A.D.A. Amiga Demoscene Archive

  Welcome guest! Please register a new account or log in

  

  

  

Demos Amiga Demoscene Archive Forum / Coding / Maximum OCS overscan

 

Author Message
Blueberry
Member
#1 - Posted: 4 Sep 2014 14:08
Reply Quote
How many (lowres, non-interlaced) pixels can the Amiga actually display?

There is an interesting thread about it here, but it could be interesting to hear if people here have any further experiences, particularly for OCS.

One thing that is immediately clear when delving into this is that it depends on the display device. And in these days of scandoublers and whatnot, there is a lot of variation to be seen here. I have tested an external (Mikronik) scandoubler on my BenQ monitor, a composite signal on my BenQ monitor, an RGB-scart cable on my (modern) TV, and WinUAE. All give different results. :)

Vertical:

It seems pretty consistent that the earliest display window start that can be displayed is $1A. WinUAE can display up to a display window stop of $139 (full PAL long frame height), but the highest I could get other displays to show is $138 (with one extra line of background color displayed below it). So we have 286 or 287 lines.

Horizontal:

I saw several displays which could go to $1D4 on the right (normal non-overscanned + 19 pixels, i.e. data fetch stop $D8 and shift value 3, last word wholly visible) but none further than that.

On the left, WinUAE can start as early as 376 pixels before this (i.e. data fetch start $20 and shift value 11, first word wholly visible), but none of the other displays would show more than 368. This seems to be the place where there is most variation.

From the right and bottom limits it seems that the maximum symmetric overscan is 358x280.

Anyone here with a CRT monitor or CRT TV who wants to test this further? :)
Paradroid
Member
#2 - Posted: 4 Sep 2014 15:59 - Edited
Reply Quote
My 1084s lets me squash the display to see more, but all my prods are limited to 352x272 as that already covers the entire screen for all the 4:3 TVs I've tested. The original Rink a Dink made use of 280 lines, but I later realised some of those were in the vblank area, which was a bit of a waste.

EDIT:
oh, just remembered.... when RAD Redux was shown at Revision they still managed to crop off some of the none overscan area, never mind the extra pixels :-/

EDIT:
The settings from one of my copperlists:

cmove (32<<8)|(113),diwstrt
cmove (48<<8)|(209),diwstop
cmove 48,ddfstrt
cmove 216,ddfstop
Blueberry
Member
#3 - Posted: 4 Sep 2014 22:50
Reply Quote
Yes, Painter's Frustration was also cut a somewhat (more on the left than on the right). It is also 352.

Maybe a CRT TV or old PAL monitor is not the best equipment to test the maximum overscan, since they typically, as you say, crop within the near overscan. We would need something which can show what the Amiga actually outputs. An oscilloscope or logic analyzer perhaps. :)
Angry Retired Bastard
Member
#4 - Posted: 5 Sep 2014 11:22
Reply Quote
Blueberry:
An oscilloscope or logic analyzer perhaps. :)


Let's not get all German now... :D
Blueberry
Member
#5 - Posted: 20 Sep 2014 10:43 - Edited
Reply Quote
Explanation for the 286/287 conundrum by Toni here:

Toni Wilen:
Last visible line can't have any sprite or bitplane data (long or short field makes no difference), only background color changes are possible. (This was not correct until 2.8.0)

Hence, anyone claiming to display 287 lines on a real Amiga is lying. ;)
Angry Retired Bastard
Member
#6 - Posted: 20 Sep 2014 17:56 - Edited
Reply Quote
*hurries off to write press release about reviewing our processes*

But yeah, that matches what I found as well. Consider the 287th line in Rubzilla an emulator-only hiddenpart. (And the 288th invisible one is done just for the hell of it). ;)

 

  Please register a new account or log in to comment

  

  

  

 

A.D.A. Amiga Demoscene Archive, Version 3.0