that not only the problem, PAL is YUV and NTSC is YIQ, YIQ is designed to give colors at lower bandwidth with manipulation and giving “I” more bandwidth than “Q” while in YUV both U and V have the same bandwidth, also YIQ didnt used in digital video like DVD, so even NTSC DVD was YUV but it will convert to YIQ before output to the TV if the device is set to NTSC (maybe it will not do this in component video output)
It’s designed for that as per the spec but in practice almost no one did this and they did actually dedicate equal bandwidth to both I and Q most of the time. It was actually cheaper to split it evenly (simpler decoding scheme).
2 Likes
I think that was the case in s-video not composite
In actuality, the bandwidth of the I channel is reduced to about that of the Q channel and each is phase-shifted an additional 33 degrees. This facilitates lower cost decoding circuitry in the receiver.
1 Like
“In actuality, the bandwidth of the I channel is reduced to about that of the Q channel”
that seems worst than using standard YIQ band! and seems they kinda convert it to YUV with that “phase-shifted an additional 33 degrees” but with less bandwidth!
anyway, I have something you may find interesting! bad PAL signal affect saturation https://youtu.be/q8c-A05qay0?t=1153 maybe that’s why there’s a difference between RGB and composite (and even s-video) in saturation in some color ranges, most likely due to bad cables or some other fault with the TV or console or it’s some analog video limitation
1 Like
It would be worse, yeah. TV manufacturers had to do a lot of weird things to make the picture look decent- without emulating the whole TV we’re kind of just eyeballing it with saturation, tint and hue controls. What I’ve seen is that manufacturers tended to over correct with saturation.
I’m still unable to find definitive evidence anywhere suggesting that NTSC composite color was inherently less saturated or vibrant compared to PAL, in practice. Everything I’m reading relates to phase/hue shifting, and in theory that is completely correctable with the tint control on an NTSC TV.
that is interesting, could be what’s going on with the Super Metroid example.
This is a good example from @hunterk. There’s a slight loss of brightness and a bit more saturation and contrast, maybe? Not washed out, though.
I was curious so I was just plugging my PS2 into the AEG TV and do a NTSC/PAL and RGB/composite check.
- Tint control also works under RGB (what?)
- Color difference between PAL and NTSC, but also when switching from 50 Hz RGB to 60. PAL looks a little more muted.
- Composite and RGB look pretty similar.
I don’t need to test my other CRT TV to know that the differences are even larger (between PAL and NTSC anyway).
I guess next thing is testing LCDs and searching for my USB grabber. Looking through my captures, I did -S-Video versus composite, but not directly PAL versus NTSC 
2 Likes
maybe you didnt use RGB part of scart and you did use the Composite?
maybe PAL-60 was treated as if it PAL-M? in any case, I think the extent of the flicker of the fields in 50hz vs 60 hz also affects how the eye sees colors
that why it’s better to make sure that the console and tv are in a good shape and the tv is calibrated, no mods and only using original cables, the tv and console and games should be from the same region, also it’s better if the TV is from the same era of the console or a bit older
Lol, no I did pay attention, it works both with RGB and composite, as well as with PAL and NTSC. Maybe the result of it being an end of era (2006-2008) cheap CRT.
Idk, I still kinda think whatever significant color changes we see between composite and RGB are caused by the decoding process.
Themaister NTSC / guest-NTSC etc is basically a very idealized composite video, very close to the “near-perfect” examples I’ve posted. It’s doing all of the steps, conversion to YIQ and then decoding etc. What it isn’t doing is the analogue hardware-specific decoding - specific notch filters and comb filters. I think that’s where we’re getting big color changes, if anywhere.
This example makes a lot of sense to me. Colors are the same. There’s no comb or notch filtering involved. It’s just a sharpness difference.
Comparison of RGB and Composite. RGB is fed via my PC and a 2nd PCI-E Radeon 6450 and VGA cable, composite via RPI 2W and Lakka. Brightness has gone down in general, a bit of saturation too, overall composite image quality is excellent. Both used fceumm on Retroarch. Probably just a tiny detail loss due to artifacts and minimal chroma bleed, at least on a RPI 2W.
3 Likes
What kind of TV?
You made sure to select NTSC composite and not PAL composite? sdtv_mode = 0
The example just posted by Jamirus basically proved that the conversion from RGB to YUV does not need to entail any loss of brightness or color. It must be something happening during the decoding step.
That’s NTSC composite. But it’s on two different systems, OS etc. In general i could live with composite if i had to.
1 Like
I think there’s too much going on between those two setups-
What happens if you change to limited range for RGB? This drastic of a change points to a setup issue, I think.
I dont mind Idealized composite video but it will be nice if we have something that act like the real hardware (just in case and for preservation)
anyway IIRC this https://github.com/happycube/ld-decode can act like the real composite video
Edit: this only decode, to encode I think this https://github.com/fsphil/hacktv will do it
dont know if there are something better (there are https://github.com/LMP88959/NTSC-CRT and https://github.com/LMP88959/PAL-CRT and there are https://github.com/Gumball2415/pally that been also used in https://github.com/L-Spiro/BeesNES, there are also https://github.com/Slamy/fpga-composite-video) also maybe we need some SECAM https://github.com/tuorqai/libsecam love 
(Personally I have not played any video game with SECAM but it’s nice to have)
edit: also there are this https://github.com/svofski/CRT
BTW, we seems forget the sub-type of Composite video like PAL-N and NTSC-J which can make different in color and even in brightness
1 Like
The ntsc video output has a color clipping, which is compensated for by saturation in the TV modulator.
Then I write about this, I am trying to clarify some doubts and terms.
2 Likes
well, seems there are 2 type of official PAL (aside from regional types like N and M) https://youtu.be/A8wNBBv_ttk?t=1067 (saturation lost in 18:50) also why PAL Chroma subsampling 4:2:0 (DV not MPEG-2)
(NTSC is 4:1:1, but both can be recorded as 4:2:2 in digital world)
I think PAL-D is not a type rather than a way to do decode? the signal is the same?
also in case of “NTSC use equal bandwidth to both I and Q” it also in the receiver (consumer TV) side not the signal? https://youtu.be/CBFlhj2UMEk?t=1343
1 Like
also in case of “NTSC use equal bandwidth to both I and Q” it also in the receiver (consumer TV) side not the signal?
I’ve looked at a few data sheets for NTSC decoders and they all describe equiband decoding. They’d need extra components (a separate delay line, for example) to do decoding with different bandwidths for I and Q.
1 Like