PALplus Sample

[jmac698]

Hey synthers, esp from Germany,
Do you have any old vcr recordings of PALplus? Could I ask you to provide a short sample? I was reading about it and it seems to be a brilliant analog hack, I'd like to write a software decoder for it. The essential information is in the U channel of the "black" borders, use utoy to check that some kind of information is there.
The rest is just a formula that could probably be done in masktools.

If you are familiar with TV shows, is there any old recordings still unreleased, but transmitted in PALplus, that would be worthy of restoring? Probably some kind of live events?

[Ghitulescu]

Are you talking about the widescreen PAL? How do you want to have the sample?
Is this still obtainable, eg from regular broadcast (cable, SAT) or commercial VHS?

[jmac698]

http://en.wikipedia.org/wiki/PALplus
Apparently ARD and ZDF are still using it. Anyhow, just record a sample (keep it in YUV!), and check that there's some kind of signal in the black borders with utoy. I don't think it was ever on commercial VHS, but it's compatible to be recorded on VHS. The best signal would be directly from cable I suppose. A short sample in lossless would be good, ffv1 in directshow compresses the "best" (I tested this in the alternative codecs thread). The sample should be widescreen obviously, good quality with a lot of vertical detail. Progressive/movie content would be good I think. YUY2 would be best (full color resolution this way).

If you capture directly in digital, of course I have no choice of format :) I think the article states that digital compression messes it up so some countries stopped using it. I can always try.

It should be possible to write a decoder/encoder and even make new VHS recordings with it, though I think that's pointless since you can just record the anamorphic dvd (player set to widescreen tv mode) onto VHS and just play it back on your widescreen tv (I've tried this and it does look better; anamorphic VHS!).

Actually, I've even made up a component format for VHS :)

[Ghitulescu]

Problem - apart from DVD-recorders (and [S-]VHS :)) I don't have any ready-to-use PC for analog capturing (I have to modify an existing one, which will take time). I think that anyway the analog cable that I have derives actually from SAT.
I can however provide untouched SAT-streams if you like.

[jmac698]

That's great, just to make it simple. If I can't find anything in the sample, then we can try something else if you like :)
Note this was an early analog technique to make compatible anamorphic widescreen - thus it doesn't apply to any HD broadcast.

[Ghitulescu]

That's great, just to make it simple. If I can't find anything in the sample, then we can try something else if you like :)
Note this was an early analog technique to make compatible anamorphic widescreen - thus it doesn't apply to any HD broadcast.

I know that :)
I'll record this week something from ARD and something from ZDF, to be on the safe side.

[TheSkiller]

PALplus can not be encoded to a lossy digital format like MPEG2 without destroying all the information in the chroma channels. It is not used for digital transmissions because for those there is no need to keep the signal compatible with 4:3 televisions that do not have the ability do display an anamorphic 16:9 image by squeezing it by themselves. The letterboxing required for those can be done in the DVB receiver.

@Ghitulescu If you're talking about digital (DVB-S) streams when you say "untouched SAT streams", those are not PALplus.


I will see if I can find something that is broadcasted in PALplus (containing "blue moving stuff" in the letterbox bars) via Astra 19.2 analog. :) I remember noticing the blue helper information on ARD a few weeks ago during a football match which surprised me because I thought I had read somewhere that ARD basically stopped broadcasting PALplus around 2004.

[Ghitulescu]

PALplus signalling is not stored in the chroma channel, but in VBI, which, true, get lost during conversion to MPEG-2. In addition, most DVD-recorders automatically set the DAR to 4:3 irrespective of this flag (Panasonic is the only manufacturer that documents this, but is equally true for most if not all of them).

I know the SAT signals are not WSS, since this is a feature for the analog signal. I offered this opportunity to jmac698 for its simplicity (mine :)). I'm thinking of recreating a PALplus signal by setting the DVD player.

[jmac698]

Try loading this into QuEnc you can see that it encodes perfectly:

#Simulate palplus signal characteristics, to test encoding
colorbars(720,432,"YV12")
ltr=last
black=blankclip(ltr,height=(576-432)/2)
u=black.utoy.subtitle("This is U",text_color=$FFFFFF)
v=black.utoy.subtitle("This is V",text_color=$FFFFFF)
top=ytouv(u,black.vtoy,black)
bot=ytouv(black.utoy,v,black)
stackvertical(top,ltr,bot)
assumefps("pal_video")
trim(0,240)

p.s. It's not exactly hidden, you can easily see the blue.

[TheSkiller]

OK, here is a PALplus recording. Recorded off ARD (German broadcaster) via satellite Astra 19.2 analog (Receiver: Grundig STR 622 Twin). Recorded in YUY2 straight to Ut video codec (4:2:2). Picked some short scenes that had moving details. 174 frames, 65.9MB

Download from Mediafire (http://www.mediafire.com/?6dnc9cfxbf41ewe)

Note that the first 4 lines are blank due to crop(0,0,0, -4).addborders(0,4,0,0) which is required to "fix" the misalignment of my capture card's capture window - it captures 4 lines too late so the first 4 lines are lost (and 4 useless ones that don't belong to the active picture are at the bottom). The information that can be found in the WSS signaling line which would be the very first of the 576 lines is lost also...


However, since my dish is only 60cm the reception is not too clean (especially the chroma). For a really good reception you need a bigger one like 75cm. From 85cm onwards the picture is literally free of any visible noise and looks excellent, noticeably better than the digital services of most broadcasters (I'm not saying every, it depends on the bitrate as always). Shame they will shut down the analog satellite next year.

[jmac698]

This is my first guess, problem is it's hard to tell - the plus signal is so weak. I wonder, if you guys could coordinate on the same show, one digital and one palplus?

#Decode PALPlus
src1=DirectShowSource("E:\project001a\palplus\PALplus_sample.avi",pixel_type="YUY2")
ltr=src1.crop(0,72,0,432)
t=src1.crop(0,0,0,72)
b=src1.crop(0,432+72,0,72)
decode=ltr.bilinearresize(720*576/432,576)
plus=stackvertical(t,b)
plus=plus.utoy.bilinearresize(720*576/432,576)
overlay(decode,plus.tweak(bright=-128),mode="add")
#plus

[2Bdecided]

PALplus signalling is not stored in the chroma channel, but in VBI, which, true, get lost during conversion to MPEG-2. In addition, most DVD-recorders automatically set the DAR to 4:3 irrespective of this flag (Panasonic is the only manufacturer that documents this, but is equally true for most if not all of them).

I know the SAT signals are not WSS, since this is a feature for the analog signal. I offered this opportunity to jmac698 for its simplicity (mine :)). I'm thinking of recreating a PALplus signal by setting the DVD player.

WSS, i.e. line 23 Wide Screen Switching, is only one part of the PALplus standard - the only part that remained in use mostly.

Full PALplus broadcasts full height anamorphic 16x9 content as letterboxed 4x3, and then encodes the missing vertical high frequency luma information in the UV/chroma information of the black bars. On a normal TV, you can sometimes see ghostly blue colours in the black (letterbox) bars above/below high frequency details of the main picture. On a PALplus TV, the extra information is used to re-build a full height anamorphic 16x9 picture (i.e. full 576 active lines, rather than the 432 active lines of the letterboxed version), and displays it stretched to full screen on a 16x9 display.

Very little used in the UK - channel 4 (UK PSB) used to use it - not sure now - I haven't used analogue for years and it'll all be switched off over the next year or so.

EDIT: The Wikiedpia article is quite good, and might even provide some hints for decoding. The links at the bottom might be useful too.

Cheers,
David.

[pandy]

http://web.itu.edu.tr/~pazarci/ETS300_731_PALPlus.pdf

[jmac698]

That's fantastic, thanks!

[scharfis_brain]

I am afraid to say this, but capturing a PALplus transmission with colour decoding enabled will most likely destroy all chances to recover the hidden information.

The subtle grey/blue artifacts in the black bars are just the QMF-Signals which fooled the luma/chroma separation of a standard PAL decoder.
IMO nothing can be restored out of this.

In order to gain access to the PALplus raw data, one needs to capture the full black&white signal without colour decoding (maybe feed composite to the Y-input of a S-Video input).
But I also doubt, that 704 pixels in width are enough accuracy to recover bare chroma let alone PALplus.

[jmac698]

Ok, the processing is a bit more complicated than my simple script. There's some companding going on.
As far as the helper signal being destroyed, the worst I can see happening is that pairs of lines are averaged. How would it be destroyed? Anyhow in the sample, I can clearly see signal in U and no signal in V.

As far as PAL not being decoded from luma alone, this has been done already, and there is a free program to do it. See here:
http://www.techmind.org/vd/paldec.html

Something even harder has been done; the BBC has invented a technique to recover color from the dots present on telerecordings.

Think positive :)

[DreckSoft]

I know this is an old thread but I just started to look for a PALplus decoder.

There are a few PAL+ LaserDiscs:
https://www.lddb.com/search.php?adv_search=*&adv_reference=&video=7&format=ld

There's also a project for software-decoding LDs, unfortunately there is currently no PALplus support. Maybe someone can help out.
https://github.com/happycube/ld-decode

[Balling]

Spec is now here. https://www.etsi.org/deliver/etsi_i_ets/300700_300799/300731/01_60/ets_300731e01p.pdf

[jmac698]

So I come back in literally 10 years and what do I find? My own thread trending! Mind blown.
Tbh I just wanted to see what it would look like. It seems there's limited material but certainly would be cool to have a decoder. Figuring out the math is kinda hard, and of course there's less interest in old analog stuff now. Might be a fun challenge one day when I retire? Check back in 10 years! :)

[SeeMoreDigital]

Over here in the UK it was mainly Channel 4 that broadcast some content in PALplus. It looked pretty impressive when viewed on a supporting Philips TV but sadly my friends S-VHS VCR was unable to record it correctly :(

[Balling]

ITU spec is here https://www.itu.int/rec/R-REC-BT.1197/e