Image cutting and poor picture clarity

[Voland]

Today I became the proud owner of OSSC. I need your help: I ​​ran into what I feared most of all – not all the lines of the image fall into the frame, 12 lines from the top are lost.
framemeister helped to solve a similar problem due to the presence of settings for vertical and horizontal scaling, as well as signal zooming.

My computer has 300 visible lines in the frame. Of these, OSSC displays 288 lines at best, could you add an additional mode of operation for 300 lines?

In addition, there is another problem.
The only mode in which my signal is recognized is the “240p / 288p proc” mode, which qualitatively displays the signal when selected:

– Line 2x -> Generic 4:3 ( https://ibb.co/zn278qN )
– Line 5x -> Generic 4:3 [1920×1200 format] ( https://ibb.co/jgk4fWT )

In the first case (see screenshot https://ibb.co/zn278qN ), 288 lines are placed in a frame, 12 lines are lost at the top, while the sharpness of the font is not very good.
In the second case (see screenshot https://ibb.co/jgk4fWT ), less than 288 lines are placed in the frame, and the frame is cut off on the right and left, but the clarity of the font is perfect.

For me, the best mode would be when the output signal would have the format of 1920×1080, but at the same time all 300 lines of the frame were displayed. Help me please.

Why is it impossible to make output resolutions of 1920×1200, 1920×1080, 1600×1200 and so on for mode “Line 2x -> Generic 4:3“? It is very important.

[Harrumph]

Why is it impossible to make output resolutions of 1920×1200, 1920×1080, 1600×1200 and so on for mode “Line 2x -> Generic 4:3“? It is very important.

The OSSC is not a scaler, it is a line multiplier ( = scan converter). If you want to understand more about how it works, you can read around on this forum and the wiki.

For your question, while you don’t provide a lot of information (e.g. it would be good to know the input line count & hor/vert frequency, which is displayed in the info window), one thing you could try is use Lx4, change V.Active to 300, reduce V.backporch until picture is centered. The output would be 1280×1200. It’s not a standard output resolution, so it might not display on your monitor anyway.

[Voland]

The display shows the following information:
AV1: RGBS 312p
15.62kHz 50.08Hz

Increasing V.Active has an effect up to 297 inclusive. Setting 298-300 does not change anything, so the 3 lines of the image are still lost. Lx4 mode on my monitor does not work.
To improve image clarity, I tried changing Sampling opt> Advanced timing tweaker> H.samplerate in increments of 1, but I couldn’t get such perfect clarity as in Lx5 1920×1080 mode.

[Harrumph]

312 lines is the standard for PAL 288p. I’m surprised your computer can generate so many active lines, because the content would have been hidden by overscan on pretty much all displays. Does the cut of lines hide any content when actually inside a game/program, or just in a test scenario?

Regarding V.Active; the total of active, backporch, sync and frontporch cannot exceed the number of input lines. So make sure V.Active+backporch+sync is less than 312 (frontporch will adjust automatically).

[Voland]

Does the cut of lines hide any content when actually inside a game/program, or just in a test scenario?

The result of further testing OSSC + retrocomputer Soyuz-Neon. Immediately make a reservation: the experiments showed that in the existing software of this computer, all 300 lines are actively used for graphics !!!

I have the usual 22-inch widescreen Dell 1920*1080, on which I began to conduct experiments.
Having tried all the combinations of parameters, I selected the optimal (V.Active = 297, V.backporch = 6, V.synclen = 4) at which the topmost (first) line of the Soyuz-Neon computer image coincides with the top line of the monitor matrix, and the screen displays 297 lines of the original image of the computer Soyuz-Neon: https://ibb.co/W5KCw04
At the same time, in the menu of the monitor, the current resolution is displayed as 710×594. Accordingly, the 3 lower lines of the original signal are lost, and when trying to increase V.Active = 297 -> 298 and higher, the image disappears for good, because the monitor cannot enter this mode.

Earlier, as an alternative to a PVM TV, I purchased a Sony LMD-1510W monitor, which turned up useless in terms of trying to connect to it the RGBS signal of any retro console. Unexpectedly, the HDMI input of this monitor opened up unprecedented multi-format capabilities, starting with the fact that the HDMI output of the OSSC, brought to it, was recognized as 576i and ending with the fact that the video stream of 301 lines, processed inside OSSC using a 3x multiplier, was recognized on my monitor as the resolution 1NNNx903 – This is despite the fact that the limit resolution of the monitor is 1366×768. NNN in this case changes arbitrarily in direct proportion to the value of the parameter H.samplerate set in OSSC, and, according to my observations, H.samplerate does not significantly affect the final result. For Lx3 mode, the output format was 16: 9, which was interpreted by the multiplier as 1280×288, where 301 multiplied by 3 was used as 288 lines.
Thus, by setting the parameters (V.Active = 301, V.backporch = 6, V.synclen = 4) + Lx3 (16:9) I received all 300 lines of the Soyuz-Neon computer image on the Sony LMD-1510W monitor: https://ibb.co/84KKK4N

Findings:
1). OSSC is a very, very interesting product, I hope the firmware will still be actively developed.
2). Since OSSC, unlike framemeister, is a multiplier, not a scaler, in order to provide better image clarity, you need to use the highest multiplication factor (Lx3, Lx4, Lx5). Accordingly, to use at least Lx4 mode, you will need a monitor that can display 1200 lines vertically. When complying with the FullHD / IPS criteria, the cheapest option is Iiyama ProLite XU2395WSU-1
3). The selected monitor should be as flexible as possible in terms of multi-format signal, otherwise it simply does not recognize most of the modes issued by OSSC.

Resume: I leave to myself OSSC, because besides the framemeister, this is the only decent and versatile upscaler that can also be bought 2 times cheaper.

My questions:
– Is there complete documentation for OSSC, which describes the purpose of each parameter, as well as a list of supported output resolutions for each of the Lx2 .. Lx5 multipliers?
– What affects the parameter H.samplerate and on the basis of which is the value calculated in which it should be set?
– What monitor needs to be purchased so that 300 lines of the original image fit on it when using the Lx5 multiplier? The fact is that I have enough of the image clarity that the Lx4 mode gives.

[Harrumph]

Some complex questions there… 🙂

The wiki is not complete documentation as far as explaining exactly how the sampling & line multiplication occurs, but at least it describes each option in the menu to some extent.
http://junkerhq.net/xrgb/index.php?title=OSSC

If you want to see some more info on the default video modes, see the bottom of this github file:
https://github.com/marqs85/ossc/blob/release/software/sys_controller/tvp7002/video_modes.h
The file is part of the OSSC firmware which describes the behaviour of the tvp7002 video chip (this chip handles the digitization of incoming analog video).

I will also copy paste something I wrote in another thread, regarding the video mode chosen by the OSSC.

The OSSC chooses a “video mode” automatically based on the input it receives and vertical multiplication mode selected, but it doesn’t matter which one it is because you can still freely change the parameters in Advanced Timing, just go into the one that is preselected for the console (it will be there when you select Sampling Opt -> Advanced Timing). Modes that end with L or LB are 288p modes also.

To clarify with your example what you see when choosing Lx3 16:9 mode: 1280×288, means that OSSC identifies the signal as putative 288p because it has 312 lines, and 1280 signifies it is set to Lx3 16:9 (or Lx4, which uses the same horizontal sampling), where the output horizontal H.active width is 1280.

What is important to remember is that the video output is actually extremely flexible in some regards (horizontal resolutions, active video area), while absolutely non-flexible in others (line count and vertical refresh rate). So one can achieve almost any custom resolution, at least as far as the area of active video described, however since total line counts and refresh rate cannot be manipulated, the output will always be more-or-less non-standard.

Another complicating factor is that for some monitors the total horizontal pixels (i.e. samplerate on OSSC) is used to calculate the appropriate display mode (disregarding settings to H.Active), while for others H.Active is more important. This can affect the representation of aspect ratio.

Without going into details, the default H.samplerate used in various modes by OSSC have been selected for best compatibility and preservation of aspect ratio in most scenarios.

[Voland]

When I connect to OSSC the Soviet computer BK-0011M, which has a frame rate of 48.8 Hz, then in Lx2 mode text running lines in demo programs move smoothly. If I turn on Lx3 .. Lx5 multipliers, then text running lines begin to twitch. How to achieve preservation of original frame frequency for Lx3 .. Lx5 multipliers?

[Harrumph]

If you mean your display performs frame rate conversion (it is possible it has different tolerances for higher resolution formats), then there’s not much to do.

If it is rather due to some unstable sync at higher line modes, you can either try turning off TVP HPLL 2x, or try adjusting sync settings (e.g. increase coast values, or increase Hsync tolerance).

junkerhq.net/xrgb/index.php?title=OSSC#Sampling_opt.
junkerhq.net/xrgb/index.php?title=OSSC#Sync_opt.

[Author]

Posts