digitized. Theoretically, the higher the sample rate the better the video will be. But whether the video quality really is better depends on the quality of the source material. No amount of processing will make real HD out of a VCR. So, the benefit of 100 MHz will only be apparant in case the content can do it justice.

Confused yet? I'll try to be more to the point in the future...

 

[Dapper Dan]

Hi Dapper Dan,
Let me try to answer your questions. I don't pretend to be an expert in this field ...

Could have fooled me W.E.C...

Thanks for your explanation, its clear as mud…! Actually, that’s a great way to put it and you answered my question very well. Once again, thank you for your input.

I guess what this really means then is that you may not really need to bother with onboard scaling on a receiver as I feel usually they will be an inferior product by comparison to outboard image processors (read DVDO, et. al). So the money that can be saved by buying say and "older" Yamaha RXV-2500 over the cost of a new RXV-2600 would mean the cost difference of a relatively higher quality video processor

How about the scalers onboard products like the Denon 3910 DVD player, which use the Faroudja “Genesis”chip. Are they better than the standard TV’s? Assuming you plug the dvd player directly into the TV input, therfore not using the reciever as a video swithcer.

Which would be the preference for video fidelity, outboard processors or those built into TV’s?

Dapper Dan…

 

[MDS]

w.e-coyote covered the relevant issues pretty well. I'll add a few more points and examples to hopefully make the distinction clear.

As he said, transcoding is the generic term for converting from one format to another and actually comes from the software world; eg., converting HTML to PDF is an example of transcoding - the information in the document hasn't changed, only the format of the document.

Converting composite to S-video is transcoding. The composite signal is a 'composite' of the Luminance (brightness) and Chrominance (color) information and is carried by one single wire. S-Video uses two wires to transmit the Luminance (Y) and Chrominance (C) separately. So to transcode from Composite to S-Video requires splitting out the Luminance part of the composite signal and sending it as the Y part of the signal and splitting out the Chrominace and sending it as the C part of the signal. Likewise for transcoding to Component, the composite signal is split into Y and two color difference signals (Pr and Pb). [Note that YCrCb is the digital format on eg. a DVD, and it becomes the analog format YPrPb when sent out the component jacks]

Up/Down-scaling is as was described - adding or removing information to match the target device. It is 'interpolation'; ie intelligent guessing as to what needs to be added or removed and there are many algorithms for doing so. If you take a 320x240 image in Photoshop and blow it up to 640x480, that is upscaling - the algorithm has to add in the missing pixels. Same deal if you use the 'digital zoom' feature of a digital camera.

Regarding the 100 Mhz bandwidth: (Analog) Video can be very high frequency. A receiver that transcodes would ideally be able to send the transcoded signal with as little degradation as possible and therefore the higher the bandwidth the better. 100 Mhz is more than double the 27 Mhz that is the maximum frequency required for HTDTV, so it should be able to send the signal relatively unchanged. The number they quote is usually the higest frequency that can be sent where the signal will be attenuated by no more than 3 dB (ie. be reduced by half).

 

[Dapper Dan]

One further point, W.E-coyote, my projector uses a Pixelworks scaler/de-interlacer chip, whilst I am not overly concerned about its performance; it can be bypassed as it is source dependent. Therefore, if I feed an HDMI input to the projector, the outboard scaler would be doing all the work.

I guess that answers my own question, video fidelity is a function of the quality of the outboard processor:

I might just have to consider the Denon 5910 with a Silicon Optix chip and DVDO scaler…HQV option now!

Dapper Dan…

I wonder how to get that passed my live in financial controller?

 

[w.e-coyote]

Hi Dapper Dan,
It's been a while since I was fully up to date on scalers and de-interlacers so my knowledge may be a bit rusty. The Faroudjia/Genesis chip is a good one by all means. It's strength is primarily in making diagonal lines look straighter. Often this is a quite noticeable artefact and the chip does a good job.

I'm not familiar with the DVDO chip but it also uses a motion adaptive de-interlacing technology (like the Faroudja by the way). Basically what it does is determine movements in the video fields of diagonals and add pixels to make the lines appear straight.

Nearly all commercial de-interlacers are using motion adaptive techniques in one form or another.

A fundamentally different solution is used by Philips in which movements in video fields are actually calculated (by motions vectors) and based on this new pixels and fields are created. This technique is known as "motion compensation".

Theoretically this method is superior to motion-adaptive methods but it really depends on the situation. In some cases motion compensation outperforms the Faroudja and in other cases it's the other way around. Each method has its own strengths and weaknesses which I won't bore you with here.

From what I understand there is more development in motion-adaptive chips than motion-compensated ones partly because of cost. Motion-compensation requires significant computational resources (ao. to accurately determine the motion vectors) and field and/or frame memories making it more expensive.

Seems to me you've got all your ducks in row. Can sympathise on the budget issue; sounds too familiar...

By the way, MDS, I was under the impression that HDTV maximum is 27 MHz. Not that any (CRT-) TV is able to display that or that it is at all visible (for mere mortals). Seems to me more of numbers game than anything else.

 

[MDS]

Yes it is 27 MHz - sorry for the typo.