Building a Standards Converter

[`FRANK.C]

Hi Terry
Yes please I would love to have a read of that.

Frank

[ppppenguin]

Frank, good to meet you at the NVCF on Sunday.

Sláinte!

This article from 1961 is also interesting if you're looking at the history of standards conversion.

[`FRANK.C]

Hi Jeffrey
Thanks for the pdf I will have a good read of it.
It was good to see you at the NVCF too. I had a great day there.

I was looking for a fairly descent 625 pattern generator as the only source of 625 I have in the workshop it a DVD player. I picked up two at the bring and buy. A Philips PM5519 and a GML MTG 205.
The GML is a really interesting generator that can produce a huge amount of patterns and can even do zone plates which I think might be useful for assessing how well a interpolater is preforming. Just need some time to play with it now.

Frank

[ppppenguin]

Having a decent TPG is more or less essential for standards converter development.

There is also a trick that is very useful - using the converter's own memory as a pattern generator. This is especially valuable with a framestore. It's not hard to rig up an arrangement that allows a PC to write to the framestore (and control the rest of the converter too). I use a parallel port on an old W98 machine to do bit bashing from programs writtenin QuickBasic but I'm sure there are plenty of more modern approaches.

The great thing about this approach is that you can separate the input and output systems of the converter. You can prove the output system using known test patterns which you can readily customise to measure particular aspects of the converter.

[`FRANK.C]

Hi Jeffrey
The right equipment and approach makes life easier.
One thing that I noticed was that the active part of the video coming from the decoder was shorter in pixel numbers that I would have expected. It looked to be approximately 700 pixels instead of 720. As the DVD player I have is a cheapo one suspicion fell on that. After trying it with the pattern generator it proved that is was the DVD player at fault as the pattern generator fill all of the 720 pixels.

Frank

[ppppenguin]

Your DVD player is not at fault.

The 720 active pixels in the digital standard is designed to encompass the entire analogue picture plus a little more so that minor displacement and blanking transients can be accommodated. Analogue PAL occupies 702 pixels as measured between the middles of blanking rise and fall times. From memory this is centred in the 720 pixel digital active line but I'd need to look it up.

For a 625>405 converter you can either impose 405 line blanking timings or simply rely on the 625 input being correctly blanked. My own approach is to rigidly impose a 720 pixel active line and let the analogue blanking take care of itself.

[`FRANK.C]

Hi Jeffery
I take it then I should have been aiming to fill the active part of the 405 line with 702 pixels and not 720 as I have been doing,  I will have to tweak the timings a bit to rectify  that. I couldn't understand then how the pattern generator had the correct length of active line until I realized that I had the converter switched to 5:4 output when the generator was connected .

I have got the 2 line interpolater going.  I am using two FIFO's, a one line to provide the delayed line for the interpolater and another which will hold over five lines for the time redistribution.  I am using the "cosine plus lift" shaped aperture, which is detailed in one of the BBC documents and the results are pretty good. I have plenty of resources left so may try a 3 line one, though a first sight at least, the aperture looks more complex.

Photos below from the converter and a screen shot of the resources used so far.

Frank

[Cardigan]

Analogue PAL occupies 702 pixels as measured between the middles of blanking rise and fall times. From memory this is centred in the 720 pixel digital active line but I'd need to look it up.

It is: if you look at Test Card J in a digital system you will note that there are two extensions to the 'diamonds' that show the edges of blanking in an analogue system. In an analogue system you would not see anything beyond the very tips of the diamonds.

This has the most accurate picture of J (scroll down).

http://www.murphy-radio.co.uk/gallery/tcf/bbc_tcf.html

[ppppenguin]

Looks good. To be honest, with 2 line interpolation I don't think it make s alot of difference what aperture you use. I used simple linear interpolation between 2 lines and it worked OK.

With 3 and 4 line interpolation things do indeed get more complicated. Too "gentle" an aperture and the picture will look soft vertically. Too "sharp" an aperture with big negative lobes and things will look over-accentuated and truly horrible.

[`FRANK.C]

Hi Cardigan
Thanks for the link, I hadn't taken notice of test card J before.

Hi Jeffery
I think the shape the cosine puts on the aperture makes little difference but the "lift" appears to be important.
I am using coefficients that 8 bits wide and first tried a aperture that was purely cosine in shape. The coefficients ranged from 0 to 254. The results were disappointing to say the least. I then tried introducing lift and am now using a lift of 51 with the coefficients ranging from 51 to 203. The lift made an huge difference. I guess that makes sense as a 405 line is wider than a 625 one so will always contain information from more than one line even if the 405 line lies directly on top of a 625 one. It makes sense now but I hadn't thought of it that way before.

Frank