On 11/26/2012 04:17 PM, email@example.com wrote: > > On 2012-11-24, at 20:25, Gerrit Heitsch wrote: > >> If you use a 6569R5 on a 250466 board and replace the modulator with a just a driver circuit for S-video, you get a very clear picture. > > http://dl.dropbox.com/u/58002657/cbm/c64/chroma_problems_0.jpg > > Here is what I mean. Note the shift from lblue into reddish domain and the dark patches between some (mostly close to each other) pixels. > > Whenever I disconnect the chroma, luma exposes no inaccuracies. Well, yes... I have seen this on all my C64, more or less clearly depening on individual VIC and monitor used. But remember how color is encoded in the signal. It's in the phase angle of the color clock signal. The problem with the C64 and other computers that produce Y/C or CVBS is that the frequency of the pixel clock is higher than that of the color clock. That means it's impossible to have the color change every pixel (The reason why all horizontal parts of the characters are at least 2 pixel wide). If you try it anyway, the color decoder in your monitor will produce unexpected results. (*) Also, color clock and pixel clock on the C64 are derived from the same master clock, but pixel clock is not twice color clock. So when you change the pixel color, it depends on the position of the pixel on the screen where the color clock is in its period when it switches from one color to the other. That seems to introduce some problems depending on the color decoder in your monitor. NTSC or PAL was never meant for such a high color resolution. Maybe the 6569 is bit more pronounced in that respect. but I have seen this on a 8565 as well. But even my 6569R5 on the testboard connected to a 1084 monitor is not as bad as your picture. (*) Set up a picture of alternating white or grey and black pixels for example and get a surprise. Gerrit Message was sent through the cbm-hackers mailing listReceived on 2012-11-26 19:00:06
Archive generated by hypermail 2.2.0.