Famicom Titler / other RGB varieties in commercials

I came across this video of old first-generation Famicom commercials: http://www.youtube.com/watch?v=PKXr_DbqHsU

The quality of the recordings is not great, but from the clear shots that can be seen (especially wrecking crew) I am not seeing any composite artifacts. Were they just years ahead of us in the US with fantastic comb filters, or do you think they are using RGB monitors and 2C03/etc PPUs?

Edit: Hmm, in the later Robot Gyro footage I can see the artifacts in the blue pillars. The SMB footage around 5:16 looks very clear, though...

Commercials that are broadcast on TV have to use a picture signal that conforms to the NTSC spec. The standard horizontal period is 341.25 dots (227.5 color subcarrier cycles so that successive lines are out of phase), not 341 or 340, and the standard picture height is 262.5 lines, not 262. There must have been some way to convert the footage for standard TV use.

How did genlock work in the Famicom Titler?

On a rewatch, it looks like they somehow scaled the video. Ice Climber, for example, is zoomed in at one point. I'm thinking something with an RGB PPU.

I don't mean I think they ran the video *through* the famicom Titler - I only mention the Titler as the only consumer-released RGB Famicom, and the footage of games running on monitors looks to be at RGB quality at times.

I think their zooming in involves a camera literally zooming in, hence the visible scanlines on Ice Climber; it is that the camera is filming an RGB system running on an RGB monitor.

Monitors are hard to film unless you are perfectly synchronized. Either they synchronized perfectly to the monitor's refresh rate, or they overlaid a video afterwards.
Usually in commercials and film, screens are simulated because they are so hard to film well.

Remember that the Titler was released in 1989. Any commercials running before that time couldn't have used it. What many magazines / CMs did use was the Sharp C1 TV, due to its RGB-generated Famicom display.

Nor did I mean to imply that anything used the Titler per se. But some of the methods of manipulating NES video may have used mechanisms analogous to the genlock mechanism in the Titler. Knowing how this genlock worked would help me guess at plausible methods of integrating NES video with other video.

Guessing blindly:
* Maybe 2C05-99 simply runs the pixel clock slower to generate compliant NTSC (seems fragile, though)
* Maybe the 2C05-99 has a STALL input, which stalls everything while asserted. This could be used to slow down scanlines (normally a little too fast for NTSC input, although too slow for RS170 input) and compensate (poorly) for the extra scanline on interlaced input.
* Maybe there's a PLL that generates the PPU pixel clock from stripped hsync input. Still needs some way to compensate for extra scanline on interlaced input
* Maybe there's a bunch of analog or digital delay lines in the extra PCBs and it uses this frame store to replay the PPU's output genlocked to the input. I think this is most likely.

Without better pictures of the whole thing (I can only find ccovell's photos here and nosuke's here) I can't really rule any of these in or out.

Yeah, sorry for not doing more technical analysis on my Titler.

IIRC, if you feed interlaced video into the Titler, the Genlocked video output is also interlaced. Feed it noninterlaced, and it remains noninterlaced.

As for how it synchs... I remember when I connected/disconnected video for overlaying, the sound on the Famicom noticeably got slower for a split-second. I reckon it does some voodoo with the CPU/PPU crystals via a PLL or somesuch to keep both video sources in sync.

Hmm, that's true about the titler's commercial release. If I remember correctly though, reading about the PPU's development, the 2C02 and 2C0x variants were designed at the same time, as the Famicom was developed to have the potential for an arcade release should the home market not pan out; it is possible that they had their own solutions / prototypes I suppose?